The E-Commerce Director's Cut

The Smartest Online Businesses Still Rely on Human Insight

2025-08-19T00:00:00Z

AI can predict a customer’s next click, generate endless content, and optimize ad spend in seconds. But if you’re building an online business and want to grow sustainably, the real game-changer isn’t automation—it’s insight.

Not just any insight, but human insight.

The kind that sees beyond the dashboard. That understands why customers abandon a cart, not just that they did. That connects cultural trends with behavioral shifts. That challenges assumptions, asks better questions, and shapes more meaningful experiences.

Human insight brings the nuance that data alone can’t capture. It’s how great digital products are built. It’s how messaging cuts through noise. It’s how brands evolve from transactional to indispensable.

At Mobile1st, we use AI where it makes sense—to analyze quant data, summarize patterns, accelerate ways to grow. But we believe real growth comes from pairing those tools with human curiosity, empathy, and perspective. Because behind every data point is a person—and behind every great business is someone who never forgets that.

This newsletter is where we’ll share those insights—the ones that move the needle because they’re grounded in something more powerful than code: human understanding.

-Jonathan Silverstein, CEO - Mobile1st

Talk to Customers...Like Your Business Depended on It

2025-08-28T00:00:00Z

”If you don’t listen to your customers, someone else will.” — Sam Walton

That’s the whole game. Customers aren’t data points,they’re the only people who decide if your product lives or dies.

Companies like Blockbuster, Nokia, Kodak, Toys "R" Us, Borders, BlackBerry, and RadioShack all became cautionary tales after failing to tune in and adapt.

The Excuses That Kill:

“We already know what they want.” You don’t.
“It takes too much time.” Not listening wastes more.
”We’ll lose our vision.” Listening isn’t surrender—it’s survival.

What To Do Instead:

“Talk to your customers” sounds obvious. The hard part is actually doing it. Here’s how it looks in real life:

Put a quick survey on the pages that matter most. Landing page, product page, checkout. Ask one question. You’ll be shocked by what people tell you.
Hand your site to a stranger and watch them try to buy something. Don’t explain, don’t help. Just notice where they get lost.
Ask simple, open questions like “What almost made you leave?” instead of “Did you like our checkout?” One gets you the truth, the other gets you polite lies.
Take what you hear and change something small. Copy, layout, offer. Then see if it makes a difference.
Do it again. And again. The moment you stop listening, your customers stop waiting around.

The companies that grow aren’t the ones with the fanciest dashboards. They’re the ones who actually shut up long enough to hear what their customers are saying.

Because the companies that don’t listen? Suffer. Of course, there are numerous digital apps that can accelerate and structure research (like AI wrote a significant part of this post), but there is also an art to this.

Blunt (and cheap) tools cut poorly. They also create messy results. Mobile1st blends a mastery of site optimization with deftly crafted research that yields insights and persistent RPV growth. Let's chat...and start really talking to your customers.

When Your Customers Only Want One Thing

2025-09-04T00:00:00Z

We all want bigger baskets. More items per order. Higher AOV. But sometimes your customers have other plans. They just want one thing.

That’s what we saw with one client. The average order size was a single item. We could have kept pushing bundles and recommendations, but it wasn’t moving the needle. So we asked a different question:

What if the fastest way to grow revenue wasn’t changing customer behavior, but making it easier for them to do what they already wanted?

Cutting Out the Cart

Here’s what was happening. Customers clicked Add to Cart. Instead of buying, they were sent to a cart page. One more step. One more chance to bounce.

Industry wide, about 70% of carts are abandoned. Every extra click is a risk. For a business where most people only buy one product, that cart page wasn’t helping. It was hurting.

So we removed it. Add to Cart → straight to Checkout. That was it. No magic trick. No complicated upsell flow. Just less friction.

The Results

The test ran for three weeks and touched every dollar of revenue. The lift was hard to ignore:

Conversion Rate: up 8.2 percent
Revenue per Visitor: up 6.8 percent
Results reached a 98 percent probability of beating the control

Yes, AOV dipped a little. But that wasn’t the story. The story was more people completing purchases. And the revenue followed.

What We Learned

Owning e-commerce growth means you’re pulled in a dozen directions. Marketing wants brand moments. Finance wants efficiency. Leadership wants results yesterday.

The instinct is to add more. More cross sells. More steps. More “features.”

But in this case, the answer wasn’t more. It was less. We took out the cart step, and customers bought more often.

Revenue didn’t come from bigger baskets. It came from aligning the flow with how people already shop.

Don’t Just Skip the Cart

That doesn’t mean every brand should start sending shoppers straight to checkout. Your cart should be designed around your products and your customers.

If people often buy accessories, that’s what the cart should highlight. If it’s a big-ticket item, the cart should reassure them on value. If you’re selling bundles, that’s where you help customers see the upside of adding more.

The only way to know what’s right is through research and experimentation. That’s how you figure out whether your cart is helping or getting in the way.

The Next Step

The lesson from this experiment isn’t “always skip the cart.” It’s that your cart should be built around how your customers actually buy. And the only way to know what’s right is to test it.

That’s the work we do with clients every day, uncovering how customers really shop, then building the funnel to match.

Why Your Header is Too Big

2025-09-10T00:00:00Z

Shoppers don’t land on your site hoping for a giant search bar. They just want to find a product and feel confident hitting “buy.”

But here’s the catch: when the information they need is buried, the whole journey slows down.

We work with a large specialty retailer in the sporting goods space. Their catalog is massive, with thousands of SKUs. On paper, that makes search the hero.

Except on mobile, the header with search built right into it was hogging almost a third of the screen. And when we pulled up scrollmaps, it got even uglier. Key details on product pages, like reviews, specs, and pricing context, were shoved below the fold.

Shoppers weren’t absorbing enough of the product story to feel ready to buy.

So we asked the obvious question no one usually asks: what if the thing that’s supposed to help is actually in the way?

The Hypothesis

Maybe conversion isn’t just about removing friction. Maybe it’s about how much information a shopper can actually absorb before they decide to buy.

So we tested a simple idea: shrink the header and give that space back to the products. The bet was that if shoppers absorbed more of the right information, conversion would rise.

The Test

We ran two versions against the conrol:

V1: Shrunk the header, collapsed search into a little icon, and cleared out wasted whitespace.
V2: Did all of the above, but also made the promo “pencil banner” taller so offers were more visible.

Both were designed to put more product information in view. The question was whether promos would add value or just compete for attention.

The Results

The two versions told very different stories.

V1 (the winner):

Conversion Rate: +3.9%

Revenue per Visitor: +6.9%
Average Order Value: +2.3%
V2:

Conversion Rate: -1.4%

Add-to-Carts: -0.2%
Average Order Value: +2.3%

Shrinking the header gave shoppers more room to absorb product details, which drove more confident buying decisions. But when we stuffed that new space with a bigger promo banner, the benefit disappeared.

And because the header sits on every page, the winning version didn’t just help on product detail pages. It impacted 100% of revenue.

Why It Works

This isn’t about minimalism for its own sake. It’s about information absorption.

More product exposure. With less header clutter, shoppers absorb more of the details that actually sell products.
Higher intent search. Search usage dropped about 19 percent, but the shoppers who did search were more serious and converted at a +3.3% higher rate.
Stronger perceived value. Cleaner design aligned with premium brand patterns, which shoppers rewarded with bigger baskets.

The big signal here is that the amount of product information a shopper absorbs may be a leading indicator of conversion.

The Takeaway

This test wasn’t really about hiding search. It was about surfacing the content that matters.

If something as “untouchable” as the site header can choke off product information, what else on your site might be guilty?

That promo banner you’ve been running for years.
The navigation that buries category details.
The product pages where reviews or specs get pushed too far down.

The real question isn’t “how big should the search bar be.”

It’s: how do we maximize the amount of meaningful product information every shopper absorbs before they decide to buy?

Every shopper arrives at your PDP with a specific to-do list — and that might be the most important leading indicator of conversion you’re not measuring.

The Hierarchy of E-Commerce Insights

2025-10-09T00:00:00Z

No one teaches e-commerce teams how to find insights that actually grow a business. Not in college. Not in your MBA. Definitely not in onboarding. You just get dropped into the role and told, “Find the insights that’ll grow revenue.” It sounds simple until you realize that nobody has ever defined what an “insight” really is, or how to tell the difference between one that actually changes customer behavior and one that just looks nice in a deck.

So, we fake it. We stare at dashboards and call them “insights.” We copy “best practices” we saw on LinkedIn and call them “insights.” We ask AI to generate a report, skim it for something that sounds smart, and call that an “insight” too. We’ve confused data with understanding. Because, in reality, a real insight isn’t a number; it’s the story behind the number. “Conversion dropped 12%” is not an insight. It’s a symptom. It tells you that something is happening, but not why it’s happening, what the customer is trying to do, or what got in their way.

Real insights take work. Messy, frustrating, slow work. You have to get obsessed, build a theory, dig through behavior data, talk to customers, and test your assumptions in the wild. Most of those theories will die. But the few that survive, the ones proven through experimentation, become the kind of insights you can confidently build on. Those are the insights that compound. They make you smarter over time.

The Pyramid: A Hierarchy of Insight Quality

If you’ve ever felt like your team is drowning in numbers but starving for direction, it’s probably because you’re operating at the wrong level of insight. I’ve found that most e-commerce teams are somewhere near the middle or bottom of what I call the “Hierarchy of E-Commerce Insights.” It’s a simple way to visualize the quality and reliability of different sources of insight, from pure opinion at the base to validated truth at the top.

At the very bottom, you’ve got the weakest forms of insight: heuristic analyses, AI summaries, and the famous “highest-paid person’s opinion.” These sources are cheap and fast, but they’re built on assumptions, not evidence. They can spark ideas, but they don’t tell you what’s true about your customers. They’re the brainstorming stage of insight generation, not the conclusion.

One level up, you reach what most teams consider “data-driven” territory: user behavior analysis (like click maps, session replays, and funnel drop-off reports) and business intelligence (dashboards full of KPIs). These tools are valuable because they reveal what’s happening, but they still stop short of explaining why. You can see that 70% of mobile users drop off at checkout, but not whether they were frustrated, distracted, skeptical, or confused. You can measure the symptom, but you can’t diagnose the cause.

Then, as you move higher up the pyramid, you enter the realm of fundamental understanding, where customer feedback, surveys, interviews, and user testing start to connect the dots between what people do and what they mean. This is where you start hearing the language of your customers: their anxieties, their motivations, and the gaps between what they expect and what your site actually delivers. It’s where data becomes empathy.

And finally, at the top, you reach the ultimate validation: the winning experiment. The experiment is the moment where your theory collides with reality. You take everything you’ve learned, the behaviors, the feedback, the hypotheses, and you test them in a way that reveals whether they hold up under pressure. A winning experiment isn’t just proof that something “worked”; it’s confirmation that you’ve discovered a truth about your customers’ decision-making. That’s the difference between guessing and knowing.

Why Most Teams Never Make It to the Top

The problem is that climbing this hierarchy takes time, coordination, and the willingness to be wrong. Many teams struggle to progress beyond the middle stage because they get stuck in cycles of reporting rather than learning. Weekly performance updates become rituals of observation rather than discovery. Everyone’s talking about conversion rate changes, bounce rates, and ROAS, but no one’s asking why. Dashboards are tidy, but the reality behind them is messy. Insights live in the mess, in support tickets, reviews, session recordings, and conversations that don’t fit neatly into a chart.

Even worse, incentives often work against the climb. Marketing teams want faster answers. Developers want precise requirements. Executives want slides that look like certainty. But the truth is, certainty only comes at the top of the pyramid, after you’ve done the hard, unglamorous work of combining research, behavioral data, and experimentation into a story that holds up.

How Mobile1st Uses the Hierarchy of Insights

Every engagement we undertake begins with an in-depth investigation that encompasses every level of the hierarchy. The goal is to uncover the kind of insight that almost guarantees winning experiments. I won’t get into the full details of our audit process here, but I’ll share how we use it once we’re working with a client.

For one of our long-term clients, a vehicle parts manufacturer we’ve partnered with for nearly two years, we recently used this process to uncover a significant win.

During our initial investigation, we found that approximately 35% of shoppers left product pages due to a lack of understanding about the installation process. When we dug into chat logs, we found a recurring pattern of questions like: “What else do I need to install this?” and “Does this come with everything required?”

Next, we layered in behavior data. Click maps showed that three of the top ten most-clicked items were accordion sections with detailed product information. That told us customers were actively hunting for clarity. Then we remembered a recent losing experiment: when we removed key product bullet points from the top of the page, conversion rates dropped noticeably. That failure reinforced the same theme: customers wanted more information, not less.

Finally, we evaluated heuristic patterns and design best practices for accordion layouts. The conventional wisdom was to close all accordion tabs by default, making it easier for shoppers to scan and find the section they wanted.

By combining all these signals —behavioral data, customer feedback, past test results, and design heuristics —it became clear what to test next. One of our guiding themes for this client was “make product information easier to find.”

So we set up a simple test:

Control: Product description tab open by default.
Variation A: The most popular tab (Product Includes) open)
Variation B: All tabs closed by default.

Closing all of the tabs turned out to be the winner. It increased conversion rate by 20% and revenue per visitor by 40%.

When we reviewed session recordings of the experiment, the story became obvious. Shoppers were able to locate key product information faster, navigate between sections more easily, and discover complementary products they might need, driving both higher conversion and higher average order value.

This is precisely how the hierarchy works in practice. No single data source provided clear guidance. The win came from combining multiple layers of insight, qualitative feedback, behavioral data, design heuristics, and experimental validation into one clear story about what customers were struggling with and how to fix it.

The Payoff: From Guessing to Knowing

Once you start operating at the top of the hierarchy, everything changes. Your team stops treating optimization as guesswork and starts treating it as an investigation. You stop testing random button colors and start testing theories about human behavior. You stop asking, “What’s wrong with the PDP?” and start asking, “What’s missing in the customer’s understanding that’s causing hesitation?” Those questions lead to changes that move metrics for the right reasons and continue to improve, because they’re built on truth.

Climbing the hierarchy won’t make your job easier, but it will make your results more predictable. It transforms e-commerce from a game of whack-a-mole into a system of learning. And once your team experiences that shift, from reacting to understanding, you can never go back.

Most directors of e-commerce don’t have time to do all of this hard work every day. That’s where Mobile1st comes in. We do this hard work to grow revenue per visitor so you can focus on getting everyone on your team on the same page.

Our Process for Compounding Revenue Without Increasing Traffic

2025-10-16T00:00:00Z

Every e-commerce team thinks they have a traffic problem. They don’t.

They have an insight problem.

Traffic is the easiest thing in the world to buy. Anyone with a credit card and a pulse can spend money on ads. Real insight, the kind that tells you why customers hesitate, why they abandon the cart, or why your best-selling product isn’t your most profitable one, that’s the rare stuff. We laid out exactly how insight quality breaks down in the hierarchy of e-commerce insights.

Most teams drown in dashboards. They can tell you conversion rate by SKU, bounce rate by device, and revenue by channel before their second coffee. They know exactly where customers drop off. But not why. And that missing “why” is usually where the money’s hiding.

Dashboards are great at telling you what’s happening. They’re terrible at telling you what customers are actually thinking.

When the Data Stopped Being Enough

Before working with us, most of our clients thought they were being “data-driven.” They had dashboards stacked to the ceiling. Reports for everything. Product pages. Checkout flows. Cart abandonment. Attribution down to two decimal places.

But when revenue flattened, all that data started to feel like noise. It told them where things were breaking, but not why.

Everyone said, “Trust the data.” But the data never told them what customers didn’t understand about the product. Or what made them hesitate to buy. Or what they believed that wasn’t true.

Once we started working with these e-commerce teams, we started talking to their customers. Real, human, unscripted conversations. We asked them: Why this product? What problem were you trying to solve by buying it? What almost made you leave?

That’s when everything changed.

We Built a System Around That Idea

That obsession with insight turned into our Digital Product Growth System. Our approach to turning customer understanding into experiments that drive real revenue.

It’s not another CRO retainer. It’s not a random pile of A/B tests. It’s a process for making your existing traffic worth more.

We launch an experiment in week one, so we can immediately start driving growth. At the same time, we start with a deep audit, launching surveys on your site, digging through behavioral analytics, and comparing your site to heuristic best practices. From there, we turn those findings into hypotheses, and we prioritize experiments that impact the most revenue. This all but ensures we help you grow.

We break those hypotheses into three big themes, the areas that really move the business, and ignore the noise.

Then we design, launch, and analyze experiments across your store. We roll out the winners, document what we learn, and scale what works.

Every client gets a dedicated growth pod: strategy, UX, data, dev. It’s basically a strike team that plugs right into your existing setup.

And since time is the most expensive resource you’ve got, we keep your time commitment under five hours a month. You *never *have to tell us what to do.

Why Clients Stick Around

You can tell a lot about a company by what its clients say behind their backs.

Ours don’t talk about “great brainstorms” or “fun workshops.” They talk about results.

The VP of Finance at GRAV said our methodology is “100% data-driven. Every test starts with insight and ends in measurable impact.”

The Senior Director of Marketing at Phone.com said, “Each team member comes to calls well-prepared with insightful analysis and excellent questions.”

And one DTC founder said it best: “They knew what they were doing and did things for a reason. They got results.”

That last line might be my favorite, honestly. Because that’s the whole point. We don’t throw ideas at the wall. We find out what matters, test it, prove it, and make it permanent.

Our clients don’t stick around because of contracts. They stick around because growth finally feels calm, clear, and predictable, three words you don’t usually hear in e-commerce.

What Our Clients Get

Every engagement starts the same way: we do the hard work. You get the wins.

You’ll have your own Growth Pod, a strategist, UX researcher, data analyst, and front-end developer, who function like an embedded RPV task force inside your team. Their only job: make every visitor more valuable.

We launch your first experiment within seven days, so you start seeing movement right away. Most clients see their first measurable lift within the first 30 days — long before most agencies finish onboarding.

At the same time, we run a deep RPV audit that finds the biggest leaks in your funnel — messaging, pricing, friction points, missed urgency cues — and build a 12-month roadmap of experiments designed to grow revenue quarter after quarter.

Here’s everything included (and why clients stick around for years):

Customer-first research that goes way beyond analytics — real user behavior, interviews, and voice-of-customer insight so we’re testing what actually drives buying decisions.
Experimentation at scale — 24-48 high-impact tests per year, prioritized by potential RPV lift, not random “best practices.”
Done-for-you delivery — strategy, creative, QA, analytics, and launch. You approve; we run it.
Business intelligence dashboards that tie every test directly to incremental revenue and ROI.
Bi-weekly growth reviews that make it easy to see what worked, what’s next, and why.
Quarterly insight workshops with your merchandising, product, and marketing teams to spread what’s working across the business.
A UX & Analytics audit worth $45K in month one, included free.
An executive RPV dashboard that updates weekly so leadership can literally watch the revenue gains stack up.

And yes, we guarantee it.

If we don’t deliver a meaningful lift in RPV within six months, you get your money back.

Average client results:

More than 12% increase in RPV in the first year
Over 10x ROI on program cost

No fine print. No “let’s revisit next quarter.” Just growth.

You’ll spend less than five hours a month managing it, less time than you spend ordering lunch. We handle the rest.

Most of our clients stay for years. Not because they’re locked in, but because once growth starts feeling predictable, they never want to go back.

The Takeaway

Most e-commerce teams don’t fail because they lack traffic, talent, or tech. They fail because they mistake data for insight.

That’s what the Digital Product Growth System fixes.

It doesn’t just increase conversion. It increases understanding. It turns customer insight into profit.

Revenue per visitor isn’t just a metric. It’s proof that you finally understand your customers.

Searching for sustainable growth, that doesn’t involve discounting before the holidays? Let’s talk.

Every Shopper Lands On Your PDP With a To-Do List

2025-10-23T00:00:00Z

If shoppers can’t finish their to-do list, they don’t buy. Here’s what they’re trying to get done:

Here’s a thought experiment: if you deleted everything on your site except your product detail pages, how much would your business really suffer?

CMOs would panic at this question. The homepage showcases our best lifestyle photography. The category pages took months to design and get right. The “About Us” page underwent 14 rewrites by the brand team.

But if you really look at where money changes hands, it’s not on any of those. It’s on the product detail page.

The PDP is where shoppers stop browsing and start deciding.

You could spend six figures on ads, build the prettiest homepage in your category, and still lose the sale here, because the PDP is where people figure out whether your product is right for them, whether it’s worth the price, and whether they trust you enough to hit “Add to Cart.”

And that’s really the key: shoppers land on a PDP to get jobs done. It’s not random scrolling. They’re trying to answer a set of questions that make them feel confident about buying. When you think of the PDP this way, not as a design exercise, but as a list of customer tasks (and don’t tell me it’s the jobs-to-be-done framework, because I don’t believe in frameworks), it becomes obvious what matters and what doesn’t.

Here’s what every shopper needs to accomplish before they’ll buy.

1. Verify that the product actually solves their problem (and brings them joy)

Every purchase begins with a tiny itch. Something they want to fix or feel. Their dog’s anxious. Their shoes hurt. Their skin’s dry. They need a new shirt (they don’t, they already have 35 in their closet).

When they hit your PDP, they’re looking for reassurance that this thing in front of them actually addresses that itch. They want to see it, read about it, imagine it in their life, and ideally, see other people like them getting results from it.

This is why clear photography, straightforward copy, and honest reviews matter so much. They help shoppers connect the dots between the idea of your product and their actual problem.

A good PDP doesn’t force people to go hunting for answers. It makes all the right information easy to find. Some shoppers will dive into specs. Others will skim reviews or look at lifestyle photos. You don’t get to choose which path they take, so create an information hierarchy that makes it easy for them to find what they need.

2. Decide whether the perceived value is higher than the price

Price is a number. Perceived value is a feeling.

The goal isn’t to make your product cheaper; it’s to make it feel more valuable than whatever number you’re asking people to pay. That feeling is built slowly, through quality cues like good photography, confident copy, and thoughtful presentation.

It’s also built through social proof: reviews, testimonials, and real-world usage that signals, “Other people like me spent this money and don’t regret it.” Add in small psychological sweeteners—free shipping, gifts with purchase, extended warranties, and you shift the mental equation from “Can I afford this?” to “I’d be dumb not to get it now.”

Perceived value is the sum of a hundred tiny signals, and your PDP is where they all have to land at once.

3. Feel trust in the brand and the delivery

Every online shopper has a few scars. They’ve been burned by a sketchy product, a broken item, or a brand that ghosted them after checkout.

So while your PDP might be about the product, part of what they’re really evaluating is you. Can they trust that you’ll deliver what you promise?

That’s why details like shipping timelines, return policies, and guarantees aren’t just fine print; they’re persuasion tools. They answer an emotional question: “Am I safe giving these people my money?”

If you’re not a household name, make it easy for shoppers to feel comfortable. Show where you ship from. Make customer service contact info visible. Use photography that features real people. Show pictures of your office, your product facilities, your team, and your life.

Trust doesn’t come from being perfect—it comes from being human and transparent.

4. Understand how and when they’ll get it

People are buying because they need something, and they may need it right away. So they’re asking: “When will it get here?” Or am I better off getting in my car and finding something similar down the street?

People want clarity. Not “Ships in 2–4 business days” buried in the footer. Real clarity: “Arrives by Friday with free returns.”

The second question is, “What happens if it doesn’t work?” This is where most brands overcomplicate things. If your return process feels like a scavenger hunt, it’s easier for the shopper to close the tab than to trust you.

Clear timelines, easy-to-understand delivery details, and visible return policies build both trust and urgency. Shoppers want to know that the thing they’re buying will be in their hands soon—and that it won’t be a nightmare if it’s not what they hoped for.

One note on returns: if you get people thinking too much about returns on the PDP, they may get cold feet. It’s a delicate balance that can only be figured out through testing.

5. Find the right product if this one isn’t it

Sometimes people land on the wrong PDP. Maybe they searched the wrong color, or your Google Shopping ad linked to a similar SKU. That doesn’t mean the sale is lost; it just means they need help getting to the right thing.

This is where related products, intelligent recommendations, and clear navigation save you. When a shopper realizes, “Actually, I need a slightly different style,” or “I want the bundle instead,” you want that moment to happen on your site, not on Google.

A good PDP doesn’t trap people. It acts like a guide, helping them self-correct without feeling lost.

Pro Tip: your Google Shopping result page can look dramatically different than a PDP because it’s a different experience. Don’t be afraid to show similar products at the top of the page for these shoppers.

6. Justify their emotional decision with logic

By the time a shopper reaches this point, they already want the product. What they’re looking for now is permission.

They need to feel that this purchase makes sense, that it’s not impulsive, but reasonable. This is where technical details, materials, certifications, and even the “why we built it” story come into play. They don’t create desire; they make desire feel safe.

This is the step where someone goes from “I love it” to “I can explain why it’s a smart buy to my partner.”

7. Feel confident they’re not missing something better

Even after they’ve decided, there’s still a flicker of doubt: “What if there’s a better one?”

That’s not logic—it’s fear of regret.

Comparison charts, “best seller” tags, expert recommendations, and social proof are all ways to quiet that voice. They tell the shopper, “You’re making the right choice.”

Your PDP should make that moment feel final, not fragile. The goal is to replace hesitation with confidence.

So they add to cart.

So how do you improve your site?

Knowing these seven tasks is the easy part. Figuring out which ones your PDP fails at is harder.

You may be able to make some quick changes, but first you need insights to prioritize what matters. Start by talking to your customers to learn what they actually care about and how they make decisions.

Once you’ve got a list of potential improvements, priotize by impact ot revenue, and run experiments. Some ideas will lift revenue. Others will quietly break things.

The only way to know is to experiment.

If any of this is overwheling, just too much, and you just don’t have the resources, email me.

If you disagree with this newsletter, tell me.

The Higher the Price, the Longer the Relationship

2025-10-30T00:00:00Z

There’s a direct relationship between how expensive your product is and how long it takes someone to buy it. The higher the price, the more your customers need to hear about your brand, understand your product, and build enough confidence to actually click “buy.” The number of interactions needed to make that happen grows exponentially as price goes up.

That’s why impulse purchases live under $27, while the stuff that costs hundreds, or thousands, requires time, repetition, and reassurance. Nobody wakes up and decides to drop $1,200 on a single piece of outdoor gear or a $4,000 espresso machine. They need to learn their way into trust.

This is true everywhere. It’s why IBM sponsors every major sporting event under the sun. When a decision-maker is watching the US Open, sees IBM plastered on the scoreboard, and thinks, “Oh yeah, I liked that demo last week, I need to follow up with the sales rep,” that’s not an accident. That’s repetition doing its job. IBM knows that closing a $4.5 million consulting deal doesn’t happen after one cold email. It happens because the buyer feels surrounded by the brand, seeing it over and over again in contexts that make them feel confident it’s a safe, smart choice.

Now, that’s an extreme example. But the same principle applies whether you’re selling enterprise software or a $300 aftermarket radiator. The psychology doesn’t change, only the scale does.

The Real Customer Journey (It’s Not Linear)

Let’s say you’re that automotive brand selling radiators. You think you’re optimizing for a clean funnel. Someone sees your ad, visits your site, and buys. Simple, right? Except that’s not how anyone actually buys anything expensive.

Here’s what’s really happening:

A customer first hears about you from a friend who installed your brand’s radiator last year. They make a mental note. A few days later, they’re scrolling and see your brand again on a Reddit post. Now they’re curious. They Google you, land on your site, skim the product detail page, and think, “Cool, looks legit.” Then they get distracted and go back to whatever they were doing.

A week later, your retargeting ad catches them mid-scroll on Instagram. They click again, but this time they’re looking for more detail: technical specs, installation instructions, anything that helps them feel confident it will fit their car. They’re not ready to buy; they’re researching.

Then they head to YouTube to find installation videos. Maybe you have one, maybe it’s someone else’s. Either way, your brand name gets mentioned. They watch it, learn more, and head back to your site to confirm what they just saw. Then they text the link to a friend to double-check they’re choosing the right part number.

They do this over and over again for three weeks. Nine site visits. Three devices. A mix of search, social, and text messages. Finally, they decide: once the next paycheck hits, they’re buying it.

Friday night rolls around. They’re watching TV, beer in hand, looking at your site again, but the Spurs take the lead in the 4th, so they put the phone down. Go Wemby! Saturday morning, they pop the hood on the car, confirm the fittings look right, and finally, finally buy.

Twenty total brand touchpoints, four conversations with friends, and three conversations with themselves, they finally pull the trigger. And a journey your analytics platform has almost no chance of stitching together into a single “session.”

What This Means for Your Site

First, your conversion rate probably isn’t as bad as you think it is. People aren’t dropping off because your site is bad; they’re leaving because they’re doing research on other devices, at different times, in different moods. Most attribution systems treat that like a loss, when in reality, it’s progress.

Second, and more importantly, your site needs to do more than just sell. It needs to educate, reassure, and remind. The more expensive your product, the more your website has to act as both a salesperson and a teacher.

Your homepage may never even enter the equation. The product detail page is where almost all of this needs to happen. But it’s not just about “reducing friction” or getting someone to find the buy button faster. It’s about making sure they can find every single piece of information they might need, no matter where they are in their journey.

That means:

Technical details that are complete, accurate, and easy to find.
Photos that show real-world use, not just perfect lighting.
Installation guides or links to relevant content.
Reviews that actually talk about performance and longevity, not just shipping times.
And maybe most importantly, content that connects your product to the real-world problems it solves.

If you’re investing in great content for TikTok, Instagram, or YouTube that explains how good your product is, make sure those videos are accessible from your product pages. A customer who finds you through social isn’t necessarily ready to buy there, but once they land on your site again, that content can close the loop.

Lastly, if you’re measuring experiments, use revenue per visitor as your key metric. While you’re not going to be able to stitch every touchpoint together, if you can at least prove your change increases revenue across all of the visitors, even those that came back multiple times, it will be a better predictor of long-term growth.

Should You Personalize Your Site Based on Where They Are in the Funnel?

Yes, absolutely. As long as you’re personalizing for every part of the funnel. We’ve written about the top seven ways to use personalization on your site right now — the same logic applies here.

The hard part is understanding the intent signals to know where someone is in the funnel. But let’s say you know that, now you need to provide the relevant content. To do that, you already need to have the content produced and accessible on the PDP. So before you start personalizing, ensure that you have created and organized content for all parts of the funnel on the PDP.

And then once you start personalizing, please, for the love of God, don’t just give discounts to people who show intent signals that they’re ready to buy. That’s just throwing away gross margin. Instead, focus on increasing value through your use of USPs, and shipping and returns messaging.

The Real Job of a Product Page

No matter how expensive your product is, your job isn’t to make the site “convert faster.” It’s to make it easier for someone to stay confident across all those interactions that happen before the sale.

When people are looking at your product page, they’re trying to get 7 things done… It’s hard to tell which one until they engaging with the page, so make it easy to do any of these tasks.

That’s a subtle but critical shift. It’s not just about optimizing checkout or tweaking buttons. It’s about designing your site to support every part of a nonlinear journey that could stretch across six weeks, multiple devices, and countless distractions.

Because people don’t buy products fast. They buy them when they’re ready; after they’ve been convinced, reassured, and reminded. Your site just needs to be there every time they come back, ready to help them take the next step in their decision.

Disagree with something? Send me an email and let me know.

The Ultimate E-Commerce Profitability Model

2025-11-06T00:00:00Z

Customer File Segmentation

Send me an export of your customer file and I’ll find your best customers in a matter of hours.

If I could have any superpower, it wouldn’t be flying or invisibility; it would be omniscience. I want to know everything. I want to see the hidden mechanics behind why something works, the root cause of why it fails, the invisible chain that connects one decision to the next. I’ve always had this obsession with finding the ultimate truth in business, the place where the numbers line up and everything finally makes sense.

In e-commerce, understanding product-level profitability is about as close as we can get to that truth. You can analyze traffic sources, campaign performance, personas, audience segments, whatever you want, but until you break down the business all the way to the product level, you’re just looking at reflections of reality, not reality itself. Our Business Intelligence product is built to automate exactly this kind of analysis.

I’ve spent years trying to get there. At one company, I built a system that pulled daily data from UPS, CRMs, Google Ads, Meta, and our order management platform so that we could see, almost in real time, which products were actually profitable once you factored in everything: the cost of goods, marketing costs, shipping, and returns. It took months to automate. But we didn’t start that way.

We started with a spreadsheet.

And that spreadsheet, simple, ugly, manual, became our bible, or Torah, if you will. Every week we’d update it by hand. It was the one report that every team trusted. The category managers used it to figure out where their margins were slipping. The traffic team used it to see which products actually deserved paid support. The optimization team used it to prioritize their experiments — and as we've written about in your most valuable customers are not who you think, the truth hiding in your customer file is just as revealing as your product data. It was the closest thing we had to truth, and it changed how we worked. This report got the e-commerce team to be literally on the same page.

The document you can download alongside this post is a version of that same sheet. It’s simple, but it forces clarity. Here’s how it works, and what each section tells you.

Product Information, Revenue, COGS, and Marketing Costs for calculating profitablity

Product Information

Start by grounding everything in clear, reliable identifiers: product name, SKU, and product category. This sounds trivial, but most businesses fall apart right here. If your systems don’t speak the same product language, if your ad platform, your analytics, and your order management system all call the same item by different names, every analysis that follows is corrupted. Unifying this data is boring, unglamorous work, but it’s the foundation. Without it, you’ll never find the truth you’re looking for.

Revenue

Once the foundation is solid, you can look at revenue. You want to track the number of visitors to each product, the quantity ordered, and the product and shipping revenue associated with each SKU. From there, calculate a simple but incredibly revealing metric: Visitor to Quantity Ordered Rate.

This single number exposes more about your merchandising and product experience than almost anything else. If a product attracts traffic but barely sells, that’s not a media problem. That’s a value perception problem. Maybe the photography is weak. Maybe the price positioning is off. Maybe it’s just the wrong product. But until you can see it, you’ll never fix it. This metric became the favorite tool of our category managers and onsite teams because it forced everyone to confront uncomfortable truths about what was and wasn’t resonating with customers.

Costs: COGS and Marketing

Now comes the part that makes most people squirm: costs. We all love revenue, but understanding profitability means following the money out as carefully as we follow it in.

Start with cost of goods sold. Then layer on the direct marketing costs—Google Shopping, affiliates, maybe even Meta or TikTok if you can classify your spend by category or product. It doesn’t need to be perfect. We used rough percentages, allocating campaign spend based on each product’s share of total sales. It’s not accounting-grade precision, but it’s close enough to show where the opportunity to improve is.

When you finally see those numbers side by side; revenue next to COGS next to marketing; you start realizing how many products are secretly unprofitable once the real costs are accounted for. Those “hero” SKUs that drive all your volume? Sometimes they’re the ones quietly burning cash.

Shipping, Returns and Profitability for each Product

Shipping and Returns

This is where profitability quietly dies. Everyone loves to talk about CAC, CPM, ROAS, but few people want to look at the real costs of logistics and returns. Outbound shipping eats into your margins right away. Inbound returns double the pain. You lose the sale, you pay for return shipping, and sometimes you refund the entire order but sometimes you get back a restocking fee.

When we started tracking this by product, the results were humbling. Certain categories that looked fantastic on the surface, high volume, good conversion rates, turned out to be nightmares once return rates were factored in. It’s not enough to know how much you sell. You have to know how much of that revenue you keep.

Profitability Metrics

With all of those inputs in place, you can finally calculate what matters:

Gross Margin: Revenue minus COGS.
True Product Profitability: Revenue minus COGS, marketing, shipping, and return costs.
Gross Margin per Visitor: The amount of profit each product generates per site visit.
Gross Margin per Quantity Purchased: The profit per sale, fully loaded with real costs.

These metrics don’t just tell you how healthy your business is—they tell you where to focus. When we first ran this report, we found that several of our top-selling products were actually losing money. Seeing that in black and white changes you. It changes what you promote, what you bid on, what you restock. It changes how you think.

Where to begin and operationalize

Just as we did in the past, I recommend starting by manually filling out this sheet using a series of VLOOKUPS. This will allow you to understand where all this data needs to come from.

Start with revenue and COGS by product. That’s in your e-commerce system, and easy to report on. Next, add in product view data using Google Analytics or your analytics platform of choice. After that let’s integrate with your shipping provider, whether it’s UPS, FedEx or use a tool like Shipstation.

You’re starting to get most of the costs, so let’s now add all the return information. This will be the most revealing for most teams, and help them see that not all sales are equal. Lastly, integrate individually with all of your marketing platforms. These are the most complex, but have the most API access for automation.

To automate this process, I would start with an N8N or Zapier to get all of this information into a single spreadsheet. Once you’ve figured that part out, I’d upgrade to a database and reporting software that can produce this report for custom dates on the fly. This is obviously a more complex job, which may need some database design to complete. Once you have proven that the worksheet is so powerful, it will be easy to wrangle the technical resources to fully automate the process.

Want to use this, but not sure where to begin?

If you want to see how this model could work inside your business, I’m happy to walk you through the process. We’ll look at your systems, your data sources, and figure out what it would take to connect the dots.

Grab a spot on my calendar, and we’ll unpack it together.

Your Most Valuable E-Commerce Customers Are Not Who You Think They Are

2025-11-13T00:00:00Z

Customer File Segmentation

Email me your customer file. I’ll sift through it, pull out your highest value customers, and tell you what they have in common. Takes me a few hours. Could change how you grow for the next year.

My favorite thing to do in any e-commerce business is to tear apart the customer file. I’ll pour a cup of coffee, open a spreadsheet, and start digging: where do our customers live, what are they buying, how big are their orders, and how often are they coming back? Our Business Intelligence product is built to surface this kind of analysis systematically. It’s the most boring-sounding task in the world, but it tells you everything you need to know about how the business really works. The patterns in those rows and columns reveal the truth about what’s driving growth, what’s holding it back, and who’s actually keeping the lights on.

After doing this across dozens of brands, I’ve found a pattern that almost no one talks about. Your best customers are not who you think they are. They’re not your loyal Instagram followers. They’re not the superfans replying to your marketing emails. They’re not even the ones leaving glowing reviews or sharing your content. Your best customers —the ones with the highest lifetime value —are, more often than not, businesses.

The Hidden B2B Engine Inside Every D2C Brand

Yep, even if you think you’re D2C. Even if your site doesn’t have a wholesale tab. Even if your entire brand story is about individual consumers. Once a company crosses around $10 million in annual revenue, I’ve seen it again and again: buried in the order history is a small but mighty group of businesses quietly buying over and over and over again. They’re not loud. They don’t ask for discounts. They rarely reach out to customer support. But they are the foundation of predictable, profitable growth.

Finding them is easier than you’d think. Start by pulling up orders with a company name listed in the shipping or billing fields. That’s your first clue. These aren’t impulse buyers, they’re companies purchasing your products in bulk for their own customers, employees, or resale.

Then, look for your serial buyers, the customers who have ordered five, ten, or twenty times. While most people buy once and disappear, these few keep coming back like clockwork. When you see that kind of repeat behavior, there’s something structural behind it.

And if you really want to see who’s driving the bus, look at your outliers. When your average order value is $200, but someone just spent $8,000 or bought 30 of the same item, that’s not a one-off enthusiast. That’s a business customer hiding in plain sight.

What Not to Do Next

This is where most e-commerce teams get it wrong. They discover these business buyers and immediately think, “We should build a B2B portal.” Don’t. That’s how six months and a hundred grand in development vanish without a measurable return. These customers already figured out how to buy from you. They don’t need a login, a dashboard, or another friction point. What they need is a relationship.

Instead of building software, build a connection. Pick up the phone. Ask how they’re using your products. Ask what made them choose you, and how you can make their life easier. You might learn they’re reselling your product, giving it away as part of a corporate program, or using it inside their own manufacturing process. Sometimes you’ll even stumble into a new distribution opportunity you didn’t know existed. But you’ll never know unless you talk to them.

Serve Them Like Partners, Not Promo Codes

Once you know who these businesses are, stop treating them like everyone else. Segment them out of your normal campaigns. They don’t care about “20% off this weekend” or new lifestyle photography. They care about reliability, lead times, delivery options, and stock availability. Change your communication accordingly.

And here’s a surprisingly effective move: send them something physical. Yes, actual mail. A postcard, a catalog, a handwritten note. Everyone’s inbox is a mess, but physical mail still gets noticed, especially by business buyers who open it thinking it might be an invoice. It’s an easy way to stay top of mind without another ignored email.

Most e-commerce directors spend their time chasing new consumers, tweaking ads, or obsessing over funnel metrics. But your fastest path to sustainable, profitable growth might already be right in front of you. Those repeat business customers are steady, loyal, and far less expensive to serve than the constantly shifting consumer base. They don’t need convincing. They don’t care about your latest ad creative. They just need a reliable partner who ships on time and makes it easy to buy again.

Find them. Call them. Learn from them. Serve them like partners. You might realize your most valuable customers were never D2C at all, and once you do, you’ll see your entire business differently.

Need help finding these customers for your store? Drop me a line.

The View to Buy Report: The One Report That Will Save Your BFCM Week

2025-11-20T00:00:00Z

Insights that Guide You

Are you struggling to pinpoint where to focus? Stuck just guessing? You’re probably not looking at the right reports. Most e-commerce teams aren’t. Want to figure out the right reports? I can solve that for you. Schedule a time here

Why This Report Matters More Than Your Funnel

You are going to be refreshing your funnel reports every five minutes next week like they personally wronged you. You will watch people add to cart, abandon, return, scroll around, and then vanish to go price-shopping somewhere else. Everyone obsesses over Added to Cart and Checkout Started, but the quieter, more important truth is hiding somewhere else entirely. The report you should be glued to is your Product View to Buy report. It tells you which products are attracting traffic, which ones are driving actual orders, and which ones are silently setting your money on fire while you keep sending more visitors to them. In the chaos of BFCM, this report becomes your flashlight. Read the rest to see how I generate the data in GA4, and download my Excel template here.

What the Product View to Buy Report Actually Tells You

How to Build This Report in GA4

Start by creating a new blank exploration in GA4. Add Item Viewed and Purchases as your metrics. Add Item Name as your primary dimension. Drag Item Name into Rows, add your two metrics, and export the whole thing to Google Sheets. This is where the report becomes actionable. In Sheets, create a new conversion rate column using Purchases divided by Item Viewed. Clean up the sheet by deleting the first five rows so your data starts cleanly at the top. Then manually insert a Site Total row in A2, which gives you a baseline for the average conversion rate. Format the conversion column as a percentage so the numbers are easy to scan without mental gymnastics.

Turning Your Sheet Into Something That Actually Helps You

Your report is already sorted by views, but you need to isolate the products that matter. I suggest focusing on items with at least one hundred views, although your threshold may vary depending on your traffic levels or catalog size. Once you have the meaningful set, turn the conversion rate column into a visual map so your brain can quickly see what needs attention. Use Format, Conditional Formatting, Color Scale, and apply a red to green scale with yellow in the middle. Set red to zero and set the midpoint as the site average. Suddenly your underperformers show up in bright red, the average performers hover around yellow, and your winners light up in green.

What To Do With the Red Products

These red products are where you should invest your time. Begin by checking traffic sources. If most of the traffic is coming from paid channels like Google Shopping and the conversion rate is terrible, pull this product from your feed immediately. BFCM is not the time to donate budget to products that refuse to earn their keep.

Next, review the product page itself. Can you upgrade the images quickly? Can you rewrite the copy to make benefits stronger and more obvious? Can you tighten the bullet points so the customer understands the value faster? Customers during BFCM shop like they are sprinting. If your product page does not answer their questions immediately, you should not expect miracles.

Finally, look outside your four walls. Go to Google and search the product category the way a normal shopper would. What are the cheaper alternatives? What objections are obvious? What competitors emphasize? This exercise forces you to confront the actual value of your product, not the version you created in a meeting six months ago. If you cannot clearly justify why your product is worth its price, your shoppers certainly will not try to figure that out for you.

The Real Reason Conversion Rates Don’t Move

You can spend all day obsessing over UX improvements, shaving seconds off load times, or adding more trust badges to the page. None of that will save a product with a weak value story. Conversion rate is a function of perceived value, not button color. If the product does not feel worth the price, your conversion rate will stay low no matter how many optimizations you stack on top of it.

The Bottom Line

Focus on the red products. Fix them if they can be fixed, promote the green ones more aggressively, and stop paying for traffic that goes nowhere. If you can do that, your BFCM week becomes less stressful, less expensive, and significantly more profitable.

We’re taking next week off, but if you have any questions you want answered, we’ll answer them in our next newsletter. Just reply to this email, and ask away.

The E-commerce Playbook is Broken. Here’s the 2026 Reality.

2025-12-11T00:00:00Z

Is your team missing this key role?

I mapped out the 7 biggest headwinds hitting e-commerce next year (and the huge opportunity hidden in them). We are rolling out 3 new AI tools to help solve them. Want a sneak peek?

If you work in e-commerce right now, you’ve probably got a nagging feeling that the ground is shifting under your feet.

You aren't crazy. It is.

I was recently mapping out the landscape for 2026, trying to get past the usual LinkedIn buzzwords and figure out what’s actually keeping retail leaders up at night. It’s not about "embracing the metaverse" or whatever the flavor of the month is. It’s about the fundamental friction points in how we operate.

The golden era of "build a Shopify store, run some FB ads, and print money" is officially dead and buried. The new reality is messier, harder, and requires a totally different organizational approach.

Here is the unfiltered reality of the headwinds, and the hidden opportunities, facing e-comm teams right now.

The "Easy Button" is Gone

Remember five years ago? Media buying was basically a math problem. You used lookalike audiences, stacked interests, and hacked the algorithm with rules.

That’s over. In 2026, targeting isn't rules-based; it's creative-based. The ad platforms’ AI is now smarter than your best media buyer. The only lever you have left to pull is the creative itself: the video hook, the image, the story.

The problem? Most e-comm teams are built for spreadsheets, not for churning out the massive volume of high-quality storytelling needed to feed the beast. Now, you need to deeply understand the audiences you’re targeting and what they actually care about just to build enough unique creative to keep Meta happy.

The Identity Crisis and the Data Fog

Compounding the creative problem is a massive identity crisis. Thanks to privacy changes over the last few years, most retailers honestly have no idea who the personas on their site actually are. We’re optimizing for an "average" customer that doesn’t exist, rather than zeroing in on the high-value cohorts that actually drive profit.

And good luck using your analytics dashboard to figure it out. Let’s be real: GA4 has been a disaster for clarity. It is harder to set up and even harder to interpret. Because so much of the data is now "modeled" (read: guessed) rather than tracked, teams have lost trust in the numbers. When you can’t trust the dashboard, decision-making reverts to whoever has the loudest voice in the room.

The Hangover: Discounts and Returns

We are also suffering from a massive hangover brought on by the "growth at all costs" era.

First, we’re hooked on the "Discount Drug." We spent a decade training consumers to never buy anything at full price. Brands are trapped in a cycle where they can only move volume during a sale, which crushes the already razor-thin margins.

Second, nobody wants to talk about the "Returns Tsunami." As sales grow, reverse logistics are eating profits alive. Too many teams treat a 30% return rate as just "the cost of doing business," rather than what it actually is: a massive failure in marketing promise or product experience. There is almost zero feedback loop between the warehouse team seeing why things come back and the marketing team hyping them up.

The Internal Squeeze

So, margins are lower and budgets are tighter. The pressure for headcount efficiency is immense.

This creates a frantic environment where teams are so busy putting out fires that they have no time to actually talk to customers. I mean really understand why people are choosing your brand, what problems they’re trying to solve, and who else they’re comparing you against. It’s hard, very time-consuming work.

We have a massive empathy gap. We rely on quantitative data (what happened) because it’s fast, and ignore qualitative insight (why it happened) because it takes time.

And then there’s AI. Management is asking teams where the AI ROI is, expecting it to replace staff. But right now, teams are mostly using AI to automate small tasks, creating more mediocre content faster, which just requires more senior oversight to quality check. E-commerce teams haven’t figured out how to deploy AI to be customer-facing yet. Those that have are usually just adding a generic chatbot, which is clunky and rarely helpful.

The Missing Link

When you look at all these headwinds combined, like the need for better creative, the blind data, the profit squeeze, and the disconnected functions, it points to one glaring hole in most organizations.

E-commerce teams lack a "Connector."

There is rarely a dedicated resource whose sole job is to collaborate across design, development, business intelligence, and user research.

We have marketing driving traffic to a leaky bucket. We have product teams building cool features that don’t drive revenue. But we lack someone focused purely on experimentation and Revenue Per Visitor (RPV) growth. We need someone who owns the outcome of the site experience, not just the inputs.

How we’re helping e-comm leaders keep their jobs in 2026

At mobile1st, we’ve been quietly increasing RPV for all of our clients by stepping in as that missing "Connector."

We do the work teams don't have time for: understanding why the customer buys and ensuring that value is expressed on key pages. We talk to the customer through deep user research, and we dive into the messy business intelligence data to find the truth.

The results speak for themselves: We've seen an over 20% increase in RPV this year for our clients.

But even with those wins, we know our clients are still feeling the pressure of 2026. So, we’re evolving our product to meet them there.

Partners who align with our core experimentation product now get access to a new suite of tools that actually deliver on the promise of AI:

True Conversational Commerce: Not just a "dumb" chatbot, but a guided AI experience that genuinely helps shoppers find key information and products, reducing friction.
Data Integrity: A tool that ensures your product data on the landing page matches your PDP, which matches your backend specs. This helps eliminate the errors that cause returns.
Persona Identification: A tool that combines public firmographic info with live browsing behavior to identify your key personas in real-time, helping you prioritize messaging across ads and on-site.

The headwinds are real, but they aren't insurmountable. You just need the right connector.

The Great E-Commerce Hypocrisy: Why You Treat Your Ads Like Science and Your Site Like a Guessing Game

2026-01-08T00:00:00Z

I know what your week looks like.

You are absolutely grinding on customer acquisition. You are feeding the beast. Your creative team is sweating bullets, churning out video assets, static images, and carousel copy faster than a fast-fashion supply chain.

Why? Because you have to.

Ad fatigue is real. Creative burns out in days, sometimes hours. Facebook and Google’s algorithms are hungry little monsters that demand fresh meat constantly to keep your CAC (Customer Acquisition Cost) from skyrocketing into the stratosphere.

You are obsessively testing. You’re throwing spaghetti at the wall to see what the algorithm catches. You trust the machine. When the dashboard says "Ad A is a winner," you don't argue. You double down. You scale the spend. You high-five the media buyer.

But then... that expensive, hard-won traffic hits your website.

And you treat your website like a museum exhibit while treating your ads like a high-frequency trading floor. It’s the single biggest hypocrisy in e-commerce right now, and honestly? It’s costing you a fortune.

The "Clean Data" Trap (Or, Why You Trust Zuck More Than Your Own Eyes)

Here is the uncomfortable truth: You trust Facebook and Google because they have better marketing than your analytics team.

When you look at Ads Manager, they’ve stripped away all the nuance. They don’t show you confidence intervals. They don’t bore you with p-values or standard deviations. They give you a shiny green arrow that says "ROI GOOD." It looks unquestionable. It looks like absolute truth.

So you act on it immediately.

But when it comes to your own site—the place where the actual transaction happens—you suddenly turn into a skeptical academic from the 1950s.

If someone suggests changing the "Add to Cart" flow or testing a new value prop on the PDP, you freeze. "Is it statistically significant?" "Are we 99% sure?" "What if we break the brand aesthetic?"

Or worse, you fall back on the HiPPO method: The Highest Paid Person’s Opinion. You make changes willy-nilly because the CEO didn't like the shade of blue, or because a competitor did something cool last week.

You demand 99% statistical certainty to change a button on your site, but you’ll let Facebook spend $50k on an audience segment based on a "learning phase" algorithm you can’t even see.

Make it make sense.

One Size Fits... Nobody

Think about this: You are paying specifically to target distinct audiences. You have creative for the bargain hunters, creative for the luxury buyers, and creative for the impulse shoppers.

Facebook has done the heavy lifting to find these specific people.

And then you drop them all onto the exact same website.

Why is your site a static monolith when your acquisition strategy is dynamic and fluid? If you wouldn’t run the same ad to every single person on the internet, why are you showing the same homepage to every single visitor? Personalization is how you close that gap.

Okay, So How Do We Fix This Mess?

Stop treating your site like a sacred cow. It’s a product. It needs to grow. We call this Digital Product Growth.

If you want to stop lighting ad spend on fire, you need to apply the same rigor to the post-click experience that you apply to the pre-click experience. But—and this is key—you have to stop guessing.

To actually move the needle on Revenue Per Visitor (RPV), you need to know what matters. That starts with understanding which insights are actually worth acting on.

Don't start with the homepage. (Seriously, stop it. That’s vanity.)
Dig into the data. Where are they dropping off?
Talk to your customers. There is a mountain of evidence regarding what your shoppers actually care about, but you’re ignoring it because it’s harder to find than a CPM metric.
Conduct user research. You need to identify what motivates your buyer, and what other options they have. Watch people struggle to understand why they should buy from you, and why they need to buy from you now. It's painful to watch, but it's the only way to learn.

The "Oh Crap, That Sounds Expensive" Part

I can hear your brain spinning already. "This sounds great, but I’m already stretched thin."

You’re right. Doing this properly isn't a side hustle for your marketing intern. To run a legitimate experimentation program that actually drives revenue, you typically need a "pod."

A Product Manager to own the roadmap and the research.
A Designer to build the variants so they don't look like trash.
A Developer to code the tests without breaking the checkout.
A Data Analyst to tell you if you actually made money or just noise.

If you are doing over $100 Million in online revenue, stop reading this and go hire that team immediately. They will pay for themselves in six months. Seriously, go do it.

But if you are in that $3M to $100M range? That headcount is a killer. You’re looking at $500k+ in salaries just to get started. That’s a tough pill to swallow when you’re trying to keep margins healthy.

There Is a Better Way

This is exactly why Mobile1st exists.

We don’t just "do CRO." We act as your Digital Product Growth partner. For less than the cost of one mid-level employee, you get the whole squad: PM, strategy, design, dev, user research, and analysis.

We own the Revenue Per Visitor metric. We do the research. We find what motivates buyers. We build the tests. We tell you what won.

Our clients like to say, "You guys just do the things that make the money."

It’s not rocket science, but it requires focus, expertise, and a refusal to rely on the HiPPO.

The Bottom Line

You are fighting a war on CAC with one hand tied behind your back.

If you’re going to obsess over new creative to decrease your acquisition costs, you owe it to your business to obsess over new site experiences to increase your revenue per visitor.

Stop letting your ads write checks your website can’t cash.

How we can help you right now

Does your site feel like a leaky bucket?

If you want to see exactly how much money you're leaving on the table, we can calculate that for you.

How to Keep Your E-Com Director Job (And Maybe Get Promoted) in 2026

2026-01-14T00:00:00Z

Here we are in 2026, and the game has changed again. The board wants profitability yesterday, your budget is tighter than a pair of skinny jeans from 2011, and you aren’t getting any more headcount.

If you want to keep your seat at the table, or hell, maybe even snag that VP title, you need to stop doing what worked three years ago.

Here is the unvarnished truth on what I think you need to master this year.

1. Stop worshipping at the altar of "Acquisition"

For the last decade, E-commerce Directors were basically just glorified media buyers. If you could spend $1 to make $3, you were a god.

Those days are over. The enemy isn't your competitor anymore; it's the ad platforms eating your margin alive.

To survive 2026, you have to pivot to profitability. You need to obsess over LTV (Lifetime Value) like your rent depends on it. If you are still high-fiving your agency over top-line revenue while your repeat purchase rate tanks, you’re walking a dead man’s plank.

2. Profitability is the only metric that matters

I don’t care what your ROAS is. I really don’t.

If you can’t tell me the Contribution Margin of your campaigns, you aren’t running a business; you’re running a charity for Google and Meta. If you want to build the model that actually shows you this, read the ultimate e-commerce profitability model.

You need to understand the profitability levers of your entire business—even the ones you hate, like logistics, returns, and COGS. If you don't understand how a 2% increase in return rate destroys your campaign's profitability, you’re going to get blindsided.

3. Get a grip on User Behavior (Stop blaming GA4)

The moment Universal Analytics died, most of you stopped understanding how people actually use your site.

GA4 is actually more powerful than Universal Analytics; you just haven't set it up right.

Want to measure which filters are being used? Stop guessing and set up the custom event.
Want to know if driving traffic to a specific PDP actually impacts profitability? Hook GA4 up to your data warehouse.

Whether you are collecting data in GA4, a warehouse, or a CDP, you need to actually get the data out to make decisions. Our Business Intelligence product is built to do exactly this. And before you say the interface is too hard: Use AI. It will write the queries for you. No more excuses.

4. You aren't Steve Jobs (Test everything)

The biggest threat to your job is the HiPPO (Highest Paid Person's Opinion). Stop guessing what customers want. You aren't that smart. Neither is your CEO.

Know thy customer: Actually talk to them. Do the user research. Understand the personas.
Test your site: If you aren't running on-site experiments, you are leaving free money on the table. Why are you obsessively testing your ad creative, but not your site?
Creative Velocity: Meta and TikTok need more creative than ever to find the perfectly matched persona. These powerful bidding mechanisms now do the targeting for you, but you need enough varied creative to help them find the right people.

5. Use AI to bypass your busy Dev Team

You aren't getting more employees this year. Your dev team is "sprinted out" and won't get to your ticket until Q3.

Use AI to write the code snippets, troubleshoot the tags, and automate the boring parts of your workflow. Be dangerous enough to solve technical problems when the tech team is unresponsive.

AI is amazing at writing SQL. Need to dissect data but the analyst is busy pulling reports for the CFO? Describe the model to your favorite LLM and it will create amazing SQL for you. Seriously. You should no longer be overwhelmed by technical problems.

Side note: Don't get caught up in the hype. AI search/shopping is cool, but it’s going to be less than 5% of your revenue this year. Treat it like a burgeoning channel, not a savior.

6. Rent the talent, don't buy it

Since you can’t get approval for a $150k FTE, stop trying.

But you can usually get approval for a partner budget. Stop trying to hire "unicorns" internally who claim they can do design, code, and strategy. They usually do all three poorly.

Hire partners to fill the skill gaps. Bring in a sniper (like a specialist agency) for specific growth goals rather than hiring a generalist infantryman who you have to train and manage.

7. Let go: you’re less in control of your promotion than you think

Getting promoted in 2026 has more to do with interest rates and macro-economic conditions than your own competency.

You can be an absolute rockstar, but if the market is trash, the promotion isn't happening. Accept this. It will save you a lot of heartache. Control what you can control (see points 1-6).

8. Go home at 5:00 PM

This is serious. Don’t work more than 40 hours a week. It’s not worth it.

There is a law of diminishing returns in knowledge work. If you burn out, the company will replace you in two weeks. Protect your sanity. Have fun. If you aren't laughing at the chaos, you're in the wrong industry.

Stop Building Online Vending Machines

2026-01-21T00:00:00Z

Almost every e-commerce sites doing between $10M and $100M are basically just high-resolution vending machines.

You know the look. You land on the site and it’s just rows of product cards. Dresses, pants, accessories, repeat. It’s efficient, sure. But it’s sterile.

Then, usually buried somewhere in the footer, there’s a link to an "About Us" page or a blog that talks about the brand’s "lifestyle" and "mission." The “About Us” page, might event talk about charities the company supports.

Unfortunately, though Less than 5% of your traffic ever gets to the About Us page.

If your brand story lives in the footer, it doesn't exist. You’re running a catalog, not a community. And in 2026, catalogs get crushed by Amazon.

The Great "Friction" Lie

I’m going to say something that might get me kicked out of the CRO club: Friction is rarely your problem.

We’ve run thousands of experiments, and the ones focused solely on "reducing friction" lose about 90% of the time. The better move is personalization — making the site feel like it knows who’s on it, and giving them a reason to connect with the brand before they decide to buy.

Your customers aren't stupid. They know how to checkout. The reason they aren't buying isn't because they can't find the button; it’s because they don't care yet. They want to know if the product solves their specific problem ("Is this teacher-appropriate?") and if they trust the brand.

You answer those questions with context, not by shaving 0.2 seconds off your load time.

From Catalog to Cult: The "Outdoor Voices" Effect

Remember early Outdoor Voices? They didn't just sell leggings; they sold the "Doing Things" club.

They hosted runs, took photos of real people, and splashed those gritty, sweaty, happy photos all over the site. When you bought the hat, you weren't buying a hat, you were buying entry into the in-crew. You saw those hats everywhere and you felt part of something.

You don't need a VC budget to do this. You just need to stop hiding your community in the blog and start putting it where the money is.

How to Kill the Vending Machine (Without a Redesign)

You don't need a massive re-platform to fix this. You need to change how you talk, where you put your content, and how your team meets on Mondays.

1. The "First Order" Copywriting Rule

Go read your top-selling Product Detail Page (PDP). Does it sound like a corporate slog written by an SEO agency? "This garment features a poly-blend conducive to..."

Boring.

Rewrite it in the Founder’s Voice. Go back to how you sold the product when you got your very first order. Why did she design this? Did she find the pattern in a vintage market? How does she wear it? When you swap "product specs" for "founder stories," you stop selling a commodity and start selling a connection.

2. The "Below the Fold" Lifestyle Layer

There is a fear that adding "community" content will distract from the Buy Button. That’s nonsense.

Keep the top of the page clean for the transaction, but as they start scrolling, that is your invitation to show them how the product fits their life.

Don't just show the dress. Show the "Teacher's Edit." Show the UGC photos of real customers wearing it in the classroom or at Sunday service. Prove to them that this product belongs in their specific world.

3. Merchandise by Lifestyle

Yes, it’s important to be able to shop by dresses, pants, skirts and shirts… But that’s not the only way people shop. They also think, where will I be wearing this outfit? Or what will I be doing?

You can merchandise in the navigation, on the home-page and even on the PDP by where will this skirt be worn? For Teachers, or Best for Church, or even Moms’ Favorites.

When you merchandise this way, shoppers start to see themselves in these groups and think okay, this will solve my problem. Make these lifestyles front and center as people browse.

4. Search means your Merchandising is Broken

This is a hard pill to swallow, but if a user has to use your Search bar or even your Hamburger menu, you have failed them.

Think about mobile. The navigation is hidden behind an icon. If a user has to tap that icon to find what they want, it means the page they were just on, the PDP, didn't give them a clear path to browse.

Your PDP should be a browsing engine. If they are looking at a floral dress, don't make them search for "more floral." Guide them effortlessly to the "Floral Shop" or the "Maxi Collection" directly from the content they are already consuming.

5. Pivot the "Dreaded" Weekly Meeting

Every Monday, your team probably sits in a room (or a Zoom) and reads a spreadsheet of numbers out loud. Everyone hates this meeting.

Keep the meeting, but change the agenda. Instead of just reporting what happened, share why it happened.

Your Social Team knows that a specific video of a mom packing a lunchbox went viral. Your Ad team knows that "Teacher" messaging is outperforming "Sale" messaging. Those assets usually die in the ad account.

Force the teams to swap those wins. Take the viral lunchbox video and put it on the PDP. Take the "Teacher" messaging and put it in the Founder’s Note.

The Bottom Line You have a story. You have a founder. You have customers who love you. Stop hiding them behind a grid of 400 silent SKUs.

How to steal budget for an experimentation program(me) and look like a genius

2026-01-29T00:00:00Z

I love words that are spelled differently in British English and American English. One of my favourites is colour. There are many words that I somehow learned to spell the British way, and now that’s just my default.

I know the feeling.

Walking into the CFO’s office to ask for more marketing budget feels a lot like being a teenager asking your dad for gas money after you already dented the bumper. It’s painful, it’s awkward, and usually, the answer is a hard "no" followed by a lecture on fiscal responsibility.

But here’s the thing. You probably don’t need new money; you already have the budget, you just aren’t quite sure where to look.

The marketing budget trap

If you’re running a brand doing anywhere between $3 million and $100 million, I can pretty much guess your P&L structure without looking at it. You’re spending somewhere between 10% to 30% of your revenue on marketing. If you’re VC-backed, it’s probably higher, and your burn rate makes me sweat, but that’s a problem for future-you.

You love paid media. We all do. It’s the safe drug. It’s predictable. You put a dollar into the Meta or Google machine, and you hope to get four dollars back. Your ROAS is likely sitting right around 4x. It’s respectable. Everyone in the building understands the model, from the intern to the board.

But deep down, you know the truth. Ad costs are only going one direction, and your efficiency is plateauing.

How to steal the money

You know you should be allocating 10% to 20% of your budget to test new channels. Maybe it’s TikTok, perhaps it’s Connected TV, maybe it’s influencers.

Here is the unlock, though. Stop thinking of "On-site Experimentation" as a separate line item. And unlike TikTok, this channel makes every other channel perform better. Our Data-Driven Experimentation program is exactly what that budget should fund.

Let’s do some napkin math on a hypothetical $7 million store.

If you’re spending 15% on marketing, that’s about $1.05 million a year. If you are doing your job right, you’re allocating 20% of that for "Testing." That means you have roughly $210,000 sitting there, waiting to be torched on a new ad platform that might not convert or pan out this year.

You don’t need $200k to start onsite experimentation. You can run a killer Digital Product Growth program for under $75k a year. That money already lives in your "Test New Channels" bucket. You need to move it from one column to another.

Please don’t try to hire this out.

Now, usually, this is the part where someone asks whether they can hire a full-time person at $75k.

Please don’t do that.

For $75k, you get a junior marketing coordinator who knows a little HTML and will accidentally delete your checkout button on Black Friday. You cannot hire a full team—Strategist, UX Designer, Developer, QA, Analyst—for that price. But you can hire a partner (hi, that’s us) for that price. And we won't break the site.

Proving you aren't an idiot

Okay, so you’ve stolen the budget. Now, the future you will have to prove to the boss that it wasn't a mistake.

Most agencies will bore you to death talking about Conversion Rate. We don't care about Conversion Rate. It’s a vanity metric. I can double your conversion rate tomorrow by lowering all your prices to $1. You’ll go out of business, but your chart will look great.

We care about Revenue Per Visitor (RPV). That’s the only number that pays the bills. If you want to understand exactly what drives RPV, the hierarchy of e-commerce insights is the best place to start.

Let's look at the impact on that $7 million business.

Let's assume we run a solid roadmap. We aren't promising magic buttons, but let's say we get 6 winning experiments in a year. If each win increases your average RPV by 3%, it may seem small.

It’s not. 3% is significant when compounded.

But here is where the corporate calculators get it wrong, and where we need to get real.

First off, you don’t get all 6 wins in January. You get one in February, one in April, one in June, and so on. The first win generates extra cash for 11 months, while the last win might only generate it for one month.

Secondly, winning experiments don't last forever—user behavior changes. Competitors change. That shiny new feature becomes background noise. We conservatively estimate that a winning test degrades by about 15% every 3 months. Why? Because entropy is real, and humans get bored. This is precisely why experimentation isn't a "one-and-done" project. You have to keep feeding the beast.

So, is it worth it?

When you factor in the timing delay and the degradation, what are you actually left with?

On a $7 million business, a program like this can conservatively generate an additional $800,000 in year-one revenue.

That means you spent under $75k (which you diverted from the ad budget) to recover $800k. That is a return better than 10x.

And the best part? That $800k lift occurs on-site. Which means your Google Ads work better. Your email flows convert higher. Your ROAS increases across the board because you fixed the bucket rather than just pouring more water in.

You have the budget. It’s sitting right there, likely about to be spent on a "Brand Awareness" campaign that nobody can track. (I actually really like that brand awareness campaign, to be honest. It’s probably better than the retargeting campaign of people who were already going to convert.)

Take the money. Bet on your own product.

If you’d like to see precisely what this ROI model looks like with your specific traffic and revenue numbers, let's chat. I’ll run the calculator for you, and we can present it to your CFO together.

The 6 Advanced E-commerce Reports To Build a $100M Brand

2026-02-05T00:00:00Z

If you’re running an e-commerce brand doing $5M, $20M, or $80M, I’m going to tell you something your Facebook ads agency won't: Your ROAS is lying to you.

It’s the great industry secret. We’ve all been conditioned to stare at the Shopify dashboard, see that "Total Revenue" line go up, and feel like geniuses. But behind the scenes? Your margins are getting punched in the throat by rising CAC, your inventory is gathering dust like a 1980s workout VHS, and you’re probably sending 20% discount codes to people who would have paid full price.

If you want to move from "scaling by luck" to "growing by design," you need to stop looking at vanity metrics and start looking at these six reports. They aren’t sexy, but they are the difference between a business that’s a cash-flow machine and one that’s just a very expensive hobby.

1. The View to Buy Report (The "Leaky Bucket" Detector)

Most people stare at their site-wide conversion rate. That’s like checking the average temperature of the entire Atlantic Ocean to decide if you should wear a wetsuit in Miami. It’s useless.

The View to Buy Report is the antidote. It zooms in on your Product Detail Pages (PDPs) to see who is actually interested versus who is actually buying.

$$View to Buy Rate = \frac{Unique Purchases}{Unique Product Views}$$

Look, if you have a product getting 10,000 hits a month but only 50 people are buying it, your marketing team isn't the problem. The page is. You’re paying for a crowd to enter the store, but when they get to the shelf, something is turning them off—maybe the photos look like they were taken on a flip phone, or the "Add to Cart" button is playing hide-and-seek.

As we’ve argued before, The View to Buy Report is the one report that will save your BFCM week because it tells you exactly where your ad spend is being set on fire. If a product has a low View to Buy rate, stop throwing money at it. Fix the page first, then turn the tap back on.

2. Product Contribution Margin (The "True Profit" Report)

If I could have any superpower, it wouldn’t be flying; it would be omniscience. I want to see the hidden mechanics of why a business actually works. In e-commerce, that "ultimate truth" is product-level profitability.

Most people look at a "Bestseller" and see a hero. But until you break it down, you're just looking at a reflection of reality. You need a report that follows the money out as carefully as it follows the money in.

$$Contribution Margin = Revenue - (COGS + Shipping + Variable Ad Spend)$$

We’ve built systems that pull daily data from UPS, Shipstation, Meta, and Google to find this number. But you should start with a simple, ugly, manual spreadsheet. When you see your "Hero SKUs" side-by-side with their real costs, it changes you. You’ll find that some of your high-volume items are quietly burning cash because the return rates are nightmares or the shipping eats the margin alive.

Stop being a "revenue at any cost" junkie. When you find the products that have a high Gross Margin per Visitor, those are the ones that deserve the paid support. The rest? They’re just "Profit Vampires" that belong in the footer.

3. Merchandising & Inventory Health

Inventory is just cash that’s sitting in a warehouse, unable to buy you a margarita. This report bridges the gap between the marketing team (who wants to sell everything) and the ops team (who has to deal with the physical reality of the warehouse).

You need to track your Weeks of Supply (WOS). Running ads for a product that only has 2 weeks of supply left is just a great way to pay money to frustrate your customers. It’s a fireable offense.

On the flip side, if you have 40 weeks of supply sitting there, it’s time to get aggressive. Bundle it, discount it, or give it away as a "gift with purchase" to move the needle. Clear the shelf so you can reinvest that cash into something that actually turns over.

4. Product "Gateway" & Affinity

Not all customers are created equal. Some buy a $10 accessory once and never come back. Others buy a specific $80 jacket and end up spending $2,000 over the next three years.

The Gateway Report identifies which specific products act as the best entry points for high-LTV (Lifetime Value) customers.

The "unsaid" truth here? Your most popular first-purchase product might be a "one-and-done" item. It brings in "tourists," not "citizens." You want to find the products that lead to the second purchase. Use those Gateway products for your Top-of-Funnel ads. Even if the initial cost to acquire the customer is higher, the long-term math wins every time.

5. Customer Discount Sensitivity Matrix

Stop being a "coupon crack" dealer. If you send a "20% Off Everything" blast to your entire email list, you are effectively handing cash to people who were perfectly willing to pay full price.

This report segments your audience by their psychological need for a discount. You’ve got:

The Loyalists: They love the brand and buy regardless of price. Stop discounting for them. Give them "Early Access" or "VIP Status" (which costs you $0).
The Bargain Hunters: They only show up when the "SALE" sign is in the window.

Treat your margin like the finite resource it is. Save the deep discounts for the people you’re trying to reactivate or to move that stagnant inventory we talked about in the inventory report.

6. User Cohort with Predictive Revenue

This is where we stop looking in the rearview mirror. Most founders scale until they run out of cash because they don't understand their payback period. They see a low ROAS today and panic, not realizing that specific cohort of customers will be worth 4x that in six months.

Using BTYD (Buy 'Til You Die) modeling, you can project the future revenue of your existing customer base without spending another dime on acquisition. This is the "CFO’s Crystal Ball."

If you know a customer acquired today for $50 will be worth $200 by Christmas, you can stop sweating the daily fluctuations of the Meta algorithm and start outspending your competitors. It’s about scaling with confidence instead of crossing your fingers.

Using this report gives you additional budget to find more new customers, because you can rely on existing ones to come back every month.

Go Build These Reports

If you’re stuck between $3M and $100M, you can’t "hustle" your way to the next level. You can’t just copy what your competitors are doing because, frankly, they’re probably guessing, too.

Most agencies will give you a PDF of "best practices" and some "button color tests." We’d rather look at your View to Buy report and tell you why your site is failing your media spend. Our Business Intelligence product is built to surface exactly these kinds of answers. It’s not always the polite conversation you’ll get at a corporate board meeting, but it’s the one that grows your revenue per visitor.

Ready to see what's actually happening under the hood?** Reach out**, we have products to build these reports for you, or guide your team.

How to Prioritize Experiments

2026-02-12T00:00:00Z

If you’re running a brand doing $20M or $50M, you know the drill. Your backlog isn't a strategy; it’s a graveyard of good intentions. You’ve got the CEO who saw a cool widget on a competitor’s site at 2 AM, the Lead Engineer who is obsessed with a technical spec no customer understands, and a Marketing team that thinks a new font is going to save the quarter.

Most agencies will tell you to "just test everything." They’ll sell you a retainer, run sixteen color-change tests on a button nobody clicks, and send you a pretty slide deck at the end of the month that says a whole lot of nothing.

At Mobile1st, we stopped playing that game. We realized that if we wanted to actually move the needle for brands in that $3M to $100M "growth trap," we had to stop being "yes men" and start being the bouncers of the roadmap. We prioritize experiments like we’re triaging a battlefield. If it doesn’t meet the criteria, it doesn’t get on the site. Period.

Here is exactly how we decide what lives and what dies.

Rule 1: The 60% Revenue Floor (The "Who Cares?" Filter)

I’m going to say the unsaid: Most of the stuff your team wants to change simply doesn't matter.

If an experiment only impacts a tiny corner of your site, like a "refer-a-friend" link buried in the footer or a specialized landing page that gets 2% of your traffic, we don’t care. I don’t care how "strategic" it feels.

At Mobile1st, if an experiment doesn't touch at least 60% of your total revenue flow, it goes to the bottom of the pile. We call this the "Who Cares?" filter.

Why 60%? Because if you’re doing $50M a year and you find a 5% lift on a page that only 10% of your users see, you haven't moved the needle. You’ve just wasted three weeks of development time to buy yourself an expensive coffee. You need to be hunting where the money is. We look at the checkout flow, the high-traffic PDPs, and the global navigation. If it doesn't have the scale to actually change your bank account balance, we aren't touching it.

Rule 2: The Hierarchy of Insights (Data Over Ego)

The second way we kill bad ideas is by looking at the quality of the insight behind them.

Every brand has a "Random Idea Generator." Usually, it’s the person with the highest salary in the room. They say, "I think we should move the search bar."

We don't care what you "think." We care about our Hierarchy of Insights. (If you haven't seen it, check it out here: /news/the-hierarchy-of-e-commerce-insights).

If your "insight" is just an opinion, it’s at the bottom. If it’s a best practice from a blog post? Still at the bottom. But if you show me a heat map where users are rage-clicking a non-functional element, or a user recording where a customer literally says, "I can't find the shipping price," now we’re talking. If we have alrady tested a similar concept on a different part of the site, we make the experiment important. That is the best insight we could ever ask for.

When you have high-quality insightsthat experiment jumps to the top of the stack. We aren't testing to see "if" something works; we're testing to solve a problem we know exists.

Rule 3: The 70/30 Split (Optimization vs. Transformation)

This is where most experimentation programs go to die: they get stuck in the "Optimization Loop."

You change a headline. You move a picture. You tweak a button color. You get a 1% lift here and a 2% lift there. It feels good, but it doesn't change the trajectory of the business.

We believe that 70% of your experiments should be optimization. You have to protect the house. You have to keep refining the current experience to squeeze every cent out of your traffic.

But the other 30%? Those need to be "Business Changing Opportunities."

These are the big swings. The experiments that challenge your entire business model or how you present your value proposition. These are the tests that make people in the office nervous.

I once had a Lead Engineer try to kill an experiment because he thought the specs he wrote were the hero of the page. He spent months on the cooling fins of a product. I told him nobody cared about the fins; they cared if it would fit in their car. We deleted his "innovative" specs and replaced them with a "What's in the Box" list. RPV went up 23%. That wasn't just an optimization; it was a realization that the way the company talked about itself was wrong.

Rule 4: Value Over Speed (The Anti-Friction Rule)

Here is where we really start to piss off the "CRO Gurus."

For the last decade, the industry has been obsessed with "reducing friction." Every blog post tells you to make the checkout faster, remove steps, and get them to the "Thank You" page in three seconds.

We do the opposite. If an experiment is just about reducing friction to help people get to checkout faster, we deprioritize it.

Wait, what?

Think about it: Making a bad experience faster just helps people leave your site sooner. Speed doesn't build brands. Value builds brands.

If an experiment improves the perceived value of your product or helps a customer find the exact right product for their specific needs, it goes straight to the top of the priority list. We want to spend our time helping the customer understand why they should pay you full price, not just making it easier for them to click "Buy" on something they aren't sure about.

Similarly, if we can personalize the experience for almost all users—not just a tiny segment—we do everything we can to test that right away. Personalization that scales is a revenue multiplier. Our Data-Driven Experimentation program is built around exactly this prioritization framework.

Stop worrying about making the "Buy" button 0.2 seconds faster. Start worrying about why the customer is hovering over it and not clicking. Are they confused? Do they feel the value? Do they know this is the right version for them? Solve for value, and the conversion will follow.

The Cold Truth

If you’re leading an e-commerce team right now and you’re still building your plans around "hacks" and "speed tricks," you're playing a game you've already lost. The "vending machine" era of e-commerce—where you put a dollar in Meta and get three out—is dead.

You win now by being more disciplined than your competition. You win by being the person who says "no" to the low-impact friction tweaks so you can say "yes" to the 60% revenue drivers and the big value-based swings.

If you’re tired of the "invisible grind" of running tests that don't matter and managing an agency that’s just a "yes man" for your CEO’s bad ideas, maybe it’s time for a different approach.

We aren't here to be your friends in the boardroom. We're here to be the ones who tell you when your baby is ugly so we can help you build something that actually grows.

Let’s stop "optimizing" and start winning.

Ready to grow RPV by prioritizing the right experiments Reach out.

Your Offer Sucks. Stop Obsessing Over Friction.

2026-02-18T00:00:00Z

Why "optimization theater" is killing your growth and the one thing you actually need to fix.

Let’s be real for a second. Imagine we’re sitting at a bar, I’ve just bought the second round, and you’re complaining about your conversion rates.

You tell me you’ve tried everything. You’ve moved the "Add to Cart" button above the fold. You’ve made it sticky so it stalks the user like a desperate ex. You’ve A/B tested three shades of green because some guru on LinkedIn said "Forest Green" conveys trust.

And yet? Your revenue per visitor (RPV) is flatlining.

Here is the hard truth that most agencies won’t tell you because they want to charge you for billable hours tweaking pixels: Friction doesn’t matter.

Nobody, and I mean nobody, ever landed on a website, saw a product they kind of liked, but left because the add to cart button was just below the fold.

The problem isn’t friction. The problem isn’t the UI.

The problem is your offer.

The "Optimization Theater" Trap

We call what we do "Digital Product Growth" here, but let's strip the jargon. It’s about making money.

If you are leading an e-commerce brand doing $50M a year, you aren't losing sleep over pixel padding. You are losing sleep over the fact that 60% of your traffic lands directly on a Product Detail Page (PDP), and most of them bounce.

Think about that. They didn't come through the homepage. They clicked an ad or a Google link because something, somewhere, promised them a solution to their problem. They arrived with intent. They want to buy.

If they leave, it’s not because they couldn’t find the checkout. It’s because the math in their head didn't add up.

The Equation is Simple:

Perceived Value > Price Asked

That’s it. That’s the whole game. If the customer thinks your product is worth more than the cash you’re asking for, they buy. If they don't, they leave.

The Commodity Crisis (aka "The Competitor Down the Street")

Here is the "unsaid" part that hurts to hear: Your product probably isn't that special.

I know, I know. You spent years developing it. But to the customer? You’re selling a widget. And the guy down the street is selling the exact same widget.

Maybe you think your "Unique Selling Proposition" (USP) is your free shipping. Unfortunately, everyone has free shipping. Amazon ruined that for us a decade ago.

Maybe you think it’s your 30-day return policy. Again, unfortunately, that’s table stakes, not a perk.

If you and your competitor are selling a similar product at a similar price with similar shipping terms, you are in a race to the bottom. You are a commodity. And commodities compete on price, which is a great way to bankrupt your business.

So, how do you win without slashing your margins? The answer is building a real offer — which is one of the four things standing between your traffic and your revenue.

Constructing "The Offer" (The Secret Sauce)

You need to stop selling a product and start making an offer.

An offer isn't just the item in the box. An offer is the complete package of value that makes the customer feel like they are ripping you off.

To build an irresistible offer, you need to add something that has high perceived value to the customer but low actual cost to you.

Let’s look at some examples of how to turn a boring product into a killer offer:

1. The Tennis Shop

The Product: A $200 Babolat racket.
The Competitor: Sells it for $200.
The Offer: You sell the racket for $200, but you include a "Pro-Ready Prep Package." This includes a premium grip (Cost to you: $2) and free custom stringing if they buy the strings (Cost to you: 15 minutes of labor).
The Result: The customer isn't buying a racket; they are buying readiness. They save time and hassle. You win.

2. The Golf Retailer

The Product: A new set of irons.
The Competitor: Sells them for $999.
The Offer: You sell them for $999, but you throw in a "Virtual Range Session" guide (a PDF or video series you made once) or a voucher for a free 30-minute hitting session at a partner indoor range.
The Result: You aren't selling clubs; you're selling a better golf game.

3. Women’s Fashion

The Product: A summer dress.
The Competitor: Sells a similar dress.
The Offer: You include a "Self-Care Sunday" kit—a branded facial mask and a sleep scrunchie.
The Cost to You: Maybe $1.50 at wholesale.
The Value to Her: "Oh, this isn't just a dress. This is a treat."

If You Have It, Flaunt It

Here is where most Directors of E-commerce fumble the ball. They create a decent offer, and then they hide it in the bullet points below the fold.

Do not be subtle.

If you are giving something away, scream it from the rooftops.

Put it in the Hero Image: The main photo on your PDP shouldn't just be the product; it should be the product plus the free gift, with a badge that says FREE.
Jam it in the Buy Box: Right next to that price, list the bonuses.
Use the Magic Word: "Free." It triggers a dopamine hit that "Complimentary" or "Included" just can’t touch.

"But I Can’t Afford Free Stuff!" (The 'Broke' Excuse)

Okay, fine. Maybe your margins are razor-thin. or maybe your warehouse manager threatens to quit every time you try to add a new SKU to the pick-and-pack list.

I get it. (Actually, I think you're making excuses, but let's play your game for a minute).

If you absolutely cannot add new value to the box, you have to uncover the value that is already there.

Your perception is your reality, but your customer’s perception is the only one that pays the bills. So, if you can’t add more, you need to "Explode the Offer."

Most brands are lazy. They list a "12-Piece Golf Set" as a single bullet point. That is one line item. One line item feels small. One line item feels expensive.

Stop making people guess what is in the box.

Break it down. List. Every. Single. Item.

Don't say "12 Clubs."
List the Driver. The 3-Wood. The Hybrid. The Putter. The Sand Wedge.

Suddenly, that one bullet point becomes twelve lines of value.

And don't stop at the hardware.

Does it come with a piece of paper telling them how to use it? That’s not "Instructions." That is a "Quick-Start Success Guide."
Does it come with batteries? That’s "Ready-to-Use Power Included."

Take what you already have and expand it until it takes up serious visual real estate on the page. Make the list look long. Make the offer look "heavy."

If you can't give them more, make sure they know exactly how much they are already getting.

The Bottom Line

If you are staring at your dashboard trying to figure out how to increase revenue per visitor and nothing is working, stop looking at your code. Stop looking at your navigation bar. Our Offer Creation program is built specifically for this.

Start looking at your value proposition.

It is rarely friction that kills a sale; it is apathy. It is the feeling that "I can get this later, or somewhere else."

Your job is to make the value so overwhelming that they feel stupid saying no.

So, here is my challenge to you: Go look at your top-selling PDP. Ask yourself, "If I were a stranger, would I feel like this is a steal?"

If the answer is no, we need to talk.

Ready to grow RPV by prioritizing the right experiments Reach out.

How E-commerce Directors Should Actually Be Using AI Right Now

2026-02-26T00:00:00Z

If you spend more than five minutes on LinkedIn right now, you’re being bombarded by tech bros and AI evangelists telling you that you need to become a full-stack developer overnight. They want you spinning up Claude Code, deploying custom apps, and fully automating your e-commerce operations.

Let me just say That is a terrible idea.

If you are an E-commerce Director at a company doing $3M to $100M, you should not be building custom software that your team will have to maintain forever. Code rots. APIs change. You are not a CTO managing technical debt; your job is to drive revenue.

But, pretending AI doesn't exist is a fast track to irrelevance. At our digital product growth agency, we use AI relentlessly to run better on-site experiments and make our clients money.

So, I wanted to write a vulnerable, "working document" of where I actually think e-commerce leaders should be deploying AI right now to get shit done.

1. Creative for the "Andromeda" Era of Meta

As we all know, targeting is dead. Creative is the new bidding. But unless you’re sitting on a massive enterprise budget, none of us have the creative teams required to keep up with the velocity Meta demands.

There are a ton of tools emerging to help close this gap. Google just launched Pomelli (currently in Labs), and it's built for exactly this. You feed it your brand DNA and your current product photos, and it spits out a barrage of creative ready to be tested on Meta.

This area is evolving fast, but here are a few other tools to add to your arsenal:

Photoroom: Absolute magic for bulk-removing backgrounds and placing your SKUs into high-end, AI-generated lifestyle scenes.
Pencil (by Brandtech): Generates predictive ad creatives based on what actually works for your specific sector.
Midjourney: Still the undisputed king for raw, high-fidelity lifestyle image generation if you learn how to prompt it right.

2. Advanced Data Analysis (The "Python" Trick)

GA4 is fine for the basics; knowing how people find you or what pages they visit. You should definitely be using the AI tools built into GA4 to quickly confirm hypotheses (if your tracking is actually set up right, which is a whole other rant).

But what happens when you export data from your CRM or ERP and it’s a million lines deep?

Trying to paste massive datasets into ChatGPT and watching it crash, or trying to do complex row-by-row calculations in Excel until your computer catches fire.

Don’t use AI to do the analysis. Ask AI to write a Python script to do the analysis for you.

Here is a concrete example from my own desk: I had a file with a million order rows (with addresses) and another file with 80 physical store locations. We needed to geocode the orders, calculate the distance between every single order and the closest store, and map it on a graph.

I dropped those files into Cursor (Claude Code works too) and asked it to build the script. After a little back-and-forth clarification, I had the Python code in 20 minutes. Funny enough, the script took longer to run (2-3 hours) than it took me to "write" the code.

Before you panic and say, "I don't know Python, I could never do that," listen to me: Let the AI guide you. It will feel intimidating the first time you open a terminal. But once you see that graph pop out, it is incredibly addicting.

3. Creating "Version Zero" for Your Devs & Designers

One of the hardest bottlenecks in development is simply explaining to developers or designers the exact change you want to make. We've all been in that meeting where you draw a wireframe on a napkin, say "make it look like this," and pray they understand.

Stop explaining. Start showing. Depending on the size of the change, here is your toolkit:

For Quick Visual Tweaks: Use Chrome DevTools AI. Right-click the element you want to change, hit "Inspect," and open the AI Assistant. It will look like terrifying HTML, but you just literally tell the AI what you want changed. It will rewrite the page locally just for you. You can even copy the new CSS/HTML to give your devs. It doesn't need to be pixel-perfect; it just needs to say, "Hey, this is exactly what I mean."
For Advanced Features: Use Cursor. Tell it you want to make changes to a specific page. It will download the necessary HTML/CSS and you can describe the exact interactive changes you want to test. It takes a bit longer, but it's perfect for complex UI shifts.
For Brand New Pages: Use v0 (by Vercel). If you are building a landing page from scratch to test a new offer, you can prompt v0 to build a pixel-perfect, fully coded UI in seconds. They look incredible and can serve as the ultimate "Version Zero" wireframe.

These are just a few ways I'm using AI every day to cut through the noise, skip the endless meetings, and actually launch experiments that drive revenue. If you're wondering how to structure the experiments themselves, our Data-Driven Experimentation program is where the rubber meets the road. (If you want to read more of our rants on digital product growth, check out our all our insights.).

What did I miss? What else should E-com Directors be doing in the trenches?

If I got something wrong, send me a message on LinkedIn.

The Four Things Standing Between Your Traffic and Your Revenue.

2026-03-05T00:00:00Z

The same four conversion killers show up on almost every e-commerce site we work on. Here's what they are. Here is what we do about them.

I talk to e-commerce directors all the time, and honestly? The conversation almost always goes the same way. They're doing a ton of things right: great ads, solid product, real investment in traffic. But the revenue per visitor just isn't where it should be, and they can't quite figure out why.

After years of doing CRO for e-commerce brands, we've seen the same four culprits come up over and over again. They're not glamorous problems. But fixing them moves the needle. Here's what they are. Here is how we tackle each one.

Problem #1: Your Ads Are Bangers. Your Landing Page Has No Idea.

Your team is crushing it on Meta. Hundreds of creatives, each one hitting a different pain point, speaking to a different type of person. Someone sees an ad about durability, they click. Someone else responds to the "perfect gift" angle, they click too. The ads are working.

Then both of those people land on the exact same product page. Same headline. Same image. Same everything. The emotional connection your ad just built? Snapped. Whatever resonated enough to earn that click just got ignored the moment they arrived.

This happens on basically every e-commerce site we've ever audited. It's one of the most expensive disconnects in digital retail, and it's almost completely invisible until you go looking for it.

Our Solution: Ad-to-Page Dynamic Content Matching

We pull the creative ID from each Meta ad, download the actual creative, and analyze what message it's leading with. Then we automatically suggest edits to key slots on your product page: the sub-headline, the hero image, the value props, so the page reflects the ad that brought the visitor there. No new landing pages to build and manage. Just smarter, more connected experiences.

We've seen a 15% lift in ROAS from this. And you stay in full control of every version. Nothing goes live without your sign-off.

Problem #2: There's No Real Reason to Buy From You Over the Next Guy

Let's say someone finds a product they want. They're comparing you to another brand selling something almost identical at basically the same price. What tips them your way?

That's your offer, and most brands don't really have one. They have a product and a price. The offer is what makes the perceived value feel bigger than the price tag, and bigger than what the competitor is offering. Most brands reach for a discount when they need this. (We've written a full breakdown of how to build an actual offer — not a discount, but a reason to choose you.) And yeah, it works in the short term. But you can't discount your way to a healthy brand. Eventually it just trains your customers to wait for the sale.

Our Solution: Strategic Offer Development

We dig into what your customers actually value, the stuff that feels premium to them but doesn't cost you much to deliver. That could be a bundle, a guarantee, a how-to guide, a community, early access. It looks different for every brand. We find what clicks for your audience, build it out, and test it. The goal is a genuine reason to choose you that has nothing to do with being the cheapest.

Problem #3: Shoppers Are Landing on One Product and Have No Idea What Else You've Got

Over 60% of e-commerce traffic lands directly on a product page, not your homepage, not a category page. These are people coming from Meta ads or Google Shopping who've already done some homework. They're not browsing. They're evaluating.

Here's the thing though. The product they landed on might not even be the right one for them. Maybe they clicked on product A, but product B is actually a better fit for their situation. But if your product page doesn't guide them anywhere, they just bounce. They never find the thing that would have made them buy.

Our Solution: Intelligent On-Page Product Discovery

We design experiences that surface the right products at the right moment, so people stay on site and find what actually works for them. Not a "customers also viewed" widget slapped at the bottom. Real, contextual product guidance built into the page flow. Less bouncing, more converting, and an average order value that actually reflects your catalog.

Problem #4: Your Site Doesn't Know Anything About the Person Shopping on It

People want to feel like a brand gets them. But most e-commerce sites show the exact same experience to a 22-year-old beginner and a 45-year-old expert who's done this a hundred times. Same page, same messaging, same recommendations.

The tricky part is that you can't figure out who someone is just by watching them browse. Clickstream data tells you what they looked at. It doesn't tell you why, or who they are, or what they actually need. To personalize well, you have to learn more. And the only real way to learn more is to ask.

Our Solution: Zero-Party Data & Personalization

The industry calls it zero-party data, meaning info a customer gives you directly, usually in exchange for something useful like a recommendation or a discount. We build quick, low-friction quiz flows right into your site. A few questions, maybe 30 seconds. Then we use those answers to personalize the on-site experience, and more importantly, to power your email and SMS in a way that actually feels relevant. The difference between "here's our weekly email" and "here's something specifically for you" is huge.

Four Problems. Four Products. One Goal: More Revenue From the Traffic You Already Have.

None of these problems are about needing more traffic. They're about what happens once traffic arrives. Most brands pour money into acquisition and leave a ton of conversion on the table because the on-site experience isn't doing its job.

We've built four products, each one targeting one of these four levers. We use experimentation to test and validate everything, so you're not just changing things and hoping. You're learning what works and doubling down on it.

**If any of these feel familiar, let's get into it. You've already got the traffic. Let's make it work harder. **Find me on LinkedIn and we can solve these problems.

The Top 7 Ways to Use Personalization on Your Site, Right Now

2026-03-12T00:00:00Z

This Week on Checkin to Checkout

This week on Checkin to Checkout, Doug Villella from APG breaks down what it actually looks like to build an e-commerce operation from scratch inside a company that sells O rings — and why talking to customers every week is the only strategy that matters.

🎧 Listen to the full episode

A note from me: We've been quietly rebuilding Mobile1st.com behind the scenes, with the goal of better expressing what we do. I'd love for you to check it out and tell me what you think. And over the next few weeks you'll start to notice some small changes to how this newsletter is structured. Same content, better delivery.

Here's the uncomfortable truth most agencies won't tell you:

Your Shopify template is fine.

Friction isn't actually the problem. You've already fixed that. The buttons are big. The checkout is fast. The mobile experience is clean enough.

So why is revenue per visitor still flat?

Because we left the era of demand capture. Every tweak you made to reduce friction was demand capture logic. Assume the customer wants to buy, remove the obstacles, collect the money.

That era is over.

We're now in the era of demand generation. And that requires something completely different. It requires making someone feel something about your brand before they've decided to buy.

That's emotional connection. And you build it through storytelling and through personalization. We wrote about the four conversion killers most brands ignore — this shift from demand capture to demand generation is at the root of all four.

Talking to your customer based on what they've told you about themselves, or what they've responded to in the past, is the highest-leverage move left in e-commerce. It's the difference between a catalog and a conversation.

Before we get into the tactics, there's something we need to address. None of this works without the foundation underneath it.

The Prerequisite: Your Data Has to Be Clean

Personalization is only as good as the data powering it.

Most brands skip this conversation entirely. They hear "personalization" and immediately start thinking about tools and tactics. Then they implement something, it underperforms, and they blame the tactic. The real problem was upstream the whole time.

Here's what you actually need before any of this is worth building:

UTM hygiene. If your ad campaigns aren't properly tagged, you can't match landing pages to ad creative. You can't attribute traffic to the right source. You're flying blind and calling it personalization.

A consistent customer identity layer. Shopify gives you a customer ID. But are you connecting it to your email platform? Your ad platform? Your on-site behavior data? If those systems don't talk to each other, you have islands of data instead of a picture of a person.

Klaviyo (or equivalent) as your single source of truth. The affinity data, quiz answers, browsing behavior, purchase history. It's only useful if it lives somewhere central and feeds back into every touchpoint. For most Shopify brands, that's Klaviyo. Not because it's the only option, but because it sits at the intersection of on-site data, email, and SMS in a way that actually works.

You don't need a full CDP to get started. But you do need clean UTMs, a connected email platform, and some agreement about where customer data lives. Without that, the tactics below are mostly theoretical.

Get the foundation right. Then build on it.

1. Treat New Visitors and Returning Visitors Completely Differently

This is the most underused personalization lever in e-commerce. It requires almost no data infrastructure. And almost nobody is doing it.

A first-time visitor and a returning visitor have completely different jobs to be done.

The first-time visitor is asking: Can I trust this brand? Is this product actually for me? Why should I buy this instead of something else? They need education. They need social proof. They need the brand story up front.

The returning visitor is asking: Where was I? Did anything change? Give me a reason to finish. They don't need the origin story again. They need to pick up where they left off.

When you serve both visitors the exact same homepage, you're failing both of them. The returning visitor is bored by the intro. The new visitor is overwhelmed by the assumption that they already care.

The fix is straightforward. Detect the visit number and adjust the experience accordingly. For new visitors, lead with trust signals, brand story, and orientation. For returning visitors, surface recently viewed products, pick-up-where-you-left-off messaging, and anything new since their last visit.

It's not complicated. It's just not the default. And that gap is your opportunity.

2. Match the Landing Page to the Ad They Clicked

You probably have 400 social ads running right now.

5 personas. 10 emotional angles. 15 different messages about why someone should care about your product.

And every single one of those ads sends the customer to... the same homepage. The same generic PDP. The same collection page that knows nothing about why they clicked.

The ad said: "Finally, a solution for X." The landing page says: "Welcome. Here are our products."

That's a broken handoff. And it's costing you conversions every single day.

The fix is simple in theory. You match the landing page experience to the ad. If the ad spoke to a specific desire, the landing page leads with that desire. If it featured a specific product angle, the page opens with that angle.

We have a software-assisted solution that does this at scale, even when you're running hundreds of ad variants. Every landing page experience matches the ad that drove the click. The handoff becomes seamless. The story continues.

It works. And it's more fun to build than you'd think.

3. Give Google Shopping Visitors More Than Just the One Product

Someone clicks your Google Shopping ad. They land on the PDP.

Cool. But here's the problem: that might not be the exact product they actually want. It might be close. It might be in the right category. But now they have to go back to Google Shopping, scroll around, hope your other ad shows up, and maybe click through again.

You just paid for that click. And you're sending them back to Google to try again.

The better play? Show them the product they clicked on AND a curated set of similar products right there on the page. Let them find the right fit without leaving your site.

Wayfair does this. Yeti does this. 1-800-Flowers does this.

They turn a single ad click into a discovery experience. The customer doesn't need to go back to Google because everything they need is already in front of them.

This is solvable. If you're sending Shopping traffic to a vanilla PDP and nothing else, you're leaving money on the table every hour of every day.

4. Change the Site Based on What the User Has Shown You

Every time someone visits a product page, applies a filter, or makes a purchase, they're telling you something.

After 2 or 3 visits, you can start to see patterns. They keep looking at a specific brand. They always filter by a specific category. They haven't bought yet, but they're clearly interested in a particular corner of your catalog.

That's affinity data. And most brands just... ignore it.

The smarter play: use it.

Sort category and search results based on the brands and categories they've shown interest in. Add a "Popular in [Category]" section to the homepage that dynamically reflects their affinities, not the site's overall bestsellers. When they come back after a few days, show them recently viewed products right on the homepage instead of making them go find what they were looking at.

This is especially powerful if you have a large catalog. A customer who comes back to your 10,000-SKU site shouldn't have to start from scratch every time. The site should already know where to start the conversation.

You've already collected the data. Now use it.

5. Just Ask Them What They're Looking For

People want help finding the right product. Most brands just never ask.

Buff City Soap is the best example I've seen of this done right. They have a quiz that isn't a sidebar widget or a popup. It's genuinely part of the experience. It asks a few questions. It learns what you like, what you're trying to solve for, what your preferences are.

And then every interaction going forward is shaped by those answers. The product recommendations. The emails. The next time you visit the site.

It's like walking into a specialty shop and having the owner ask you four questions before pointing you toward anything. That's a white glove experience. And it makes the sale feel obvious instead of effortful.

If you have a catalog with any real complexity, multiple use cases, multiple customer types, multiple problems you solve, a quiz is one of the highest-ROI things you can build. Not because it reduces friction. Because it builds connection.

6. Let Email and SMS Finish What Your Site Started

This one gets left out of almost every personalization conversation because it lives "off-site." That's exactly why it's a blind spot.

Everything on-site, the quiz answers, the affinity signals, the browsing behavior, the recently viewed products, has a second life in your email and SMS flows. But only if you've connected the systems.

Think about what's possible when you do:

A customer takes the quiz and doesn't buy. Your email flow picks up exactly where they left off, leading with the product the quiz recommended, not a generic welcome sequence. A customer browses a specific category three times but never adds to cart. Your SMS fires at the right moment with the right product, not a 15% off coupon that trains them to wait for discounts.

The on-site experience plants the seed. Email and SMS water it.

Most brands treat these as separate programs. The ones winning right now treat them as one connected system. The quiz answer that lives in Klaviyo should shape every email that customer ever receives. The affinity data from their browsing should inform which SMS campaigns they're enrolled in.

On-site personalization without the off-site follow-through is a leaky bucket. You're doing the hard work of learning about the customer and then forgetting everything the moment they leave.

Don't let the data go to waste.

7. Use AI, But Not an Open Chatbot

Yes, AI. Obviously.

But not the way most brands are implementing it.

An open chatbot sitting in the corner of your site is not personalization. It's a support ticket waiting to happen. Nobody wants to type a question into a box and wait for a response.

What actually works is a guided tool. Something that proactively helps customers figure out if a product is right for them. Not a general-purpose assistant. A purpose-built experience that walks someone through the right questions and surfaces the right product for their specific situation.

Done right, this increases conversion rate. Not because it removes friction. Because it builds confidence. The customer leaves the interaction knowing they made the right choice, which means they're far less likely to return it and far more likely to come back.

The tool does the work of a great salesperson. And unlike a great salesperson, it scales.

The Through-Line

Notice what none of these have in common with the old playbook.

None of them are about removing friction. None of them are about faster load times or cleaner checkouts or bigger "Add to Cart" buttons.

They're all about creating demand. About making a shopper feel seen, understood, and guided before they've decided to buy.

That's the era we're in now.

The brands that figure this out first are going to pull away. The brands still optimizing button placement are going to wonder why their revenue per visitor keeps flatlining.

Personalization isn't a feature. It's the strategy.

The Friction Playbook Is Broken. Here's What Replaced It.

2026-03-19T00:00:00Z

For a solid decade, e-commerce had one job: remove friction.

Shave 0.2 seconds off the load time. Move the Add to Cart button above the fold. Kill one more step in the checkout flow. The whole industry operated on one assumption. Demand already exists. Your only job is to not screw it up.

And honestly? It worked. For a while.

The Demand Capture Era Was Real

Here's what was actually happening during the golden years of e-commerce growth.

Distribution was the unsolved problem. For the entire history of commerce (we're talking the Silk Road, the British East India Trading Company, Walmart opening its 4,000th store) the game was always getting the product in front of the person who wanted it. Then Amazon came along and solved distribution for almost everyone. Suddenly any brand could reach any customer, anywhere, within a day.

That created a stretch of growth that felt like genius but was mostly timing. You put a dollar into Google Shopping, you got three dollars back. You listed your product on Amazon and sales showed up because people were already searching for what you sold. You optimized your site by removing friction because the demand was already walking through the door.

The entire optimization industry, my industry, was built on this reality. Reduce friction. Capture more of the demand that already exists. Revenue goes up. Everyone's happy.

But here's what nobody wanted to say out loud: we weren't growing demand. We were just getting better at catching it.

Then the Music Stopped

At some point (and you probably felt it before you could name it) the demand capture playbook stopped delivering.

You're still running the same A/B tests. Still shaving milliseconds. Still moving buttons around. But RPV is flat. Conversion rate won't budge. The friction is already gone, and the results aren't improving.

That's because the problem changed underneath you.

The more friction you remove, the less each removal moves the needle.

Distribution is solved. Amazon, Shopify, social commerce. A customer can buy almost anything from almost anywhere in almost no time. The brands that grew by riding that wave have now hit the ceiling. Every competitor has the same access to the same customer through the same channels.

When everyone has distribution, distribution stops being the advantage.

And when the demand that's out there has already been captured (by you, by your competitors, by whoever got there first) removing more friction doesn't create more customers. It just polishes an empty room.

We're in the Demand Generation Phase Now

The brands that are actually growing right now? Not reshuffling sales between channels. Actually growing. They've figured out something the rest of the market is still catching up to.

They're not optimizing for capture anymore. They're building demand that didn't exist before.

And it looks nothing like the old playbook.

I was talking to Dan Pierce, CEO at Sovereign Naturals, on a recent episode of Checkin to Checkout. He put it simply: "You can buy transactions, but you can't buy transactions where there isn't demand."

His team owns 70% of their category on Amazon. They've captured about as much demand as exists. So what does his team spend their time on now? Not bid adjustments. Not checkout optimization.

They brought a videographer on staff to document a product development process that's taken 14 years and counting. They're interviewing the doctors who recommend their products, not to pitch the product, but to tell the doctor's story and make their world bigger. They found out their team had been washing medical-grade bottles as an extra purification step for years, and instead of burying that detail, they turned it into content. Because that kind of specificity is what builds trust at scale.

Their content split is flipping. Less "here's 20% off." More "here's what it actually takes to make this product and why we're obsessive about it." More employee stories. More behind-the-scenes. More celebrating the people who use the product and letting them tell their own story.

Dan told me he obsesses over search volume now. That wasn't even on his radar five years ago. But if more people aren't looking for you, you're not growing. You're just fighting over the same pool.

That's what the demand generation era looks like in practice. It's not one channel. It's not one tactic. It's a team that wakes up every day asking "are more people looking for us than yesterday?" instead of "how do we convert more of the people who already showed up?"

The measurement changes too. You stop living and dying by ROAS on every campaign and start tracking brand momentum. Search volume trends. Social engagement that isn't tied to a promo code. You start investing in things that won't convert this week but build the audience that converts next quarter. And that's a hard conversation to have with a CFO who wants to see a return in 30 days. But it's the only conversation that matters right now.

Not because it's trendy. Because the math requires it.

Why This Matters for Your Site

Here's where this hits home if you're running e-commerce.

If you're still operating like demand capture is the game... if your optimization roadmap is still about reducing friction, removing steps, making the checkout 0.3 seconds faster... you're solving the wrong problem.

Your customers aren't leaving because they can't find the Add to Cart button. They're leaving because they don't have a reason to stay.

The site experience now has to do something it never had to do during the capture era. It has to make someone feel something. Build confidence. Tell a story that makes the price feel irrelevant.

That's a completely different optimization problem. It's not "how do we reduce friction?" It's "how do we increase the value someone perceives before they hit the buy button?"

And the teams that get it? The ones who stop kicking the vending machine and start asking why nobody wants what's inside it? Those are the ones who are going to win the next five years.

The Question Worth Asking

So before you run the next A/B test, before you rearrange the PDP one more time, before you try to squeeze another 0.1% out of checkout... ask yourself:

Am I solving for demand capture or demand generation?

Because if the demand isn't there, no amount of friction reduction is going to save you.

Directors of E-Commerce Don't Get Fired for Missing Revenue. They Get Fired for Not Managing Up.

2026-03-26T12:00:00Z

On this week's Checkin to Checkout, Lauren Goodwin joins Justin to dig into the question every e-commerce brand needs to answer before any tactic makes sense: why you, why now? If you're building the case internally for where to invest next, it's worth a listen before you keep reading.

Someone I know hit 94% of their revenue target last year. The business was genuinely growing. The team was executing. The roadmap was smart.

They were gone in 18 months.

Not because the strategy was wrong. Because the CMO didn't feel informed. Because the CFO got surprised by a Q3 dip they should have seen coming. Because the CEO walked into a board meeting and didn't have the story.

Nobody fired them for bad performance. They fired them for making leadership feel out of control.

This happens constantly in e-commerce. And the people it happens to are almost never the ones doing a bad job. They're the ones so buried in doing the actual work that they forget to tell anyone about it.

Most directors learn this lesson after it costs them something. Here's how to learn it before it does.

Your actual job is making your boss look smart in front of their boss

Most Directors of E-Commerce think managing up means "don't surprise people." That's the floor. Not the ceiling.

The real job is giving leadership a story they can retell.

Your CMO needs to walk into a board meeting and sound like they understand what's happening in e-commerce. Your CFO needs a number they can point to when someone asks about ROI. Your CEO needs to feel like someone on the team has a plan and it's working.

If you're not providing those things proactively, someone else is filling the narrative vacuum. And their version of the story usually ends with your budget getting cut or your role getting "restructured."

Directors of E-Commerce already have two full-time jobs. The first is the one you were hired for: grow revenue. The second is the one you actually do every day: getting people who don't report to you to do work they don't care about, on your timeline, so you don't miss your number.

There's a third job nobody mentions. Keeping the people above you feeling confident about you. Not impressed. Not wowed. Just confident. That's the bar. And most directors never clear it because they're too busy doing the first two.

Six things that keep Directors of E-Commerce employed

1. The Friday 3-bullet Slack

A director I know has been in her role for four years. Two VPs have come and gone above her. She's outlasted both of them.

Her secret is boring. Every Friday afternoon she sends her CMO a Slack message. Three bullets. What's working. What's not. What's next.

Nobody asked her to do this. That's the point.

It takes her ten minutes. It gives her CMO everything they need to walk into any meeting and sound informed. It means her boss is never caught off guard. And when a VP above her gets replaced, the new person inherits a trail of receipts showing she's been delivering consistently for years.

Three bullets. Every Friday. That's it.

2. The "What If" model

Every director has been in this meeting. You're reviewing the quarterly scorecard. Revenue is up 30%. Traffic is up 40%. Looks like a massive win.

Then someone points at the Revenue Per Visitor column. It's down.

Right on schedule: "So did all those experiments you ran actually do anything? Or did we just burn cash?"

If you don't have an answer ready for that question, you lose. Not because the question is fair. Because the perception of failure is now in the room and you didn't control it.

Build a simple model that shows what RPV would have been without your experiments. Here's the logic: when you scale traffic aggressively, especially cold paid social traffic, RPV doesn't stay flat. It dips. That's not a failure. It's math. New visitors don't spend like returning customers.

Your "What If" model shows the gap. This year we ran 12 winning experiments. Combined, they lifted baseline RPV by 23%. Without those tests, with this much cold traffic, we wouldn't just be "down." We'd be underwater. And we would have missed the revenue target by 20%.

That's not a defense. That's a proof point. Have it ready before anyone asks.

3. Pre-announce the dip

If Q3 is going to miss, tell them in July. Not August. Not in the post-mortem.

A CFO who gets surprised is a CFO who fires you. A CFO you warned in advance is a CFO who defends you.

The framing matters: "Here's what's happening. Here's why. Here's the plan to recover." That's three sentences. You can deliver them in a hallway conversation or a one-paragraph email. The medium doesn't matter. The timing does.

Every director I've seen get fired for a bad quarter actually saw the bad quarter coming. They just hoped it would fix itself. It almost never does.

4. Own the channels you don't control

A director I was working with missed his Q1 goals last year. We did the whole forensic review. Paid channels were fine. RPV was decent. Site was performing.

Then I asked about email. He said, "Oh, I don't own email. That's Marketing."

Turns out they were sending one email a month. One. And he thought he could still hit his number by optimizing what he did own.

His CFO didn't care that email wasn't technically his department. 30% of revenue didn't show up. That was his problem.

If a channel affects your number, you need a relationship with whoever runs it. A weekly check-in. A shared dashboard. Something. You don't need to own the team. You need to own the conversation.

The best directors I know are internal lobbyists. They're in every room where their revenue is being discussed, whether they were invited or not.

5. Keep a win inventory

Start a running document. A shared Google Doc, a Slack canvas, a Notion page. Every time something meaningful happens, log it. The date, the result, the metric, and one sentence on why it matters.

It doesn't need to be polished. It just needs to exist.

When performance review season arrives, you don't scramble to reconstruct the narrative from memory. You pull up the inventory. When a new VP walks in and asks "what has this team actually accomplished?" you don't freeze. You send them a link.

The directors who survive long enough to get promoted are the ones who can prove what they've done without having to think about it. The win inventory is how you build that proof in real time instead of trying to recreate it under pressure.

6. The first 90 days with a new boss

Most of the advice in this post assumes a stable relationship with leadership. But the highest-risk window for any director is when a new boss arrives.

A new CMO, a new CEO, a new VP above you. They don't know you. They don't know what you've built. They're forming an opinion of you in the first three weeks whether you realize it or not.

The director I mentioned earlier who outlasted two VPs didn't just survive because of her Friday Slack. She survived because every time a new VP started, she did the same thing: she interviewed them.

Not formally. Just a conversation in the first week. "How do you like to receive updates? What level of detail do you want? What's the one thing that would make you feel like e-commerce is in good hands?"

Then she over-communicated for the first 60 days. More updates than she'd normally send. More context. More proactive flags. Not because the new VP needed it, but because she was establishing her rhythm before they established their expectations.

By the time they settled in, she was already the person they trusted. The person who made them feel informed. The person who never let them get blindsided.

If you wait for the new boss to tell you what they want, you've already lost the first 90 days. You set the tone. You define the relationship. You teach them how to work with you before they figure it out on their own.

Three ways directors accidentally get themselves fired

The dashboard dump.

You send a 40-slide deck full of metrics nobody asked for. You think thoroughness earns trust. It doesn't. Leadership's eyes glaze over. They walk away feeling less confident, not more. I once watched a CEO start writing an email while the director was mid-slide. Nobody told the director because nobody wanted to hurt their feelings. He was gone six months later.

The lone wolf.

You execute brilliantly in your silo. Numbers are up. Tests are winning. The site is humming. And nobody upstairs knows. You think the work speaks for itself. It doesn't. Work doesn't have a publicist. If leadership doesn't hear about it from you, they assume it's not happening. This is exactly what the win inventory solves. You can't publicize what you haven't logged.

The surprise miss.

You knew in June that Q3 was going to be rough. But you hoped the back-to-school push would cover it. It didn't. Now you're explaining a miss in October that leadership is hearing about for the first time. A miss they can plan for is a business reality. A miss they didn't see coming is a firing offense.

The directors who survive aren't always the most brilliant. They're the ones who never let leadership feel blindsided.

The person who got fired at 94% didn't lack skill. They lacked a communication system.

The director who's been in her seat for four years while two VPs came and went above her doesn't have a secret. She has a Friday Slack message, a win inventory, and the discipline to interview every new boss in week one.

Managing up isn't politics. It's the infrastructure that lets you keep doing the work you're actually good at.

Send the updates. Pre-announce the bad news. Own the channels that affect your number. Log your wins. Set the tone with every new boss before they set it for you.

Do those six things and you get to keep doing the part of the job you actually like.

Justin Aronstein is the Chief Product Officer at Mobile1st, where his team helps e-commerce brands increase revenue per visitor through experimentation and customer insight. If you're a Director of E-Commerce who wants a partner that helps you build the story, not just run the experiments, reach out at mobile1st.com.

Your Ads Made a Promise. Your Landing Page Broke It.

2026-04-02T00:00:00Z

This week on Checkin to Checkout: Josh Johnston, Senior Director of Online Experience at Trail Appliances, talks about surviving replatforms, weaponizing customer data in executive forums, and building an AI chat widget that drove a 4x conversion lift.

You've done the work.

Your Meta ads account is a thing of beauty. You've got 47 creatives running across six audiences. You've tested hooks about protein content, hooks about taste, hooks about convenience for busy parents. You've got UGC, polished studio shots, and that one weird iPhone video your founder shot at 6am that somehow outperforms everything.

Every week, your media team is in there cooking. New angles. New formats. New offers. New objection-busting copy. They're testing carousel vs. single image. Short-form vs. long-form. Pain point leads vs. aspiration leads.

They're doing it because they know the game has changed. Creative is the new targeting. And they're right.

But here's the part nobody talks about.

All of that beautiful, segmented, persona-matched creative is driving traffic to the exact same landing page.

The same static page. With the same headline. The same hero image. The same value props in the same order. No matter which ad someone clicked. No matter what promise convinced them to tap through.

Your ads are having 47 different conversations with 47 different people. Your landing page is having one conversation with everyone.

And it's the wrong one for most of them.

The Protein Oatmeal Problem

Let me give you a real example because I've seen this exact scenario play out with enough clients that it haunts me.

A brand sells oatmeal. Good oatmeal. They've got a strong Meta ads operation. One of their top-performing creatives leads with the protein angle. "28g of protein in every bowl." The ad shows a gym bag, a shaker bottle, and a bowl of oatmeal. The copy talks about how hard it is to hit your protein goals and how this is the easiest breakfast swap you'll ever make.

Someone who cares about protein sees that ad. They click.

They land on a page that says: "The Best Tasting Oatmeal You've Ever Had."

The hero image is a family at a breakfast table, smiling. The first three value props are about flavor variety, kid-friendly options, and a money-back taste guarantee.

Protein? Mentioned once. Below the fold. In the fourth bullet point of a section most people never scroll to.

That customer came looking for a protein solution. They landed on a taste pitch.

Some of them will convert anyway. The product is good. The price is fair. But a meaningful percentage of those protein-motivated visitors just bounced. Not because the product was wrong. Because the page didn't finish the conversation the ad started.

Now multiply that by every creative you're running.

The ad about convenience for busy parents lands on the same page as the ad about clean ingredients. The ad about the subscription discount lands on the same page as the ad about the founder's story. Every single ad, with its carefully crafted hook and its specific persona target, lands on a page that ignores all of that context.

It's like running a personalized email campaign where every link goes to your homepage.

If this sounds like your ad account right now, we built something that fixes it automatically. Keep reading and I'll show you the data.

Your Site Doesn't Know What Ad They Clicked

This is the core of the problem, and it's so obvious that most teams just… accept it.

Your media team spends hours every week building creative that speaks to specific people about specific things. They vary by persona, hook type, format, offer, persuasion angle, and objection. They know exactly who they're talking to in each ad.

Your site knows none of this.

When a visitor lands on your page, your site has zero context about where they came from or what convinced them to click. It treats the protein-obsessed CrossFit guy exactly the same as the taste-focused mom looking for something her picky toddler will eat.

And look, your site is good. You've optimized it. You've thought about the big problems customers are trying to solve. You've got solid product photography, clear pricing, and a checkout flow that doesn't make people want to throw their phone.

But it's static. It's a fixed experience that tries to be everything to everyone, which means it's not quite right for anyone.

The ad did the hard work of finding the right person and saying the right thing. Then the landing page undoes half of it by changing the subject.

What Happens When the Ad Matches the Page

We started noticing this pattern a few years ago. Client after client, the same disconnect. Incredible ad programs feeding traffic into landing experiences that had no relationship to the creative that drove the click.

So we tested it.

We took clients who were already running strong Meta creative and matched the landing page experience to the ad. When someone clicked an ad about protein, they landed on a page that led with protein. When someone clicked an ad about taste, the page led with taste. Same product. Same price. Same checkout. Just a different conversation based on what the visitor already told us they cared about.

The results were hard to argue with.

Revenue per visitor went up 20% or more for those matched experiences.

Not conversion rate. Revenue per visitor. The number that actually tells the truth about whether your site is working.

And it makes intuitive sense if you think about it. You already did the hardest part. You already found the right person and got them to click. All you have to do is not change the subject when they arrive.

That's a low bar. But almost nobody clears it.

We're giving newsletter subscribers a free month of Throughline to see this for themselves. Subscribe to The Director's Cut and we'll set you up.

Why Nobody Does This (Even Though Everyone Should)

If matching ads to landing pages is so obviously valuable, why aren't more teams doing it?

Because it's a nightmare to execute manually.

Think about what it would take. You'd need to catalog every active creative. Classify each one by persona, hook, funnel stage, and offer type. Then build or modify landing page variations for each classification. Then set up the routing logic so the right visitor sees the right page based on the right ad.

For a brand running 50+ creatives, that's an enormous amount of work. And it changes every week as your media team launches new creative and pauses underperformers.

Most e-commerce teams are already stretched thin. The Director of E-Commerce is already juggling platform migrations, email calendars, inventory headaches, and a CMO who wants to "protect the brand story." Nobody has time to build and maintain a manual ad-to-page matching system.

So teams just accept the mismatch. They optimize the landing page for the broadest possible audience and hope for the best.

It's not a strategy. It's resignation.

We Built Throughline Because We Got Tired of Watching This Happen

After seeing this same problem with client after client after client after client, we built a solution.

It's called Throughline.

Here's how it works.

We download all of your Meta ads. All of them. The images, the videos, the CTAs, the copy, the headlines. Everything that's live in your account.

Then we ingest all of that creative and classify each one across multiple dimensions. We calculate the persona it's targeting. We identify the funnel stage. We determine the hook type. We classify the persuasion angle, the objection being addressed, the offer structure, and several other dimensions that matter for landing page relevance.

Once every creative is classified, we can dynamically update the landing page to match the ad for anyone clicking through that specific creative.

Protein ad? Protein-led landing page. Taste ad? Taste-led landing page. Convenience-for-busy-parents ad? You get the idea.

Same product. Same checkout. Same brand. Just a landing experience that actually continues the conversation your ad started instead of starting a new one from scratch.

The visitor doesn't know anything changed. They just feel like the page "gets" them. Because it does. It knows exactly what brought them here, and it speaks to that.

The Math Is Simple

Your media team is already doing the hard work. They're testing hooks, iterating on creative, finding the angles that resonate with different audiences. That work is generating clicks.

But every click that lands on a mismatched page is a leak. You paid for that click. You earned that click. And then your static landing page fumbled the handoff.

When you match the landing experience to the ad creative, you're not adding a new channel or increasing spend. You're making every dollar you already spend work harder.

Our clients have seen 20%+ gains in revenue per visitor when they follow this process. That's not a rounding error. On a $10M revenue base, that's $2M you're leaving on the table because your landing page doesn't know what your ads are saying.

Want to See It for Yourself?

If you subscribe to our newsletter, The Director's Cut, we're giving away a free month of Throughline so you can see what matched landing experiences do for your numbers.

No pitch meeting. No demo you have to sit through. Just connect your Meta account, let us classify your creative, and watch what happens when your ads and your landing pages finally start speaking the same language.

Click here if you want in →

Because your media team is already doing the work. Your landing page just isn't listening.

Justin Aronstein is Chief Product Officer at Mobile1st, where he helps e-commerce brands grow revenue per visitor through experimentation and ad-to-page personalization. He also co-hosts Checkin to Checkout, a podcast for e-commerce operators who'd rather build than babysit dashboards.

Stop Personalizing. Start Being Relevant.

2026-04-08T00:00:00Z

Every year, someone invents a new word for the same idea.

Personalization. CRO. Site optimization. Offer strategy. Dynamic content. AI-powered experiences.

We love the new words. We build decks around them. We buy tools that promise them. We hire agencies that specialize in them.

But strip the jargon away and you're left with one question that has always been the question:

Is what this person is seeing right now relevant to them?

That's it. That's the whole game.

Revenue per visitor goes up when relevance goes up. Conversion rate goes up when relevance goes up. Click-through rate goes up when relevance goes up. Every metric you report on, every KPI your CEO cares about, every number that keeps you employed... they all move in the same direction as relevance.

Not "personalization." Not "optimization." Relevance.

The Difference Between Relevant and Personal

Here's where most teams get it wrong.

They hear "personalization" and they think: capture the customer's name. Show it on the homepage. Maybe reference their city. Maybe a "Welcome back, Sarah!" banner.

That's personal. It is not relevant.

Knowing someone's name doesn't help them buy the right product. It doesn't reduce the mental math happening in their head. It doesn't answer the question they showed up with.

But knowing their height, their favorite driver brand, and their handicap? Now you can suggest the exact club they should buy next. That's relevant. That changes the buying decision.

The difference isn't subtle. One makes the customer feel recognized. The other makes the customer feel understood. And understood is what opens wallets.

We Could Always Read the Signals. We Just Couldn't Manage the Complexity.

Here's the thing nobody says out loud: the data has been there for years.

We knew which ad someone clicked. We knew what they searched on Google. We could see their browsing behavior in real time. UTM parameters, referral sources, session data... none of this is new.

The problem was never reading the signals. The problem was doing something about them at scale.

Because if you wanted to create a unique landing page for every ad variation, a different PDP experience for Google Shopping traffic, and a personalized navigation bar based on browsing behavior... you were looking at hundreds of experiences to build, test, and maintain.

For the last decade, the best we could do was test for the whole audience. Look at session replays, form a hypothesis, build one variation, and run an A/B test for the entire population of visitors. If Version B beat Version A, ship it.

That worked. It still works.

But it treats every visitor the same. The person who clicked a Meta ad for running shoes and the person who typed your URL into their browser and went straight to hiking boots... they saw the same homepage. The same hero image. The same "Shop Now" button.

We knew it was a compromise. We just couldn't operationally manage the alternative.

That's what changed. Not the data. The management layer.

We can now build and manage thousands of experiences without the operational nightmare that used to come with it. The tooling caught up. What used to require a dedicated team to maintain 50 landing pages can now run seamlessly across hundreds or thousands of variations, adapting in real time without someone manually wiring up every rule.

Where the Relevance Actually Lives

Think about what happens when someone clicks one of your Meta ads.

Meta's algorithm is terrifyingly good at figuring out who wants to see what. It analyzed thousands of signals you'll never have access to, determined that this specific person would respond to this specific creative, and served it to them. And they clicked.

That click is a massive relevance signal. It tells you exactly what message resonated.

So what do most brands do with that signal?

They send the person to a generic landing page. Or worse, the homepage.

All that precision targeting. All that algorithmic intelligence. And you reward the click with a page that says "Welcome to Our Store. Shop Our Collection."

The relevance dies the second they land.

Now imagine the alternative. They clicked an ad featuring a specific product, with a specific message, highlighting a specific benefit. They land on a page that continues that exact conversation. Same product front and center. Same language. Same benefit emphasized.

Relevance maintained. Conversion rate jumps. Not because you redesigned the page. Because you stopped breaking the promise the ad made.

Google Shopping is the Same Story

Someone searches for a specific product on Google. They see your listing. They click through. They land on the PDP.

Great. They're on the right product page. But they got there through a comparison shopping mindset. They were looking at your product next to four others, comparing prices and ratings.

What if, instead of just showing them the standard PDP, you surfaced similar products right away? Gave them the comparison experience they were already in the middle of? Showed them that you have depth in this category, not just one option?

You're not changing the product. You're matching the context they arrived with. That's relevance.

The Bar Across the Top

Here's one that sounds simple but almost nobody does well.

A visitor has been browsing your site. They've looked at three categories. They've hovered on two brands. They've added something to their cart and kept shopping.

You know what they're interested in. You have the data. It's sitting right there in their session.

So why does the navigation bar look the same as it did when they first showed up?

Surface the categories and brands they've been exploring. Put them right across the top. Make it effortless to go deeper into the things they've already told you they care about.

You're not guessing. You're listening. And then you're making it easier for them to find more of what they already want.

The Old Way Still Works. The New Way Compounds.

Here's what I don't want you to hear: "A/B testing is dead" or "CRO doesn't matter anymore."

That's not what this is.

Traditional experimentation, testing a hypothesis against the full audience, is still one of the highest-ROI activities in e-commerce. If you're not doing it, start.

But relevance-driven changes layer on top of that. They compound.

You run a test that lifts RPV by 8% across all visitors. Great. Now you layer in ad-to-page relevance and lift RPV another 12% for paid traffic. You add session-based product surfacing and conversion rate ticks up for return visitors.

These aren't competing strategies. They're the same strategy operating at different levels of resolution.

The old way increased relevance for everyone by finding the best single version. The new way increases relevance by reading the same cues we've always had access to, but managing the response at a scale that used to be impossible.

The signals didn't change. Our ability to act on thousands of them simultaneously did.

Why This Matters Right Now

We're past the era where you could grow by just buying more traffic. CAC is up. Margins are down. The CFO wants growth without new spend.

The only way to grow revenue without growing traffic is to get more out of the visitors you already have. And the fastest way to do that is to stop showing everyone the same thing.

Every visitor who lands on your site is carrying context with them. The ad they clicked. The search they ran. The pages they've browsed. The products they've lingered on.

That context is a gift. It's telling you exactly what's relevant to them.

The brands that win in 2026 and beyond are the ones that actually listen.

Not with a name tag. Not with a popup. Not with a "Hey Sarah, welcome back!"

With a page that feels like it was built for the exact reason they showed up.

That's content relevance. And it's the only thing that matters.

Latest from Checkin to Checkout

Ep. 13: Everything Is Downstream from Your ERP - Wayne Stratbucker

Wayne Stratbucker has seen both sides of the omnichannel equation. He's helped brands sell on Amazon, Walmart, TikTok Shop, and Instagram. In this episode, he breaks down why the brands winning right now are the ones who got their backend right first.

Listen now →

How to Use Experts to Build Community

2026-04-15T00:00:00Z

I spent a long weekend checking Instagram every hour.

Not for friends. Not for memes. For a race.

Specifically, the Go One More Backyard Ultra Marathon, held just outside Austin at Nick Bare's ranch. I found out about it through a few fitness influencers I follow. Didn't plan to get hooked. And yet, every hour on the hour, I was opening Instagram to watch the BPN live stream, scanning faces to see who lined up for the next lap and how they looked.

After searching for Bare Performance Nutrition ten times in a single weekend, I just followed them.

What Is a Backyard Ultra?

If you don't know the format, here's the short version: competitors run a 4.2-mile loop every hour, on the hour. You keep going until you're the last one standing.

On paper, it sounds manageable. 4.2 miles? Anyone can walk that in an hour.

In reality, you're competing against athletes who've lasted over 72 hours and covered 280+ miles, still running. Every lap you complete, you have to start the next one. Miss the start gun and you're out. There's no finish line. The goal is just to go one more.

Who Is Bare Performance Nutrition?

If you've been to a grocery store lately and hit the supplement aisle, you've probably seen BPN. Protein powders, pre-workout, electrolytes, creatine. They sell DTC but have made their way into HEB, Vitamin Shoppe, Scheels, and Central Market.

I've tried the protein. It's solid. But honestly? I'm not sure it's meaningfully different from the 15 other options sitting next to it on the shelf.

Nick Bare founded the company in 2012 out of a college apartment in Indiana. $20k loan, no experience, a lot of grit. He went on to serve as an Army infantry officer, completed Ranger School, ran 100-mile ultras, competed in bodybuilding. His book, Go One More (a New York Times bestseller), captures the philosophy behind all of it: growth is a choice, decisions compound, and the only way forward is to go one more. One more rep, one more mile, one more hard decision.

The race is the book. The book is the brand. The brand is the man.

What I Kept Thinking About

The runners in that race aren't just customers. They're expert users. Not necessarily of BPN specifically, but of the entire category of products BPN competes in. These are people who have spent years dialing in their nutrition. They know what works at hour 30. They know the difference between a product that holds up and one that doesn't. If you want to know how your electrolytes perform under real duress, ask someone who's been running loops since Friday.

These athletes are the power users. The ones who have developed real opinions.

And every one of them has their own audience. Their own following. Their own community that watches them train, race, and document the whole thing. That audience followed the race. That audience watched the BPN live stream. That audience searched for BPN, probably the same way I did.

BPN didn't sponsor a runner. They brought the runners together.

The Community Problem Everyone Shares

Every e-commerce team worth anything is trying to build community. The reasons are straightforward: community reduces customer acquisition costs, increases LTV, generates organic content, produces reviews, and creates a base of loyal buyers who aren't just waiting for a promo code.

We all want followers who actually care. We want customers who talk about us without being asked. We want the kind of brand attachment that doesn't disappear when a competitor runs a better ad.

Most of us are trying to build community for the masses: something everyone can participate in. A hashtag campaign. A loyalty program. A Facebook group. A Discord.

And most of us find it hard, because it is hard. It takes real work to build a community from scratch around something people can join but don't have a strong reason to.

BPN flipped this.

They didn't try to build a community for everyone. They built an event for the experts and let everyone else watch.

The Expert-Audience Model

Think about what actually happened here.

The competitors in that race are the most credible validators BPN could ever have. They're not influencers paid to hold up a shaker bottle. They're ultra-endurance athletes actively testing their nutrition at the extreme end of what a human body can do.

Meanwhile, every fan of those athletes (and there are thousands of them) is watching the BPN live stream. Not because they were targeted with a Facebook ad. Because someone they already follow is competing, and the race is compelling enough to check back every single hour.

At one point I opened the stream and over 15,000 people were watching. On a brand Instagram account. For a supplement company. Who wouldn't want that kind of viewership?

The experts are the show. The masses are the audience. And the brand owns the stage.

Sponsorship vs. Ownership

Sponsorship deals have always done something similar. Brands attach themselves to elite athletes or events because the association carries credibility.

But there's a key difference: sponsorship rents the stage. Running the event owns it.

When you sponsor a race, people watch the race. Your logo is in the background. The narrative belongs to the event organizers. You get impressions but not identity.

When you host the race, on your founder's ranch, with your brand name in the title, with your team running the live stream, the event and the brand become inseparable. People don't search for "that ultra marathon." They search for Bare Performance Nutrition.

Even if every competitor isn't sponsored by you, they are now a part of your community.

What This Looks Like for Other Brands

You don't need an ultra marathon. You don't need a ranch outside Austin. You just need to ask: who are the most expert users of products in my category, and how do I bring them together in a way that's worth watching?

Sell gear for fly fishing? Don't just send influencers to Montana. Bring the top 20 guides in the country together for a weekend on a legendary stretch of river. Stream it. Own the content.

Sell outdoor apparel? Gather elite mountaineers to summit something together. You don't need to sponsor it. You need to create it.

Sell radiators? Host a car show with the top cars in your area.

You don't need your brand to be the only brand represented. Other brands will be there. That's fine. The experts will use what they use. What matters is that you are the one who brought the experts together. People will appreciate that.

People will become part of your community because you gave them something worth gathering around.

So What Do You Do With This?

Most community-building efforts fail because they ask too much of the audience at the start. You're asking people who barely know you to opt in, engage, generate content, and care.

The expert-audience model asks the opposite. Start with the people who are already obsessed: the power users, the extreme adopters, the practitioners who have made your category a major part of their identity. Give them a stage. Make their expertise visible. Create something worth watching.

The masses don't need to participate to become part of your community. They just need to watch something compelling enough that they come back.

Every hour, on the hour.

Related Episode

Why Your $200K Personalization Platform Isn't Moving RPV

2026-04-29T12:00:00Z

A walk through the 7 types of on-site personalization, and the one most brands skip.

A Director of E-Commerce I know spent $200K on a personalization platform last year.

They did everything right on paper. Built 14 audience segments. Trained a recommendation model. Launched a quiz. Hired a CRO agency to run experiments on top of it. Their vendor's QBR slides were beautiful.

Six months in, RPV was flat.

Not down. Just flat. Which, after $200K, a platform implementation, and a year of internal political capital, is its own kind of disaster.

When we dug in, the answer was almost embarrassing. 60% of their paid traffic was hitting the same generic homepage. The ad with the most budget promised "shoes built for marathon training." The landing page said "Welcome. Shop the collection." The visitor arrived, the relevance gap was enormous, and they bounced before any of the $200K worth of personalization had a chance to fire.

They had personalized everything except the moment that mattered most.

This is the failure pattern I see at almost every $5M to $100M e-commerce brand I work with. Personalization investment is back-loaded toward the sophisticated end of the stack (predictive models, audience segmentation, identity resolution) while the simplest, highest-leverage layer gets ignored because it is not sexy.

The simple layer is making sure the page matches the ad.

You cannot out-personalize a broken first impression. And yet most personalization strategies are built as if the first impression is already handled. It almost never is.

Start with the entry point. Everything else is a layer on top.

The type of personalization you should start with is contextual. Closing the relevance gap at the entry point, between what the ad promised and what the landing page delivers. It is the cheapest, highest-leverage, most ignored layer in the personalization stack.

Every other type works harder when you fix this one first. Most brands skip it because their vendor is not selling it.

The rest of this article walks through the seven types, the failure modes of each, and a framework for deciding which secondary layers to add based on your stack.

The relevance gap is biggest at the entry point

The relevance gap is the distance between what the visitor was promised and what the visitor sees.

Every personalization type tries to close this gap somewhere in the funnel. Predictive recommendations close it on the PDP. Behavioral triggers close it after the visitor has shown intent. Audience segmentation closes it for known customers across sessions.

But the gap is largest, and most expensive, at the entry point. The moment a visitor arrives from an ad, an email, or a search result is when the promise is freshest in their mind and the cost of mismatch is highest. Bounce rates on landing pages are not a UX problem. They are a relevance gap problem. (You can see what your own gap looks like with your real Meta ads here.)

Most personalization investment goes toward closing gaps that occur after the entry point. Those gaps are real. But it is like waterproofing the basement while the roof is open. You are solving the wrong layer first.

Seven types, listed in the order brands tend to buy them

Which is roughly the inverse of the order they should be bought in.

1. Inferred / Predictive Personalization

What it is: Model-driven personalization. Propensity-to-buy scores, churn risk, category affinity, next-best-product recommendations, intent prediction. This is what most "AI personalization" vendors are actually selling. The implementation pattern is fundamentally different from the others on this list. You are not writing rules. You are consuming model outputs and trusting them to render the right thing.

A newer entrant worth calling out: MadeWithIntent approaches predictive differently than most. Instead of training models on what visitors have done in the past to predict what they will buy, they score live intent in the moment, using behavioral signals to identify visitors who are at risk of bouncing or ready to convert. The activation is in-session rather than catalog-based. Same category, different angle, and arguably more honest about what predictive personalization can actually deliver.

What it pulls: Primarily CVR when it works. Predictive personalization is supposed to identify the right visitor for the right intervention at the right time. When the model is good, it pulls more visitors through the funnel. The AOV story (better recommendations = bigger baskets) is real but secondary, and frankly, oversold by most vendors.

Data prerequisites: Significant traffic and transaction volume. Models need training data. Sub-10K sessions per day means most of your model's predictions are noise dressed up as signal.

Tech lift: Low to medium. Most vendors are turn-key. Drop a script, configure a few placements, go.

Maintenance burden: High and underestimated. Models drift. Inventory changes. Catalog updates break recommendations. Without constant QA you will, eventually, recommend out-of-stock products, items the customer just bought, or worse.

Failure mode: Black-box embarrassment. The model surfaces a fertility tracker to a male shopper, or a winter coat to someone in Phoenix in July, or the exact product the customer returned last week. The technology is great. The lack of governance around it is what kills brands.

Tools to start with: MadeWithIntent for in-session intent scoring. Nosto and Rebuy for Shopify-native product recommendations. Bloomreach and Dynamic Yield for mid-market. Algolia Recommend if you are search-heavy. Fast Simon for catalog-driven brands.

The honest take: I am skeptical of a lot of this category. Strip away the vendor decks and a meaningful chunk of "predictive personalization" reduces to identifying customers who would have bought anyway and giving them a discount. That is not personalization. That is margin erosion with a model attached. The vendors that are actually doing something interesting (MadeWithIntent's in-session intent scoring being one example) are the ones identifying real interventions for real moments, not just optimizing discount distribution. Buy carefully. Ask hard questions about what the model is actually predicting and what you are actually doing with the prediction. If the answer is "we discount the people most likely to convert," you are paying a vendor to lower your margin.

2. Behavioral Personalization

What it is: Personalization based on what the visitor does on-site. Pages viewed, time on PDP, cart additions, scroll depth, search queries. Strong for in-session relevance. The visitor signals intent through their actions, and the experience adapts.

What it pulls: CVR (rescuing high-intent visitors who are about to bounce) and AOV (post-add-to-cart upsells based on the cart contents).

Data prerequisites: Traffic. Real traffic. If you are running 5,000 sessions a day and trying to behaviorally personalize three different segments across four page templates, you do not have enough volume to learn from any of it.

Tech lift: Medium. Most personalization platforms include behavioral triggers as a standard feature. The rules are easy to configure. The hard part is deciding which behaviors actually predict intent.

Maintenance burden: High. This is true for most of the rule-based categories on this list, but it shows up first here. Every behavioral rule is something a human decided. Every variation of an experience is something a human built. Behavioral rules accumulate. The PM who set them up leaves. The new PM does not know what is running. Without aggressive pruning, you end up with 200 active rules, no idea which ones are working, and a personalization stack that is mostly tax.

Failure mode: Behavioral personalization fires after the visitor has already engaged. By definition, it cannot help you with bounce. If your relevance gap is at the entry point, behavioral does nothing for you.

Tools to start with: Rebuy and Nosto for Shopify. Optimizely Web Personalization, Dynamic Yield, and Bloomreach for mid-market and up. Most A/B testing platforms (VWO, Convert) include behavioral targeting.

The honest take: Useful, well-understood, and worth building. But it is a middle-of-funnel tool. If your top of funnel is broken, behavioral personalization is solving for a smaller and smaller pool of survivors.

3. Zero-Party Data Personalization

What it is: Personalization based on what the customer explicitly tells you. Quizzes, preference centers, on-site surveys, profile completion. The visitor declares their preferences, goals, or constraints, and the experience adapts in response.

What it pulls: CVR (better product matching reduces decision paralysis) and RPV (visitors who complete a quiz convert at multiples of those who do not).

Data prerequisites: A reason for the customer to engage. The quiz has to feel like it is for them, not for you.

Tech lift: Low to medium. Plenty of Shopify-native quiz apps. The harder part is what happens after the quiz: does the result actually change the experience, or does it just route to a generic collection?

Maintenance burden: Medium. Lower than behavioral because there are usually fewer rules, but every quiz path is still a manually designed experience. Someone decided what each answer combination routes to. The catalog changes. Products go out of stock. Quiz logic gets stale faster than people notice.

Failure mode: Collecting the data and never using it. Quizzes that ask 12 questions and route everyone to the same three bestsellers are worse than no quiz at all. They train the customer to ignore the brand's promises about understanding them.

Tools to start with: Octane AI, Rebuy quiz builder, and Shopify's native quiz features for smaller brands. Jebbit for more sophisticated, branded quiz experiences. Klaviyo for tying quiz responses into email and SMS flows.

The honest take: Underused, especially for high-consideration categories. If you sell mattresses, supplements, skincare, or anything where the customer has a real internal monologue about what is right for them, zero-party data is your highest-leverage move after contextual.

4. Audience / Segment Personalization (including Lifecycle Stage)

What it is: Personalizing based on who the visitor is rather than what they are doing. Logged-in customer tier (VIP vs. new). CRM or CDP segment. Lifetime value bucket. B2B vs. B2C. Where they sit in the lifecycle: first-time, considering, active, lapsed, win-back. The defining feature is that it persists across sessions and channels. Behavioral personalization resets every visit. Audience does not.

What it pulls: AOV (surfacing higher-tier products to known high-value customers) and RPV (matching the experience to the customer's actual relationship with the brand).

Data prerequisites: Identity. Either the visitor is logged in, or you have stitched their anonymous session to a known profile (more on that in type 6). Without identity, audience personalization does not exist.

Tech lift: High. You need a CDP or a strong CRM with on-site activation, segment definitions that are actually useful, and the engineering to render different experiences per segment.

Maintenance burden: High and human-driven. Every segment is something a person defined. Every experience that segment sees is something a person built. The most common failure is rule sprawl: a brand builds 47 segments, designs custom experiences for 12 of them, and then nobody maintains the other 35. Quarterly segment reviews are non-negotiable. Almost nobody actually does them.

Failure mode: Building segments you cannot activate. Most brands have beautiful segmentation in their CDP and a generic homepage on their site. The two systems never talk. The data exists, the experience does not change.

Tools to start with: Klaviyo for Shopify-native audience activation (the on-site personalization features are underused). Bloomreach Engagement, Dynamic Yield, Optimizely, and Salesforce Personalization for mid-market and enterprise. Segment or Rudderstack as the CDP layer underneath.

The honest take: Powerful when the data infrastructure supports it. Most brands are at least 18 months away from doing this well. That is not a reason not to build toward it. It is a reason not to lead with it.

5. Geo / Temporal Personalization

What it is: Personalization based on location, weather, time of day, and local inventory. Free shipping thresholds by region. Weather-triggered product surfacing (rain jackets in Seattle in October). "In stock at your nearest store" callouts. Local pickup options. Sometimes folded into contextual personalization, but it deserves its own callout because the implementation is distinct. You are consuming external feeds, not reading the visitor's referrer.

What it pulls: CVR (relevance to the visitor's physical context). Sometimes AOV (free shipping threshold optimization).

Data prerequisites: Geo-IP (mostly free), weather APIs (cheap), and inventory feeds tied to location (the harder one).

Tech lift: Low to medium. Geo and weather are easy. Local inventory is harder because it requires real integration with your OMS.

Maintenance burden: Low to medium. The rules themselves run autonomously once configured. The variations that fire (the regional offers, the weather-triggered hero copy) are still manually built. Every region you add is a creative production task.

Failure mode: Privacy creep. "We see you are in Austin" feels useful. "We see you are in 78704" feels invasive. The line is not where you think it is.

Tools to start with: Most personalization platforms include geo-targeting natively. Tomorrow.io and OpenWeather for weather APIs. For local inventory, your OMS or commerce platform usually has the data; the work is exposing it on-site.

The honest take: Concrete, narrow in scope, and underused. Worth building when you have the right use case (regional shipping, climate-driven catalog, brick-and-mortar footprint). Not worth building if you are a single-warehouse DTC brand selling one SKU type.

6. Identity-Resolved Personalization

What it is: Personalization unlocked by tying anonymous on-site visitors back to known email addresses or customer profiles. Vendors like Retention.com and Customers.ai use various forms of identity resolution to recognize 30 to 40% of "anonymous" sessions as identifiable visitors. Suddenly, audience and lifecycle personalization apply to people who never logged in. This is not an experience layer. It is a data layer that makes other personalization types work harder.

What it pulls: RPV through compounding effects. Identity resolution does not directly change the experience. It expands the population on which other personalization types can fire.

Data prerequisites: A vendor relationship and a clean integration with your ESP, CDP, or personalization platform. The data is only useful if it flows into the systems that use it.

Tech lift: Low to medium for the resolution itself. Higher for downstream activation, because you need a personalization layer that can consume the resolved identity and render against it.

Maintenance burden: Low at the data layer. The vendor handles match rates, decay, and refresh. The work is on your side: making sure resolved identities are actually being used. Which loops back to the audience segmentation problem above. Identity resolution just gives you more inputs for a system you may or may not be running well.

Failure mode: Two failure modes here, both serious. The first is buying identity resolution and never activating it. The second is regulatory and brand risk. Some customers find this useful. Some find it deeply uncomfortable. The reputable vendors have consent layers and legitimate-interest frameworks. The disreputable ones do not. Choose carefully.

Tools to start with: Retention.com is the most prominent in the e-commerce space. Customers.ai, Black Crow AI, and Lebesgue offer adjacent capabilities. Klaviyo and Sendlane have begun building this in natively.

The honest take: Genuinely powerful when stacked on top of audience and lifecycle personalization. But it is a force multiplier, not a foundation. If your audience personalization is weak, resolving more identities just gives you a bigger group of people to under-personalize for.

7. Contextual Personalization

What it is: Personalization based on the link and context the visitor brought with them. UTMs, referrer, ad creative, campaign source, email click. The system reads the signal the visitor arrived with and adapts the landing experience to match. Hero, headline, social proof, product surfacing, offer. Every paid campaign gets a landing experience that mirrors the ad's promise instead of dumping every visitor onto the same generic homepage.

What it pulls: CVR (massive reduction in bounce when the page matches the ad) and RPV (every retained visitor becomes signal for every other personalization type downstream). Contextual is the only type on this list that compounds. Every visitor you do not lose at the entry point becomes data for behavioral, audience, predictive, and identity resolution. Skip this layer and the rest of your stack runs on a smaller and smaller pool of survivors.

Data prerequisites: Almost none. You already have UTMs. You already have referrer data. The data is sitting in every visitor's URL bar and HTTP headers.

Tech lift: Theoretically low. Practically, this is where most brands fall over. The reasons are organizational, not technical. The team that owns the ad and the team that owns the landing page rarely talk. Paid media is judged on CPA. CRO is judged on on-site conversion. The handoff between them, which is exactly where the relevance gap lives, is owned by nobody.

Maintenance burden: This is where contextual gets misunderstood. Done the wrong way (hand-building a landing page for every ad variant), maintenance is brutal and unscalable. A $20M brand running 50 ad variants cannot produce 50 landing pages. So they produce one, and send all 50 ads to it. Done the right way (dynamic, parameterized personalization driven off the ad signal), maintenance is low and scales with campaign volume rather than collapsing under it.

Failure mode: The failure mode is not running it at all. The B2B-native tools (Unbounce, Instapage, Mutiny) are excellent products built for someone else: their pricing, integrations, and product surface are not designed for Shopify e-commerce. Native solutions for closing the relevance gap across paid, email, and owned links have been essentially absent. The result is the most consistent revenue leak in mid-market e-commerce. Every Director knows it intuitively. Almost nobody has the tooling to fix it.

Tools to start with: Throughline. It is an app that turns every paid campaign, email, or link you control into a contextually relevant landing experience without requiring a new page build for each one. Throughline reads the email or ad context, the UTM, and the referrer, and dynamically composes changes to the page including hero, copy, social proof, and product set to match the promise the visitor was sold. Brands of any size can run it. The lift compounds with every campaign you launch.

If you want to see what your relevance gap looks like with your real Meta ads, you can run a live demo at throughline.mobile1st.com/demo.

The honest take: This is the foundation. Walk through the math: if 40% of your paid traffic bounces because the landing page does not match the ad, no amount of downstream personalization recovers that revenue. Behavioral needs the visitor to engage. Audience needs the visitor to be known. Predictive needs page views. Identity resolution needs time on page. They all fail when the visitor bounces in three seconds. Contextual personalization is not competing with the other six types. It is the floor they sit on. Every dollar you spend on personalization beneath a generic landing page is leverage you are leaving on the table.

How to decide which type to start with

Step 1, regardless of size, traffic, AOV, or stack: Run contextual personalization. Close the relevance gap at the entry point. The math is universal. Visitors who bounce do not become customers, and they do not become signal. There is no personalization stack that works around a broken first impression.

Step 2, layer based on your stack:

Under 10K sessions per day? Skip predictive and behavioral. The volume is not there. Add zero-party if your category is high-consideration (mattresses, skincare, supplements, B2B). Add geo if you have regional logistics differences.

Have a CDP and logged-in users? Audience and lifecycle are next. This is where compounding shows up in the data.

Over 50K sessions per day with strong transaction volume? Predictive becomes worth the cost. Below that, the model has nothing to learn from. And even above that, ask hard questions about whether the predictive layer is generating real lift or just identifying customers who would have bought anyway.

Have a meaningful email list? Identity resolution is the multiplier on whatever audience layer you already run.

AOV under $50? Lean heavily into contextual and zero-party. The economics of mid-market personalization platforms are punishing at low AOV.

AOV over $200 and high-consideration category? Zero-party is your highest-leverage non-contextual move. Buyers in those categories want to be understood before they buy.

The framework is simple: start with contextual. Layer based on what your stack and economics actually support.

Personalization is a system that compounds. Or it falls apart at the entry point.

Every visitor who does not bounce becomes a data point for behavioral. Every behavioral signal becomes input for predictive. Every predictive output gets validated against audience. Every audience gets enriched by identity resolution.

The whole system falls apart if the entry point is broken.

Most brands will keep investing in the top of the personalization stack. The brands that compound will fix the foundation first.

The relevance gap at the entry point is the largest, cheapest, most ignored revenue leak in mid-market e-commerce. Closing it is not the most sophisticated thing you can do. It is the most leveraged.

If you want to see what your relevance gap looks like with your real Meta ads, run a live demo at throughline.mobile1st.com/demo.

This piece is part of The Director's Cut, a newsletter for Directors of E-Commerce who want to compound their advantage instead of chasing tactics. If you want more like this, subscribe.

How to start running tests today with the help of AI

2026-05-06T00:00:00Z

(Why I'm all in on personalization)

Let's get the obvious out of the way.

AI is powerful. Genuinely, absurdly powerful. It can process more information, hold more variables, and make more connections than any human alive. That is not hype. That is just math.

But here is the thing nobody is saying: AI is only as useful as the context you give it.

Right now, you still have to be the one asking the right questions. You still have to bring the data. You still have to know enough to know what you're looking for.

That will probably change in the next few years. But today, the e-commerce manager who wins is the one who learns how to use AI as a thinking partner, not just a search engine.

This guide is for you if:

You know you should be running experiments, but you don't know where to start
You've thought about hiring an optimization agency, but the budget isn't there yet
You suspect your product page is leaking revenue, but you can't prove it

You might actually not need an agency. You might just need Claude.

Here's exactly how I'd run an experimentation program if I were starting from scratch today.

First, a quick framework you need to understand

Not all evidence is created equal.

At Mobile1st, we use something called the Hierarchy of Insights. You can read the full breakdown here. The short version:

Heuristics (expert opinion, best practices, research): useful, but the weakest signal
User Behavior Analysis (scroll maps, click data, session recordings): stronger
User Research (real conversations with real customers): stronger still
Experiment Results (actual A/B test data from your site): the strongest signal of all

Most brands stop at heuristics. They read a blog post, make a change, and call it optimization.

The goal of this workflow is to stack all four layers, so when you launch an experiment, it's backed by evidence, not just instinct.

Let's build the stack.

Step 1: Start with a heuristic audit

Before you touch your data, audit your pages against a research-backed framework.

I'd start with your product detail page (PDP). Almost every single visitor who buys something will land on a PDP. Fix problems there first, then work up to category pages and your homepage.

I built a 22-tactic PDP audit sourced from Baymard Institute, Nielsen Norman Group, and Shopify platform data. You can use mine or build your own. The tactics are organized by impact tier (Very High, High, Medium, Low) so you know where to focus first.

Here's how to run it with Claude:

Open Claude. Paste in the audit framework.
Give it your product page URL (or screenshots).
Ask it to score the page across all 22 tactics.

You'll get a scorecard, a letter grade, and a ranked list of fixes. In about two minutes.

Will it catch everything? No. Heuristics never do. But it will surface the obvious problems fast. Missing in-scale imagery. No sticky add-to-cart on mobile. Shipping cost buried in the footer. The stuff that's clearly broken.

This is your starting list. Not your final list.

Step 2: Layer in real user behavior

Heuristics tell you what should matter. Behavior data tells you what's actually happening on your specific site, with your specific customers.

If you haven't set up Microsoft Clarity yet, do it today. It's free. It records sessions, generates heatmaps, and surfaces scroll depth and click data automatically.

Once you have it connected, hook up the Clarity MCP to Claude and start asking questions:

"Where do visitors stop scrolling on this product page?"
"What's the click rate on the Add to Cart button versus the image gallery?"
"Are visitors rage-clicking anywhere on this page?"
"What percentage of visitors see the reviews section?"

You're not looking for answers yet. You're looking for questions worth asking.

If 70% of your visitors never scroll past the hero image, your above-fold value prop is the problem. Not the upsell widget below the fold that you've been obsessing over. Behavior data keeps you honest.

Step 3: Pull your GA4 event data

If you've got custom events set up in GA4, this is where things get interesting.

Hook up the GA4 MCP and ask Claude to help you find patterns in your event data:

"What's the drop-off rate between product page views and add-to-cart events?"
"Which traffic sources have the highest add-to-cart rate but the lowest purchase rate?"
"Is there a difference in behavior between mobile and desktop visitors?"

Even with basic GA4 tracking, you'll start to see where the funnel breaks. That's the data you need.

If you don't have events set up yet, that's okay. Start simple. Add an add_to_cart event and a purchase event. Even those two will tell you a lot.

Step 4: Bring in your customer voice

This is the step most brands skip. It's also the step that separates good experimentation programs from great ones.

Two sources of customer voice:

Post-purchase reviews. You have them. You're probably not mining them. Take your last 100 reviews, paste them into Claude, and ask: "What are the most common concerns customers had before purchasing? What almost stopped them from buying?"

The answers will surprise you. Customers will tell you exactly what objection your PDP failed to address.

User research. I know. Nobody has time. But even five 20-minute conversations with real customers per month will give you more signal than 10,000 rows of behavioral data. Ask them to walk you through how they shop your category. Record it. Bring the transcript to Claude.

AI cannot replace this. The nuance in a customer saying "I wasn't sure if it would fit my lifestyle" is not something a heatmap will ever show you. This is the part where you still have to be human.

Step 5: Build your experiment roadmap

Now you have four layers of evidence. Time to put them together.

Take everything (your audit results, your behavior insights, your GA4 data, your customer voice) and bring it all into a single Claude conversation. Then ask:

"Based on everything I've shared, build me a prioritized experimentation roadmap. Rank experiments by estimated revenue impact. Weight experiments higher if they're supported by multiple evidence sources, especially user research and behavior data."

You'll get a ranked list of experiments, each with a hypothesis and a rationale.

This is not a gut-feel list. This is evidence-backed prioritization. Every experiment on the list exists because multiple data sources pointed to the same problem.

That's the difference between optimization theater and a real program.

Step 6: Write the code

"But Justin, I don't know how to code."

It's 2026. Anyone can write code.

Take your top hypothesis from Step 5. Bring it back to Claude with your page URL and say:

"Here's my hypothesis: [paste it]. Please write the JavaScript and CSS to create this variation that I can use in my A/B testing software."

Most A/B testing platforms (Optimizely, VWO, Convert, AB Tasty, and the Google Optimize alternatives) all work essentially the same way. You paste custom code into a visual editor, set your traffic split, and launch.

Claude will write the code. You paste it in. You don't need to understand every line. You need to understand the hypothesis.

Step 7: Launch and let it run

This part requires patience.

Do not look at the results for two weeks. I mean it. Statistical significance takes time, and peeking early is how you make bad decisions and convince yourself bad tests are good.

Set a minimum runtime of two weeks. Ideally three. Let it breathe.

Step 8: Interpret the results

When the test is done, take a screenshot of your results dashboard and bring it to Claude.

Ask: "Here are my results. Was this statistically significant? What does this tell me about visitor behavior? Should I call a winner, extend the test, or kill it?"

Claude will walk you through the math and give you a clear recommendation.

Step 9: Update your roadmap

Here's the part everyone misses.

Experiment results are not just wins or losses. They are information. Every test teaches you something about your customer.

Go back to your experiment roadmap. Share the results with Claude and say:

"Here are the results from this experiment. Here are my previous hypotheses. Please re-rank my priority list based on what we learned."

Your roadmap should be a living document. Every experiment makes the next one smarter.

This is the compounding effect that agencies charge you a lot of money for. You're building institutional knowledge about your customer with every test you run.

Step 10: Grab a coffee

Seriously.

You just ran an evidence-backed experiment, built on heuristic analysis, behavior data, GA4 events, and customer voice. You interpreted the results. You updated your roadmap.

That is an experimentation program.

Not a one-time optimization. Not a "we changed the button color and called it a day." A real, repeatable system.

You built it with a free AI tool and some patience.

One more thing: what happens after your experiments start winning

Once you're running experiments and the data starts compounding, you'll hit a new question.

Your experiments are telling you what works on your site. But what about the traffic coming in from Meta ads? Those visitors don't arrive cold. They clicked something specific. They saw a specific message. And then they land on a generic page that has no idea what they just saw.

That gap between the ad and the page is where a lot of revenue disappears quietly.

It's called the relevance gap. And it's the next frontier after experimentation.

Throughline is how we solve it. It's a Shopify app that reads the Meta ad a visitor clicked and dynamically updates the landing page to match. Same product. Different headline, value prop, and imagery based on what brought them there.

If your experiments are working and your paid traffic still isn't converting the way it should, that's usually why.

Check it out at throughline.mobile1st.com.

Your Customers Don't Care About Your Features (Even When They Tell You They Do)

2026-05-13T00:00:00Z

Something funny happens when you ask customers why they bought a product.

They don't say "the 400-thread-count Egyptian cotton" or "the 20-hour battery life." They say "I finally sleep through the night" or "I stopped worrying about finding an outlet."

They describe their life after the purchase. Not the product itself.

And yet most product pages are still organized around the product. Specs. Materials. Technical callouts. Everything that made sense in the development backlog or the supplier catalog, presented as if the customer cares about the same things the product team does.

They don't. And that gap is costing you conversions.

Features are what you built. Outcomes are why someone buys.

Here's the distinction:

A feature is a 400-thread-count Egyptian cotton sheet set with temperature-regulating technology.

An outcome is "wake up feeling rested, even if your partner runs hot."

A feature is a 13-inch laptop with 20-hour battery life and a 1.4kg chassis.

An outcome is "work a full day at the coffee shop without hunting for an outlet or stressing about your bag weight."

One of those sentences explains the product. The other one sells it.

Most product pages are still leading with the what.

Here's the test we run with our clients

We work across industries (SaaS, D2C, B2B) and we've run enough experiments at this point that it's practically a standing heuristic:

Take your feature list. Replace it with outcomes. Add a "See full specs" button below.

That's it. That's the hypothesis.

The logic is simple: a customer who isn't yet convinced your product will change their life has no reason to care about specs. But once they believe it will? They'll click that button. They want to see the proof. The specs become evidence for a decision they've already emotionally started to make.

We recommend testing it rather than just taking our word for it. You want the data to back the change. But in the experiments we've run, the results tend to go one direction. Conversion goes up. Revenue per visitor follows.

How do you know what outcomes to put on the page?

This is where most brands overcomplicate it.

Ask your customers.

Your post-purchase survey is sitting there right now with untapped product insight. Two questions are all you need:

How do you use this product?
Why do you love it?

Read through the responses. Patterns will emerge fast. You'll see the same phrases showing up: the real jobs this product is doing in people's lives. The way they describe your product to a friend. The outcome they actually bought.

Now put those words on the product page. Not your marketing team's version of the outcome. Not the brand story from the homepage. The customer's own language, reflected back at them.

That's the relevance signal that converts.

Most product pages are written for the wrong audience

They're written for buyers who already understand the category, who know what the specs mean and can mentally translate a feature into a benefit.

But most of your traffic isn't that buyer. They're earlier in the decision. They need to see themselves using the product before they're ready to evaluate it.

Features are the how. Outcomes are the so what.

Lead with the so what.

What to do this week

Pick your top five products by traffic. Pull the last 90 days of post-purchase survey responses. Look for outcome language.

Then go to those product pages and rewrite the above-the-fold feature section as outcomes. Move specs below. Add the button.

Run it as an A/B test if you want the data. Or ship it and watch the trend line.

Either way, you're starting to sell the way your customers actually buy.

Want help figuring out the right outcomes to feature or how to test these outcomes? Reach out to Justin!

Your Product Recommendation Slider Is Lying to You

2026-05-20T00:00:00Z

Pull up a stool, grab a drink, and let's talk about the biggest, most expensive lie currently sitting on your e-commerce site.

I'm talking about that product recommendation slider. You know the one. It lives on your Product Detail Pages (PDPs), your homepage, and your cart page, quietly cycling through individual products. Some SaaS sales rep swore up and down that its "proprietary, AI-powered, lookalike-behavioral algorithm" would automatically print money for your brand.

So you dropped a chunk of your budget on it, slapped it on the page, and called it a day.

The Multi-Million Dollar Lie in Your Tech Stack

Here is the unsaid truth your tech vendor doesn't want you to whisper in a board meeting: Product recommendation sliders are almost entirely useless.

They are the wrong tool for the job.

Think about how these algorithms actually work. They track past shopping behavior, look at what other people who viewed this item also clicked on, and try to guess the exact single product this specific human wants to see next.

It sounds smart on paper. But in reality? It's like trying to find a needle in a digital haystack. The algorithm might get warm, it might get close, but 9 times out of 10, it's recommending the wrong product. And when a user sees a slider full of single items they don't want, they don't keep clicking. They tune out. Banner blindness sets in. You lose them.

Stop Handing Them Needles. Show Them the Haystack.

So, what do you do instead? Stop trying to hand them the needle. Show them where the haystack is.

Instead of recommending individual products, you should be recommending groups of products: categories, cohorts, and pathways.

Let's look at an example. Say you run a golf gear brand and a shopper is looking at a premium Callaway driver PDP. The standard, lazy play is to show a slider with 10 other individual Callaway clubs. But they are already looking at a driver. Showing them a slightly different driver just triggers choice paralysis.

Instead, give them helpful, thematic pathways. Give them links to broader groups: Callaway Irons, Beginner Drivers, Customizable Drivers, or Left-Handed Woods.

Each of these choices doesn't force them into a dead-end product page. It takes them to a curated set of products where they can easily find the exact item they're looking for. It drastically widens your net. If they don't like the exact driver on the page, you aren't gambling your whole conversion on a single lookalike item. You're giving them a logical next step to stay on the site.

Context Is Everything: The Best Buy Checkout Rule

Of course, the strategy changes based on where they are in the funnel. It's all about contextual relevance.

On the PDP: You want to offer alternative category paths to ensure the customer can find their right fit or build out their setup.
In the Cart: You aren't trying to upsell them on an entirely different category of core product. You want quick add-on categories. Think of it like the candy bars, lip balm, and phone chargers lined up in the Best Buy checkout lane. You don't show them another TV in the cart. You show them a pathway to Cables & Power Protection.

The Proof: A 4.5% RPV Bump for a $100M Brand

We don't just sit around bars theorizing about this stuff at Mobile1st. We test it. We call our flavor of conversion rate optimization Digital Product Growth because we don't care about vanity metrics. We care about closing the gap between the click and the buy to scale RPV.

We recently ran a version of this exact category-routing concept for a sports retailer doing over $100M online. We completely bypassed the traditional single-item recommendation blocks. Instead, we built a new browsing mechanism in the form of dynamic "bubbles" right in the header.

These bubbles shifted and changed in real-time based on the user's live browsing and shopping behavior. The more they browsed, the more personalized the set of bubbles became. Crucially, these bubbles never took a shopper to a specific, single product. They routed users to targeted brand and category pages that the data suggested this specific shopper would care about most.

We ran the experiment for over a month. The result? A 4.5% increase in RPV.

But the headline number doesn't tell the most interesting part of the story. The secondary metrics are where it gets really interesting.

Only about 20% of first-time visitors interacted with the new bubble bar. That's expected. A first-time visitor doesn't have enough behavioral signal yet for the personalization to feel earned, and they're often still figuring out what the brand even is.

But for repeat visitors, interaction rates jumped to over 40%. The bar got smarter the more someone browsed, and that compounding relevance is exactly what kept them clicking.

Here's why that matters: returning visitors (people who'd been to the site more than twice) made up over 75% of total revenue. So the RPV lift on the segment that actually drives the business was significantly higher than the blended 4.5% number suggests. We were moving the right people, not just inflating an average across cold traffic.

And because this mechanism was built into the global header layout, it impacted 100% of the site's traffic. That is massive revenue left on the table if you're just letting a black-box widget guess individual SKU preferences. There is still a ton of optimization work to be done on fine-tuning our algorithm and content sets, but this exact architecture can, and should, be applied to every single part of an e-commerce site.

Real Personalization Requires Real Experimentation

For us at Mobile1st, 2026 is entirely focused on one thesis: Personalization means nothing unless it is verified through experimentation. Real site growth doesn't come from buying bloated personalization platforms. It comes from making every touchpoint contextually relevant.

That's the exact reason we built our newest product, Throughline. It's designed to automatically message-match your landing pages to the precise ad creative that drove the click in the first place, keeping the emotional hook alive.

We're currently recruiting e-commerce leaders (especially those doing between $3M and $100M who want to see what actual, data-backed experimentation can do) to participate in a study. We want to ensure we're building Throughline to solve your exact, real-world growth bottlenecks.

If you want to stop burning ad spend, ditch the lazy sliders, and actually move your RPV, drop your info below. No long forms, no sales gauntlet. We'll reach out personally, and fast.

The Post-Click Audit Nobody Runs: A 7-Part Scorecard for Ad-to-Page Fit

2026-06-03T00:00:00Z

You have a dashboard for everything that happens before the click. CTR, CPM, frequency, ROAS by audience, creative-level breakdowns. You can tell me which thumbnail outperformed which hook down to the cent.

Then the visitor clicks, lands on the page, and your instrumentation goes quiet. You know the conversion rate. You don't know why the page converts or doesn't. And more importantly, you can't see the single most common reason it doesn't: the page breaks the promise the ad just made.

It's the cheapest revenue you'll ever recover, because you've already paid for the click. And almost nobody audits it, because there's no number for it. So let me give you one.

Your landing page is the second half of a sentence

Here's the reframe that makes this easy to think about. The ad isn't a separate asset from the landing page. The ad starts a sentence in the visitor's head, and the landing page has to finish it.

If the ad says "50% off your first box, this week only," the visitor arrives already holding three expectations: there's a discount, it's 50%, and there's a deadline. If the page they land on leads with "Premium ingredients, sustainably sourced," it's a non-sequitur. The visitor doesn't consciously think "this violates message match." They just feel a half-second of friction, lose the thread, and bounce. You paid for that click and got nothing, not because the product is wrong, but because the page answered a question nobody asked.

The fix isn't "make a better page." It's "make a page that delivers what this specific ad set up." And to do that systematically, you need to break the ad's promise into parts and check the page against each one.

The seven components of a promise

Across a few thousand ad-to-page pairs, the promise an ad makes almost always decomposes into the same seven components. Not every ad uses all seven (a brand-awareness ad might set up two, a flash-sale ad might set up six), but these are the universal building blocks. I've ordered them by how much they matter when they're missing.

1. Primary promise (the heaviest). The single benefit or outcome the ad leads with. Not a benefit: the one. "Sleep through the night." "Cut your closing time in half." If the ad's lead promise isn't the first thing the page reinforces, nothing else matters. This is the component that most often breaks, and it's the most expensive when it does.

2. The offer. The concrete incentive: the discount, the free trial, the bonus, the free shipping. If the ad dangles "$50 off" and the visitor has to hunt for it on the page (or it's not there at all), you've created a bait-and-switch feeling even when the deal is real. Offers need to be above the fold, not in the footer.

3. Specific claims. The proof points the ad named explicitly: numbers, features, comparisons, a testimonial. "Rated 4.8 by 12,000 customers." "Works in 30 seconds." If the ad made a specific, checkable claim, the page has to back it up. A specific claim that evaporates on the page reads as a lie, even if it was true.

4. Offer constraints. The urgency or eligibility attached to the offer: the deadline, the scarcity, the "new customers only." This is what converts intent into action. An offer with no constraint on the page loses the reason to act now.

5. Audience signal. Who the ad said this is for. If the creative targeted "for runners over 40" or "built for Shopify stores," the page should make that visitor feel seen in the first screen. A generic page after a tightly-targeted ad feels like you walked into the wrong store.

6. CTA continuity. Does the action the page asks for match the action the ad implied? If the ad's energy was "shop the sale," the button should say something like that and lead to a product or cart, not "Learn more" routing to a blog post. Mismatched CTAs quietly lose you the most motivated visitors.

7. Tone and identity. Does the page sound and look like the ad? A punchy, irreverent ad followed by a stiff corporate page is a tonal whiplash that erodes trust before the visitor reads a word.

If you want a rough weighting to argue about in your next page review: primary promise is worth roughly a quarter of the whole thing on its own, the offer another fifth, specific claims and constraints the next chunk, and audience/CTA/tone round out the rest. The exact points matter less than the order: fix the top of the list first.

Two rules that keep the scorecard honest

Most "landing page audits" produce a number that means nothing. Two disciplines are what separate a real diagnosis from a vanity score.

Rule 1: Only grade what the ad actually set up. A brand-awareness ad that never mentions a discount should not be penalized for the page lacking an offer, and it shouldn't get free credit for it either. You grade each ad against only the components it raised. This is why "our pages all score in the 70s" is a useless statement: a one-size rubric gives easy points for components the ad never invoked, and the scores cluster meaninglessly. Extract the promise the specific ad made, then grade against that. A direct-response ad gets a tough six-part rubric; a brand ad gets an honest two-part one.

Rule 2: "Delivered" means delivered well, not "mentioned somewhere." This is the one that changes behavior. The temptation is to check a box the moment the page references a component at all. Don't. Set the bar at something like 70%: the component has to be prominent, clear, and hard to miss, not buried in paragraph four. A CTA that exists but sits below three scrolls of copy is not delivered. When we moved our internal threshold from "is it present?" to "is it delivered well?", the number of fixable gaps we found on the same pages roughly tripled. The gaps were always there. A lenient bar just hid them.

A worked example

Take an ad: "New: protein bars with 20g protein, 1g sugar. 20% off your first order, code FIRST20. Built for people who train."

Score the page it points to:

Primary promise ("20g protein, 1g sugar"): is that the hero, or is the hero a generic "fuel your day"? Usually the gap.
Offer ("20% off"): visible above the fold with the code, or missing?
Specific claim ("20g / 1g"): repeated and reinforced, or vague?
Constraint ("first order"): clear, or absent?
Audience ("people who train"): does the imagery and copy speak to that, or is it generic wellness?
CTA: "Shop with FIRST20," or "Learn more"?
Tone: energetic and performance-driven, or soft and lifestyle?

Walk those seven in order and you'll almost always find the problem in the top three. Now you have a prioritized fix list instead of a vague "the page feels off."

The mistake to avoid: more changes ≠ a better page

When teams finally see the gaps, the instinct is to fix everything at once and pile on tweaks. Resist it. We learned this the expensive way: when you optimize for number of changes, you get a busy page full of low-value edits that pad the count and move nothing. The pages that actually win make fewer, larger changes that close the highest-priority gaps. One change that puts the primary promise and the offer above the fold beats six cosmetic tweaks. Close the top of the list, ship it, measure it.

And notice where the real work is. Spotting the gap is the easy half. Closing it well, rewriting a live page so the promise lands in the first screen without breaking your brand voice, your layout, or the rest of the funnel, is the hard half. Doing that once is a good afternoon. Doing it across every ad, every page, and every time the creative refreshes is the problem that actually eats your team.

And measure it for real, with live traffic, not your own judgment. A scorecard is a fantastic diagnostic for where to look and what to fix first. It is not proof that the fix worked. Internal scores, even good ones, are noisy and easy to fool yourself with. The only honest verdict is a conversion-rate test on real visitors. Use the scorecard to decide what to change; use live data to decide whether you were right.

Your Monday-morning version

Do it by hand first, on a handful of ads, so the pattern becomes real to you. Pull your five highest-spend ads. For each, open the landing page it points to, and on a sticky note write the seven components in order. Mark each one: delivered well / present but weak / missing. Be strict: "weak" and "missing" are where you're losing money.

You'll find your worst problems are almost never at the bottom of the list. They're at the top: the promise the ad led with, missing from the page the visitor paid you to see.

We built Throughline to run this scorecard automatically across every ad you're running, and then to close the gaps. It reads the promise in each ad, grades the landing page against it, and shows you exactly which components are missing or weak. Then it rebuilds the page to deliver them, you review and approve each change (it's not a black box), and approved versions ship live at the edge with a single snippet, no developer and no site redeploy. Every change runs as a live A/B test against the original, so the only versions that stay are the ones that actually lift conversions. If the sticky-note version above turned up gaps you want to close at scale, that's what it's for. Either way, run the audit on your top five ads this week. The gaps are already there.

We Scored 21 Top DTC Brands on Message Match. The Average Was 62/100.

2026-06-10T00:00:00Z

You can have a great ad and a great landing page and still lose the click in the half-second between them.

That half-second is where the visitor asks one question: "Is this what I was just promised?" When the answer is no (the discount code vanished, the claim isn't repeated, the page sells to everyone instead of the person the ad targeted), they bounce. Not because anything was bad. Because the message broke.

We call the discipline of keeping that promise message match, and we built a scoring rubric for it. Then we pointed it at 21 of the best-known DTC brands in America and published the results as the Message Match Index.

The average score was 62 out of 100. Nine of the twenty-one scored below 60. The promises these brands spend millions putting in front of people are, on average, only about 60% kept by the pages those people land on.

How we scored each brand

Everything in the index comes from public data, the same two things every shopper sees:

The ads. We pulled each brand's live Meta ads straight from the Meta Ad Library: up to five active ads per brand, exactly as they're running today.
The landing pages. We followed each ad's destination link and captured the page the way a first-time visitor lands on it.

Then our scoring engine does something deliberately strict: it reads the ad first, extracts every commitment the ad makes, and only then looks at the page. The page is scored blinded, with no idea whose page it is or how it's "supposed" to do. Each commitment becomes a scored component:

Component	The question it asks	Weight
Primary promise	Does the page hero deliver the specific benefit the ad led with?	25
Offer / incentive	Is the ad's offer visible with matching terms?	20
Specific claim	Are the numbers, features, or comparisons in the ad confirmed on the page?	15
Offer constraint	Is the ad's urgency, deadline, or redemption mechanic reinforced near the offer?	10
Audience signal	Does the page confirm the visitor is in the right place for them?	10
CTA continuity	Does the page's primary action match the action the ad invited?	10
Tone / brand voice	Does the page sound like the same brand as the ad?	10
Trust signal	Is the social proof or authority the ad leaned on confirmed on the page?	10

Not every ad commits to every component (an ad with no discount can't fail on "offer"), so each ad is scored only on the promises it actually makes. The overall score is the share of promised points the page delivers, 0 to 100. Anything earning 70% or less of its possible points on a component is flagged as a gap.

Two details make the rubric hard to game.

The creative outweighs the caption. A claim that only lives in body copy the visitor may never read counts for a fraction of a claim carried by the image or video itself. If your video promises "30-day results" out loud, the page is on the hook for it.

Video and image claims count. We transcribe spoken claims in video ads and read text printed on image creatives, so brands don't get credit for pages that match the caption but ignore what the creative actually says.

A brand's index score is the average across its scored ads. Each brand's page on the index shows the full component-by-component breakdown for its richest ad, including the rationale for every point lost.

What we found

The spread is enormous. Quince leads at 85. Fabletics sits at 37. The median is 64. Six brands, including household names like Rothy's, OLIPOP, and Fabletics, scored below 50. Great advertising and great message match are clearly different skills.

The thing brands do best: the button. CTA continuity scored 92% across every ad we graded. If the ad says "Shop now," the page almost always lets you shop. The mechanical part of the funnel is solved.

The thing brands do worst: everything they said before the button.

Offer constraints delivered at just 43%. The urgency mechanics ads lean on (deadlines, code-at-checkout instructions, minimum-spend tiers) usually evaporate on the page. Magic Spoon's ad offered "$5 off any order, or $10 off orders $63+ with code TREATSVIP"; the landing page showed a subscription discount and a free bowl set instead. A completely different deal. Vuori's ad told visitors to apply the code emailed to you at checkout; the page contained no code, no email mechanic, no mention of the offer at all.

Primary promises delivered at 58%. The single benefit the ad led with is barely half-kept. Mejuri ran an ad promising "guaranteed compliments wherever you go"; the landing page is a standard product page with zero aspirational copy. The emotional hook that won the click is abandoned at the door.

Audience signals broke for 16 of 21 brands. Ads that call out who the product is for ("parents of young children," "for serious home cooks") usually land on pages that sell to everyone.

Specific claims went missing constantly (16 of 21 brands). Jones Road's ad says free from sulfates, parabens, and phthalates; the page never mentions any of the three. Hiya ran ads for a greens powder with "55 whole-food ingredients" that pointed at... a probiotic product page where no greens powder exists.

The deeper pattern: ad-page pairing, not page quality. The most brutal scores came from ads pointed at the wrong page: a sleep-product ad landing on a probiotic PDP, a tiered-discount ad landing on a subscription pitch. The page is fine. The match is broken. This is an org-chart problem. The media team ships ads, the site team ships pages, and nobody owns the half-second between them.

(Methodology honesty: 28 brands went into the run; 7 were excluded for data-quality reasons, including destination links that were bot-blocked or Meta-owned, a landing page that hangs automated browsers, a page that serves visitors a full-screen cookie wall, and four brands whose public Ad Library lookups returned a different advertiser than the brand itself. Every exclusion is documented, and no brand was scored on data we couldn't stand behind.)

Why this matters for your CAC

Every gap the rubric finds is a place where money you already spent on the click gets thrown away by the page. If your ads promise something your page doesn't repeat, you're paying full price for half a message. Message match isn't a branding nicety. It's the cheapest conversion lever most teams haven't touched, because it sits between two teams and neither owns it.

See where you'd rank

The same engine that built the index runs as a free, self-serve analysis. Give it your Facebook page and you'll get your live ads, your landing pages, your score, and the exact components where your message breaks, in about two minutes.

Run your free message match analysis →

Methodology notes: ads sourced from the public Meta Ad Library on June 10, 2026; landing pages captured at scoring time; scoring performed by Throughline's blinded component rubric (no human adjustment). Scores are a point-in-time snapshot and will be refreshed as campaigns change. Brands shown are not customers and did not participate; all creative shown is their own public advertising.

You've only made 20% of the Meta Creative you need

2026-06-24T00:00:00Z

You did everything right.

You've got your pages dialed in: your core product pages, and maybe you even spun up special collection pages for that product drop. Tight, specific, highly targeted. While you were at it, you built creative for the drop. Some ads point at the collection page, some point at the product pages. Clean setup.

Then you built the campaign. A handful of ads, five image variations, a few different headlines and primary text options. You shipped it. Good work.

The catch: you made one set of creative.

Five image variations of the same idea. Different headlines saying the same thing. All hitting the same point, all aimed at the same person, all leaning on the angle your landing page already makes. You didn't make five pieces of creative. You made one piece of creative, five times.

You've built about 20% of the creative you actually need.

Stop iterating. Start brainstorming.

The move isn't making more versions of the same ad. Step back and brainstorm the whole space of who could buy this product, and why:

Who are all the personas who might buy it? The gift-giver and the self-treater aren't the same person. The first-timer and the loyalist aren't either.
What are all the angles on the offer? Free shipping, the bundle, the financing, the limited drop, the guarantee. Each one pulls a different buyer.
What are all the selling props? Every spec, every benefit, every reason to believe is its own potential hook.
What's the primary promise for each of those people? The transformation a runner wants isn't the one a new mom wants.
What proof moves each of them? Reviews, press, the founder story, the before-and-after.
What objection is each one stuck on? Price, fit, trust, timing.
What tone lands? The same product sells with humor to one segment and with authority to another.

Now multiply it out. Every combination of persona, offer, selling prop, promise, proof, and objection is its own creative set. That's not five ads. That's dozens, and each one is a real, distinct shot on goal.

That's how you scale creative: not more reps of one idea, but coverage across the whole space, so you can find out what actually works for each campaign instead of guessing which version of the same ad won.

Now launch the matrix and let Meta do its job.

This matters for more than volume.

When every ad says the same thing, Meta has nothing to optimize toward. It's just picking the prettiest version of one idea. Put a real matrix into market instead: the speed angle, the no-hassle angle, the year-round angle, each tuned to a different buyer. Now you've handed the algorithm something to chew on. It starts routing the speed angle to people who respond to speed, the price angle to deal-hunters, the proof-heavy ad to skeptics.

A brand we work with sells heated outdoor furniture. Same core product, very different buyers. One ad led with raw speed: warms up in 15 seconds. Another threw out the whole "fire pit" mental model: no propane, no waiting, snaps together in three steps. Different promise, different buyer, different reason to click. One product, many genuinely distinct angles.

Launch it, let the algorithm sort it, and within a couple of weeks you know something real: which angle wins which audience. That's the prize: signal, not volume.

Then you notice the problem. Every winning ad makes a different promise. The 15-second ad promised speed. The no-propane ad promised zero hassle. They're all pointing at the same landing page.

"But now my landing page doesn't match my ad."

Correct, and that's the point.

One landing page should never be able to match all the different ads you're running. If your single page somehow matches every ad you've got, that's the tell that you haven't thought through all the creative combinations available to you. You narrowed your creative to fit your page, instead of expanding your creative to find your buyer.

That mismatch is intentional. Someone clicks the "no propane, snaps together in three steps" ad, lands on a page whose hero is all about year-round comfort, and the one thing that won the click, the easy setup, is nowhere in sight. The message that earned the visit vanishes the second they hit the page.

That gap is where your conversions leak out.

Close the gap, dynamically.

Once all that creative is live, now is the moment to change your landing page based on which ad someone clicked.

By the time they land, you already know everything. You know the persona. You know which offer pulled them. You know the selling prop, the promise, the objection the ad was answering. You know exactly what the page needs to say to keep the conversation going instead of dropping it.

Back to the heated-furniture brand. The visitor who clicked the "15 seconds" ad lands on a page where the hero now reads "warms up in just 15 seconds," the proof gallery confirms that exact claim, and even the sticky buy bar carries it as they scroll. The visitor who clicked the "no propane, three steps" ad lands on the same page, reshaped to lead with the setup steps and "no propane, no waiting." One page, a different promise kept per click, each matching the ad that earned the visit.

This is what we do for our customers through Throughline, and it's a genuinely fun thing to watch work. The ad sets the expectation, the page delivers on it, and the visitor never feels the seam.

It's improving conversion rate by about 40%.

Make all the creative. Let the page keep up.

If you want to see what this looks like on your own store, email me at justin@mobile1st.com.

How to Tell If Your 100 Meta Ads Are Really Just One Ad

2026-07-01T00:00:00Z

Open your Meta account and count the active ads. Maybe you have a hundred. Maybe two hundred. That feels healthy. It looks like you're testing, iterating, and giving the algorithm plenty to work with.

Now try something uncomfortable. Line those ads up and ask a harder question. How many of them are actually different?

For most e-commerce brands we analyze, the answer is not good. You don't really have a hundred ads. You have one ad that got rebuilt a hundred times. Same customer, same product angle, same core claim, with a fresh hook or thumbnail on top. It's volume that looks like a portfolio but behaves like a single point of failure.

Variation isn't diversity

Your testing culture is working exactly as designed, and that turns out to be the problem. You find a winner and then you spin off twenty versions of it. A new headline here, a new opening frame there, a different background. Each one is a fair test. But they all orbit the same idea. Same persona, same product spec, same offer, same emotional register.

That is variation, and variation is not diversity. If you swap the hook but keep the persona, stage, format, offer, and claim the same, you have just repainted the same car.

So how many should you have?

Real diversity has more than one dimension. When we grade an account, we don't count ads. We check how many distinct values you cover on each axis, and how concentrated your spend is. These are the bands we recommend.

Axis	Options	Minimum to be "diverse"	Ideal
Persona	your defined set	all of them active	all, balanced spend
Funnel stage	TOF / MOF / BOF	all 3, none under ~15%	roughly balanced
Format	5 (UGC, testimonial, demo, lifestyle, polished)	3	4 to 5
Offer	5 (value prop, guarantee, discount, FOMO, aspiration)	3	4
Hook	6 (problem-aware, curiosity, social proof, story, urgency, aspiration)	3	4 to 5
Persuasion	6 (scarcity, authority, social proof, reciprocity, commitment, novelty)	3	4
Objection answered	5 (price, trust, complexity, fit, risk)	all 5 covered somewhere	all 5
Distinct product claims/specs	open	3 to 5 different specs	more
Concentration cap	none	no single cell over 40%	no cell over 25%

One reframe matters here. The floor is not a hundred ads. If you serve three personas across three funnel stages with two distinct concepts each, that already gives you around eighteen genuinely different ads. Once those are running, you make variations of whatever wins. Eighteen diverse concepts will beat a hundred clones.

Three lenses to see it

The first lens is the recipe. Is each ad right for the stage it runs in? Cold traffic needs UGC or lifestyle content, a problem-aware or curiosity hook, and a value or aspiration angle. People in the consideration phase respond to testimonials and demos, social proof, and guarantees. Retargeting wants urgency, a discount or FOMO, and a strong risk reversal. A guarantee-heavy discount shown to someone who has never heard of you is simply the wrong recipe.

The second lens is the coverage grid. Put your personas down the side and your funnel stages across the top, then drop every ad into a cell.

Persona ↓ / Stage →	TOF	MOF	BOF
Persona A	✅	✅	✅
Persona B	✅	✅	⬜
Persona C	⬜	✅	⬜

A healthy account fills most of the cells and spreads spend across them. Then you apply one rule. No single cell should hold more than about 40% of your active creative. Once one persona and one stage swallow the account, you have a monoculture with good production values.

The third lens is the spread. Are you actually hitting different hooks, claims, and offers, or does it just look that way? This is the per-axis table above. You can run five formats and still collapse into one hook and one product spec, restated fifty times.

What this looks like in the wild

We recently analyzed a brand running 122 active ads. On paper it had five formats and three funnel stages, which sounds diverse. But 61% of the classified creative sat in a single cell, one persona at the mid-funnel stage in a lifestyle format. When we looked closer, 23 of the ads were the exact same combination. Same persona, same stage, same format, same aspirational offer. Top-of-funnel had two ads total. Only 6% of the account carried a real offer, and only 16% carried any social proof.

That brand did not have a volume problem. It had a diversity problem wearing a volume costume.

Why this quietly kills accounts

You fatigue your audience fast. Twenty-three versions of one idea wear out the same people at the same time. Frequency climbs, click-through drops, and you end up blaming the algorithm.
You cap your own scale. Meta's delivery system is hungry for variety, so one idea only finds you one kind of buyer.
You go invisible to everyone else. When every ad speaks to one persona at one stage with one claim, the other personas you could be serving never hear from you.

Run the audit on yourself

Pull your active ads and tag each one honestly. Note the persona, the funnel stage, the format, the offer, the hook, the product claim, and the objection it answers. Then look for the collapse.

Is any persona and stage cell over 40% of your ads?
Does every persona get all three funnel stages, or is one persona carrying the whole account?
How many genuinely different product claims are you making, or is it one spec restated?
Is your top-of-funnel starved while retargeting is bloated?

If the grid comes back lopsided, your next creative brief has already written itself, and it is not "make twenty more of the winner."

Coverage was always the metric

Volume was never the real metric. Coverage is. A hundred ads that say one thing to one person add up to a beautifully produced blind spot. The brands that actually scale run ads that reach different customers and meet them at different points in their journey. They are not re-dressing one ad a hundred times.

How to Run a User Research Study

2026-07-08T00:00:00Z

And now, to the blog.

A step-by-step field guide — and every quiet decision point that determines whether you walk away with the truth or just expensive noise.

You need to be running user research. Not "someday when we have bandwidth." Now.

Here's why I'm confident saying that, even though I don't know your business. In 2026, the diagnostic layer is basically free. AI can crawl your site and hand you a list of everything technically wrong with it — contrast failures, layout shifts, broken heuristics, missing labels. Your analytics can tell you where people fall off: which step of checkout leaks, which page has the ugly bounce rate, which plan card nobody clicks. Both of those things are real. Both are useful. Use them.

But neither of them can tell you the one thing that actually moves revenue: how people feel about your product, and whether the way you explain it lands the way you think it does.

Analytics tells you that someone left the pricing page. It will never tell you they left because they assumed the underlined feature names were clickable links, hovered, got nothing, and quietly decided the page felt broken. An AI audit will tell you your comparison matrix is "well-structured." It will never tell you that a customer did the add-on math in his head — "$15 plus $8 for call recording is $23, that's only ten bucks under Pro, might as well go Pro" — and talked himself into a different plan than the one you were steering him toward.

That information doesn't live in your codebase or your GA4. It lives inside a human's head. The only way to get it out is to talk to them.

And here's the part everyone racing to automate their way out of a conversation is missing: in an AI world, talking to your customer is worth more, not less. Everyone has the same audit tools now. Everyone can point Claude at a landing page. The heuristic checklist has been commoditized. What hasn't been commoditized is knowing — actually knowing, from their mouth — what your customer was afraid of, what almost stopped them, and what finally made them trust you enough to put in a credit card. That understanding is the last durable advantage, and it's the raw material for every relevant experience you'll ever build.

There's a second reason, and it's bigger. AI-human interaction is no longer a niche — it's becoming the interface. Your customers are already talking to chatbots, copilots, AI search, and agents inside your product and everywhere around it, and that behavior is only going one direction. Nobody has this figured out yet. How people actually feel about talking to an AI, when they trust its answer versus quietly route around it, where it delights them and where it makes them abandon — that is entirely uncharted, and it is precisely the kind of thing you can only learn by observing and researching real humans in the act. If you're going to build for AI-human interaction — and you are, whether you meant to or not — you have to study the interaction itself. We watched this happen in the Phone.com sessions: one participant said she'd sooner use the on-site chatbot than a general model like Gemini because she trusted it not to blend in everyone else's wrong information; another said he'd just "use AI to find out" what a feature meant rather than read the page. Those are AI-human interaction findings, and no amount of auditing your own site would surface them. The only way to build good AI experiences is to research how humans actually behave inside them.

And underneath all of it is the thing no model can do for you: empathy. Empathy is the core of research practice — not a soft add-on, the actual engine. It's the researcher's job to sit in another person's frustration, to feel the confusion of the underlined-link-that-isn't, to understand why a customer's trust broke and not just log that it did. AI can transcribe empathy's outputs; it cannot feel. And because meaningful, relevant, creative experiences — the ones that keep customers coming back — are built on genuinely understanding how people feel, UX research doesn't get automated away in an AI world. It becomes one of the few things that still has to be done by a human, for humans. That's not nostalgia. It's the whole reason the discipline survives.

So: talk to your customers. Here's how it's actually done.

Fair warning before you scroll. This is going to look simple. A few emails, a few calls, a spreadsheet of quotes. It is not simple. Every phase below contains a dozen forks, and the maddening thing about research is that the wrong fork doesn't announce itself — it quietly poisons everything downstream in a way you will not detect until you've shipped the wrong thing. I'll flag the forks as we go. To keep it concrete, I'll walk you through a study we recently ran on the Phone.com pricing page, start to finish.

Step 0: Decide what you're answering — and how you'll answer it

Before you write a single email, you have to know what kind of question you have. Get this wrong and everything downstream is beautifully executed research aimed at the wrong target.

Two axes matter.

Generative vs. evaluative. Generative research asks "what's going on here, what should we even build?" Evaluative research asks "here's a thing — does it work?" You run these completely differently. Generative work is open, exploratory, and you shut up and listen. Evaluative work puts an artifact in front of someone and watches them struggle with it.

Attitudinal vs. behavioral. What people say and what people do are different data, and they routinely contradict each other. If you only ask people what they think, you'll get a tidy, confident, and frequently wrong story. The best studies capture both — ask about the experience, then watch them actually try to use the thing.

Our Phone.com study was deliberately a hybrid. The goal, in plain language: understand what created enough confidence for a small business owner to trust Phone.com with their business and commit to a plan. The first two-thirds of every session was attitudinal and generative — tell me the story of how you got here. The last third was evaluative and behavioral — here's the live pricing page, now show me what you'd actually do. One session, two modes, and the design of the whole thing flowed from that decision.

But knowing what you want to learn is only half of Step 0. The other half is choosing the method — and this is where a lot of otherwise-smart teams skip straight to "let's do some interviews" without asking whether interviews are even the right instrument for this intent. Moderated interviews, unmoderated usability tests, diary studies, surveys, card sorts, tree tests, contextual inquiry, A/B tests — each answers a different kind of question, and each one distorts the data in its own characteristic way. Method selection is research design. We chose moderated 1:1s for Phone.com specifically because we wanted the why behind a high-trust financial commitment, and only a live conversation lets you follow the thread when someone says something surprising. A survey would have given us scale and told us nothing about the moment a customer's confidence formed or broke.

And every method drags its own biases in the door. You don't get to avoid bias; you only get to know which ones you've signed up for and design against them. The big ones to name out loud:

Temporal bias. Memory decays and, worse, it reshapes — people smooth a messy real experience into a clean story that never happened. It's the entire reason we required customers who'd signed up within the last 90 days. The further out you recruit, the more you're studying a reconstruction, not an experience.
Framing bias. How you frame the study and each question changes the answer. Tell someone up front you're "improving the pricing page" and they'll dutifully find things wrong with the pricing page — even if their real problem was three steps later.
Question order bias. Ask about price early and you've just told the participant that price is what matters here; every answer after that bends toward it. Sequence is a variable you're setting whether you think about it or not.
Social desirability bias. People want to be liked by the friendly person on the call, so they soften criticism and inflate praise. Left undefused, it turns your whole study into flattery, and flattery is worthless data.

Different methods amplify different ones — an in-person session juices social desirability, a survey lets framing bias run wild with nobody there to catch the confusion, a diary study fights temporal bias but invites people to perform for the journal. Picking the method means picking your battles.

Forks that shape everything after this:

Are you trying to discover a problem or validate a solution? (Different studies. Don't blend them by accident.)
Do you want to know what people believe, what they did, or both?
Which method actually fits this intent — and which biases does that method invite that you'll need to design against?
What's the single sentence describing what you'll know at the end that you don't know now? If you can't write it, you're not ready to recruit.

Step 1: Recruiting — the part that quietly kills most studies

Ask anyone who does this for a living where studies die, and they'll say recruiting. Not analysis, not moderation. Recruiting. It is unglamorous, it takes longer than you budgeted, and if you cut corners here nothing else matters, because you'll be interviewing the wrong people flawlessly.

The core question is: who do you need to talk to, and how do you find enough of them without contaminating the sample?

For Phone.com we wanted recent self-serve customers — people who'd been through the exact experience we were studying while it was still fresh. That single requirement generated a cascade of recruiting decisions. We reached out by email, offered an Amazon gift card in the $20–$50 range as a thank-you, and gave a one-click scheduling link. Then a follow-up to non-responders five to seven days later — because your first email will underperform and one nudge recovers a meaningful chunk of your sample.

Sounds mechanical. It isn't. Look at the decisions baked into that paragraph.

Forks that shape everything after this:

Incentive amount. Too low and you get nobody, or worse, only people with nothing better to do. Too high and you attract professional study-takers who'll tell you whatever keeps the gift cards coming. The right number depends on how hard your segment is to reach and how much of their time you're asking for. A busy business owner is not a $10 conversation.
Recruit from your own customers or a panel? Your customers give you real, relevant experience but they're biased toward people who stayed. Panels are faster but you're often paying for strangers pretending to fit your criteria.
Timing of the ask. We required people who signed up within the last 90 days. Memory decays fast and it decays dishonestly — people don't forget, they confabulate, smoothing a messy real experience into a clean story that never happened. Ninety days is a judgment call balancing recall against sample size.
The follow-up. One nudge, or three? Too few and your sample is tiny. Too many and you're annoying the exact people whose goodwill you need.

None of these has a "correct" answer written on a card somewhere. Each is a tradeoff, and each one silently tilts who ends up on your calls.

Step 2: The screener — who you let in is your study

If recruiting is where studies die, the screener is where they get quietly compromised. A screener is the set of gate questions that decides who qualifies. It looks like a formality. It is actually the single highest-leverage document in the entire study, because your findings can only ever be as good as the people you let into the room.

Here's a taste of the logic we built into the Phone.com screener. Every one of these is a deliberate exclusion, and every exclusion is defensible for a specific reason:

"Who made the decision to sign up?" If someone else at their company chose Phone.com, this person cannot tell you why the pricing page worked — they never evaluated it. They're disqualified. And a joint decision? We didn't auto-include or auto-exclude; we told the recruiter to use judgment and only continue if the person was actively involved in the evaluation. That's a human call, mid-screen, that a checkbox can't make.
"Did you sign up directly through the website?" If they were set up by someone else, they didn't experience the flow we're studying. Out.
"How many people use your account?" We capped it. A 40-seat deployment is a fundamentally different buyer than a solo operator, and mixing them muddies every theme.
A recruitment priority cascade. Basic-plan customers were the primary target and got scheduled first. Only if we couldn't fill the Basic quota after two weeks would we open to Plus, and Pro was last. That ordering isn't bureaucracy — it's protecting the sample so the study answers the question we actually asked.

Forks that shape everything after this:

Every criterion trades sample purity against how long recruiting takes. Tighten the screener and you get cleaner data more slowly; loosen it and you fill the calendar faster with people who blur your findings.
Watch for the participants who game the screener — the ones who figure out which answer keeps them eligible and give it to you. A well-built screener disguises which answer is the "right" one.
The prioritization logic is itself a design decision. Get the target segment wrong and you've run a perfect study about the wrong customer.

I want to be blunt about this one: most in-house teams write a three-question screener in ten minutes and wonder later why their findings feel mushy. It's because they let the wrong people in, and by the time you're staring at the transcripts, there is no undo.

Step 3: The first three minutes — consent, framing, and defusing the bias you're about to introduce

You've got the right person on a Zoom call. Before you ask them anything of substance, the way you open the conversation determines the quality of every answer that follows. Most people wing this. You shouldn't.

A good intro script is doing several jobs at once, quietly:

Consent and recording, handled cleanly. You explain the recording is internal-only, not for marketing, and you get a clear yes. Notice a subtle decision here: in our study, recording was preferred but not required. If someone declined, we still ran the session rather than lose a hard-won participant. That's a deliberate tradeoff — data richness vs. sample size — decided in advance so the moderator isn't improvising it live.
Lowering the stakes. "Nothing you share today will affect your account or billing in any way." That single line removes a confound you might not have thought about: a customer who's worried that criticism will cost them will soften everything they say.
Killing social desirability bias before it starts. The best framing we used: "Honest reactions are more useful to us than positive ones. This is not a test." People instinctively want to be nice to the friendly person on the call, and "nice" is worthless data. You have to actively give them permission to be negative, or they'll flatter your product straight into a bad decision.
Setting up the think-aloud. You tell them, up front, that later you'll ask them to narrate their thoughts out loud — even the small or off-topic ones. Warming them into that habit early means they're actually good at it when it matters.

Forks that shape everything after this:

How much do you tell them about what you're looking for? Too much and you've primed them — now they're hunting for the thing they think you want. Too little and they're disoriented. There's a narrow band of "just enough."
Do you reveal that an agency, not Phone.com, is running the session? (We did — it subtly frees people to criticize the product without feeling like they're insulting you to your face.)
Recording required or optional — and who decides in the moment if they hesitate?

Step 4: The interview guide — how the questions are built, and why the order is everything

Now the actual questions. And here's where amateur research does the most damage, because writing questions feels like the easy part and it is the hardest.

The problem is that questions leak. A leading question — "Did you find the pricing page easy to use?" — hands the participant the answer you want and they'll take it, because it's the path of least resistance. You have to build questions that extract the real story without steering it.

Our Phone.com guide was structured as a funnel, wide to narrow, and the sequence was load-bearing:

Context first. Tell me about your business. Warm, easy, and it grounds everything that follows in who this person actually is.
The trigger. What made you start looking for a phone system when you did? Was there a specific moment? We're mining for the real event that kicked off the journey — not "I needed a phone" but the actual moment of pain. (For several participants it turned out to be wanting to separate business calls from personal ones. That's not a feature request; that's the emotional core of the purchase.)
The decision process. Did you look at other options? What made you choose this one? What almost stopped you? That last question — "what almost stopped you" — is gold. It surfaces the friction and the fear, the stuff that's invisible in analytics because those people either barely converted or bailed silently.
The reveal, held until the end. Only now do we show the live pricing page.

That ordering is not an accident, and it's the detail I'd most want you to steal: we deliberately did not show anyone the pricing page until the final third of the session. If you put the artifact in front of someone first, you've contaminated their whole story — everything they tell you afterward is shaped by staring at your design. By getting their unprompted journey first and the page reaction second, you get two clean datasets instead of one muddy one.

Forks that shape everything after this:

Open vs. closed questions. Open questions get you stories; closed questions get you yes/no dead ends. Almost everything should be open.
Question order and priming. Ask about price too early and you've told them price is what matters. Every early question frames the ones after it.
How hard do you probe? "Tell me more about that" versus leading them somewhere. The line between good probing and putting words in their mouth is thin, and you cross it without noticing.
What you deliberately don't ask, so you leave room for them to volunteer the thing you didn't know to look for.

Step 5: The think-aloud — watching, not asking

The final section is a different skill entirely. You share your screen — or better, have them share theirs — put the real page in front of them, and ask them to narrate every thought while they use it. This is the behavioral half of the study, and it's the closest you'll get to watching the truth happen live.

The technique has a name and a pedigree — the think-aloud protocol comes out of decades of cognitive research on how to externalize what's happening in someone's head in real time. The discipline is this: you ask them to do something ("walk me through how you'd choose a plan if you were seeing this for the first time"), and then you stay quiet and let them struggle. The struggle is the data. Every instinct will scream at you to jump in and help. Don't. The moment they get stuck and don't know you're watching is the most valuable ten seconds of the entire study.

This is exactly the section that produced our sharpest findings on Phone.com — the ones no audit tool on earth would have surfaced:

Multiple participants, independently, assumed the underlined feature names were clickable links. They tried to click. They tried to hover. Nothing happened, and the page briefly felt broken to them. No heuristic scanner flags "this text implies interactivity it doesn't have," because it's not a bug — it's a mismatch between the design's signal and a human's expectation. You only find it by watching a human expect the wrong thing.
Participants did add-on math out loud — "Basic at $15, plus $8 for call recording, that's $23; Plus is $22.50, basically the same, so I'll just take Plus." That real-time mental arithmetic is the actual decision mechanism, and it was completely invisible in the funnel data.
One returning customer narrated the exact moment his confidence broke — after signup, when an SMS setup required a carrier-registration step nobody had disclosed, and his personal and business numbers started "fighting each other." That's a retention risk hiding one step past the checkout event where every analytics dashboard stops looking.

Forks that shape everything after this:

Do you give them a task ("choose a plan") or let them free-scroll first? (We did both — free scroll to capture first reactions, then a directed task.)
How long do you let silence sit before you prompt? Too soon and you rescue them out of the finding; too late and they shut down.
Their screen or yours? Theirs is more real but adds tech friction that can eat your session time.

Step 6: Moderation — the invisible craft

Everything above is preparation. Moderation is the live performance, and it's where experience separates from enthusiasm. A good moderator is doing several things simultaneously, in real time, for thirty minutes: following the guide but knowing when to abandon it for a thread worth pulling; staying warm without being agreeable; hearing a throwaway comment and recognizing it as the most important thing said all session; and — hardest of all — keeping their own opinions, hopes, and body language completely out of it.

Your face is data to the participant. If you light up when they praise the design and flatten when they criticize it, you've trained them, mid-session, to praise. Neutrality is a skill, and it's exhausting.

There's no clean list for this one, because moderation is the part that genuinely doesn't reduce to a checklist. It's reps. It's the difference between a session that produces a quote you'll build a roadmap around and a session that produces forty minutes of politeness.

Step 7: Synthesis — the discipline of not fooling yourself

You've run five interviews. You now have hours of transcript and a strong temptation to skim it, pull the three quotes that confirm what you already believed, and call it insight. That temptation is the enemy. Synthesis done right is a formal discipline designed to stop you from finding only what you went looking for.

The method is thematic coding, and the sequence matters:

Raw data. Every notable quote and observation, pulled verbatim, attributed to a participant. You work from their words, not your memory of their words.
First-level codes — inductive, from the data up. This is the non-negotiable rule and the one most people break: codes come from what participants actually said, not from your hypotheses. You are not checking boxes you drew in advance. You're letting the labels emerge from the transcripts. On Phone.com that produced codes like "pricing transparency as differentiator," "false interactivity signals," "add-on math narrows price gap," "social proof irrelevance." Some of those we expected. Several we did not — which is the entire point.
Themes — patterns across people. A code becomes a theme when it shows up across multiple participants. "Transparent self-serve pricing is the primary conversion driver." "Post-signup complexity is the primary retention risk — the pricing page isn't the problem, the setup after it is." That second one reframed the whole engagement: we came in to study a page and the data pointed a floor below it.

And then the piece of intellectual honesty that separates real research from theater: directionality. Every finding in our study was explicitly labeled directional until confirmed across three or more participants. Two people saying the same thing is a hypothesis, not a fact. We tracked what we didn't yet know in a running Open Questions list — is the post-signup SMS friction a pattern or just one person's bad week? Does outcome-framing land with less technically confident users? You state your uncertainty out loud instead of laundering two anecdotes into a confident recommendation.

Forks that shape everything after this:

Code inductively (from the data) or deductively (against a pre-built framework)? Deductive is faster and will cheerfully confirm your existing beliefs. Inductive is slower and is the only way to be surprised.
What's your bar for "this is a theme"? Set it too low and every stray comment becomes a mandate.
How do you handle the finding that contradicts your favorite hypothesis? (You keep it. Especially that one.)
How honestly do you represent sample size? Five interviews is directional. Anyone who hands you "customers want X" off five sessions with no caveat is selling you confidence they haven't earned.

Where AI helps — and where it will confidently lie to you

Because this is 2026 and someone's already thinking it: can't I just get AI to do all of this?

Parts of it, genuinely yes, and you should. AI is excellent at transcription. It's a real accelerant on first-pass coding — feed it clean transcripts and it'll surface candidate codes fast. It can help screen recruits and draft your guide. It compresses the grunt work enormously.

Where I find it most valuable, though, is as a thinking partner around the work rather than a replacement for it. It's a genuinely good brainstorming collaborator for the craft decisions: give it your draft questions and it'll offer ten different ways to phrase the same thing so you can hear which one leads and which one opens. Ask it to reorder your guide and it'll surface a sequencing you hadn't considered. Prompt it to "challenge my assumptions here" or "what am I not asking that I should be?" and — when you explicitly invite the pushback — it's sharp at flagging the gap in your study design before you've spent your recruiting budget discovering it live. It's also excellent downstream: brainstorming which visualization actually fits a given shape of data, pressure-testing how to tell the story so the finding lands and doesn't drown, and — the one I lean on constantly — reframing the same insight for different stakeholders. The way you present a trust-breakdown finding to a design lead is not the way you present it to a CFO, and AI is a fast, tireless partner for working through that translation.

And then there's the line you cannot cross, which is the whole reason this article exists:

Synthetic users are not users. The pitch that you can skip recruiting and just ask an AI to "role-play your customer" is seductive and wrong. A model will generate the statistically average plausible answer — the smooth, confabulated story — which is the exact thing real research exists to get past. It will never tell you it assumed the underlined text was clickable, because it never actually looked at your page with a human's expectations and got quietly annoyed.
AI summarization flattens the outlier, and the outlier is frequently the finding. Ask a model to summarize five interviews and it'll regress toward the consensus. The single participant who narrated the exact moment his trust broke gets averaged into a bullet point and lost.
AI will hand your bias back to you, dressed as insight. Prompt it toward the conclusion you want and it will oblige, fluently and with citations. It doesn't have the disinterested spine to tell you that your favorite hypothesis just died on the transcripts.

The honest framing: AI makes a skilled researcher dramatically faster. It does not make an unskilled one safe. It removes the labor, not the judgment — and this whole discipline is the judgment.

So — should you run one yourself?

You now have the whole map. Define the question. Recruit the right humans without contaminating the pool. Build a screener that guards the sample. Open the session so people tell you the truth. Sequence questions that don't lead. Hold the artifact until the end. Shut up during the think-aloud. Moderate without leaking your own hopes. Code inductively so you're capable of being surprised. And label your confidence honestly instead of manufacturing it.

Every one of those steps has three to twelve forks in it. And here's the thing I most want you to sit with: a wrong turn at any fork doesn't fail loudly. It doesn't throw an error. It produces a clean-looking deck full of confident findings that happen to be about the wrong customer, or shaped by a leading question, or built on two anecdotes wearing the costume of a trend. You will not be able to tell from the output that it's wrong. You'll ship against it. You'll wonder why the winning test didn't win.

That's the real cost, and it's why this is a craft and not a checklist.

Can you run a user research study yourself? Absolutely. Should you run the one your revenue decisions depend on without someone who's made all these mistakes already? That's a different question — and now you know enough to answer it honestly.

If you'd rather not learn which fork was the wrong one after you've already shipped, that's what we do. We ran exactly this study for Phone.com, and we can run it for you — or sit alongside your team and make sure the forks get taken correctly the first time.

Either way: go talk to your customers. In a world where everyone has the same audit tools, the people who actually understand the human on the other side are the ones who win.

Your Returning Customers Are Carrying the Business. You've Never Asked Them Who They Are.

2026-07-15T00:00:00Z

I was digging through GA4 for a specialty retailer client recently, the kind of pull I do at least once a quarter for every account, and one number stopped me.

Returning visitors made up 20% of sessions.

They made up 77% of revenue.

I sat with that for a minute. Every experienced e-commerce person has a rough sense that repeat visitors convert better. But 77% isn't "convert better." That's four out of every five dollars, coming from one out of every five people, and when I asked the team what they actually knew about that 20%, the answer was nothing. No name. No email. No mechanism for anyone to say "hey, it's me again."

Why This Number Gets Ignored

It's easy to miss because it doesn't show up as a problem. Revenue is fine. Conversion rate looks normal in the dashboard. Nothing is on fire.

But flip the number around and it gets uncomfortable. If 77% of your revenue depends on people you can't identify, you don't actually have a customer base. You have a temporary majority of anonymous repeat visits that happens to be propping up the P&L.

Here's the part almost nobody says out loud: the reason people don't identify themselves isn't that they don't want to. It's that you never asked.

Go look at your own site right now. Is there an actual reason for someone to log in before they're ready to check out? Does your homepage know the difference between someone's first visit and their ninth? Most sites are built as if every session is session one, forever, no matter how many times that browser has been back.

What Not to Do About It

The instinct here is to bolt on a loyalty program. Points, tiers, a "Rewards" tab in the nav. I get why. It feels like the obvious fix for "we don't know our customers."

Don't start there.

A points program is a mechanism for retaining people who already identify themselves. It does nothing for the anonymous 20% who are already coming back on their own, without any incentive, and still aren't raising their hand. If people are already returning at that rate without a rewards program, the problem isn't motivation. It's recognition. You're solving the wrong layer of the funnel.

The other wrong move is forcing an account creation wall before checkout. That trades a data problem for a conversion problem.

What Actually Works

Start with the visit, not the transaction. The lowest-friction identity signal is simply knowing someone has been on your site before, even before you know their name. Cookie-based recognition alone lets you shift the homepage and PDP experience for returning sessions: skip the brand-story intro, surface what they were last looking at, tell them what's changed since they left. None of this requires a login. It just requires building two versions of the top of the funnel instead of one.

Make the entry discount earn its keep. Almost every brand already runs some version of "10% off your first order." That discount is doing one job right now: buying a transaction. Make it do two jobs. Instead of just an email address, ask for the two or three pieces of information that would actually let you personalize the next visit. For a golf retailer, that's handicap, how often they play, and what driver is already in their bag. For a skincare brand, it's skin type and the one thing they're trying to fix. For a supplement company, it's the specific outcome they're chasing. You're not making the discount harder to get. You're making it smarter. Same conversion event, but now you walk away with a customer profile instead of just an email address.

Use the membership program you already have, don't build a new one. If you've already got a Players Club, a Rewards tier, a VIP list, anything with an account behind it, that's your on-ramp. Most of these programs collect a name and maybe a birthday and call it done. Rebuild the signup flow to ask two or three real questions at the moment someone's already motivated to join, not buried in a "complete your profile" page nobody visits later. The membership program was never the problem. The account creation flow inside it was.

Consider putting a quiz at the front door, not the back. Most DTC brands already do this, a short quiz as the actual first touch, before any product grid. Sporting goods and specialty retail rarely do, and there's no obvious reason not to test it. Someone lands on the site, answers four questions about how they play or what they're solving for, and gets routed to the right products instead of a generic category page. It's a bigger swing than the other three, worth treating like a real experiment rather than a redesign: run it against your current homepage for a segment of traffic, and see what it does to conversion and to how many people hand you usable data. And don't ignore the signal that's already free: reviews and sizing feedback often contain the exact detail you'd otherwise have to ask for. Most brands just aren't looking.

The Real Ask

You don't need to solve identity for 100% of your traffic. You need to solve it for the 20% who are already proving, with their wallets, that they're coming back.

Pull your own GA4 numbers this week. Compare session share to revenue share for returning visitors. If the gap surprises you, you're not alone. Almost every brand I've looked at has some version of this.

If you disagree with any of this, send me an email and tell me why.

The Message-Match Gap: What the Best DTC Stores Do That the Worst Don't

2026-07-23T00:00:00Z

We call the distance between what an ad promises and what its landing page delivers the message-match gap. To see how much it actually costs brands, we analyzed the Meta advertising of 22 leading DTC brands: reading what each ad promises (the words baked into the creative, the words spoken in the video) and checking, piece by piece, whether the destination page delivers it.

This isn't a vibe check. Every ad-to-page pair gets broken into the specific promises the ad makes: the primary promise, the offer, specific claims, audience signals, tone, urgency, and social proof. A page either carries each one or it doesn't. A promise the page drops is a gap.

Then we ranked all 22 brands by how many gaps they rack up per ad. The distance between the top and the bottom isn't subtle.

The Best: Quince (1.6 Gaps Per Ad)

Promise kept

The ad

The page it links to

quince.com/…/european-linen-shirt

The exact shirt from the ad, $34 (down from $98, "you save 65%"), 3,056 reviews and a "Best seller" flag, all settled above the fold.

Quince runs a warm, word-of-mouth style ad: "My husband loves these linen shirts from Quince." Click it, and you land on the exact product being talked about, the 100% European Linen Relaxed Short Sleeve Shirt, with the whole promise settled before you scroll:

The exact shirt from the ad, front and center.
$34, marked down from $98 ("you save 65%") right next to the price.
3,056 reviews, a "Best seller" flag, free shipping, and 365-day returns, all above the fold.

The ad made a promise. The page paid it off, in full, instantly. Across all five of Quince's ads, we found 8 gaps total, 1.6 per ad, and every one of them was cosmetic: a missing trust badge here, a slightly under-specified audience cue there. Quince never drops the promise, never drops the offer, never leaves a visitor wondering if they're in the right place.

Bombas ties Quince for the fewest gaps in the entire index (also 1.6 per ad). The pattern at the top is consistent: the core message always survives the click.

The Worst: Fabletics (6.6 Gaps Per Ad)

Promise lost

The ad

The page you actually get

fabletics.com/scrubs

A specific ad about one color for summer leads to a generic "join VIP" page, with a "tap a gift box" pop-up sitting on top of the hero.

Fabletics runs a genuinely good, specific ad:

"Introducing Vintage Grey, your new go-with-everything scrubs color this summer! For a limited time, get your first scrub set for just $15 as a new VIP signup."

A color. A season. A versatility angle. A price. Click it, and you don't land on Vintage Grey. You land on a generic "join VIP" membership page, greeted by a gamified pop-up ("A gift just for you, tap a gift box to unwrap your discount") sitting on top of the hero. Vintage Grey survives as one swatch in a row of ten. The "go-with-everything" pitch, the summer story, the specific reason you clicked: gone. The ad sold a product. The page sells a subscription.

That's the pattern for every Fabletics ad we looked at: a sharp, specific hook poured into the same one-size-fits-all funnel. Across their five ads we counted 33 gaps, 6.6 per ad, roughly four times Quince's rate. And these aren't cosmetic. Six or seven of every ad's eight promises were gaps.

The Pattern Holds Across the Index

Zoom out from the two extremes and the difference between the best and worst halves is structural, not random:

The best brands never drop the load-bearing pieces. Across the six lowest-gap brands, we found zero dropped offers and only a handful of weak primary-promise or tone components. Their misses are refinements.
The worst brands drop exactly those pieces. Across the six highest-gap brands, the most common gaps were the primary promise, the tone and identity of the brand, and the offer itself: the things a visitor needs to see in the first two seconds to believe they're in the right place.

Message match isn't a tie-breaker that separates good stores from great ones. It's the difference between a page that confirms the click and a page that silently asks the visitor to start over.

Closing the Gap: A Live Client

Knowing where the gaps are is one problem. Closing them, across every ad, every week, without rebuilding your site, is the harder one. That's what Throughline does. Here's one live client, a DTC apparel brand.

We ran the client's Meta ads through the same diagnosis, then generated a matched landing-page variant for each one. Across 37 rebuilds, Throughline diagnosed and closed the majority of the gaps standing between the brand's ads and its pages:

gaps diagnosed

gaps closed

60%+

of the gap, gone

How It Closes a Gap: One Real Example

The client ran a Father's Day ad: your dad wants a new polo, plus the gear from their latest sneaker-brand collab. The collection page it linked to led with a generic "the new drop is now live, shop now" banner. None of the gifting occasion. None of the collab. Five gaps, one for each promise the ad made and the page ignored.

Throughline's fix didn't invent new marketing copy. It borrowed the ad's own words back onto the page, the principle the whole engine runs on:

Primary promise: a hero subhead delivering the Father's Day gifting line the ad opened with, above the fold.
Urgency: the announcement bar rewritten from "now live" to the time-sensitive Father's Day occasion the ad established.
Audience and tone: a short intro reinstating the gift-buyer framing and the brand's playful voice.
Specific claim: a callout naming the exact sneaker collaboration the ad led with.

Four of the five gaps closed. Same products, same brand, same page. Now it finishes the sentence the ad started.

The Early Results

Every one of these variants runs as a true A/B test against the original page, so the lift is measured, not asserted. In the first weeks of live testing on Meta, the rebuilt pages are running ahead of the originals: a variant conversion rate of roughly 4.3% versus 3.1% on the control, +38% relative.

The honest caveat matters. That's an early, directional signal built on about two dozen conversions. It's not yet statistically significant, and on a sample this small the number will move as more traffic accumulates. We're reporting the direction, not planting a flag. But it's the direction you'd expect when the page finally says what the ad promised.

The Takeaway

The best DTC stores and the worst aren't separated by better creative or bigger budgets. They're separated by whether the promise survives the click. Quince's pages finish every sentence its ads start. Fabletics' pages make you start over.

Closing that gap doesn't require a redesign or a new copywriter. It requires reading what your ad actually promised and making sure the page delivers it. That's a measurable number. Most brands have never looked at it.

See where your own ads and pages don't match. Run the free demo. →

Half Your Meta Ad Clicks Might Not Go Where You Think

2026-08-12T00:00:00Z

Product extensions are the grid of catalog products Meta staples onto your ad. Someone taps the ad, and alongside your landing page they get a little storefront of tiles pulled from your product feed. Tap a tile, land on a product page. Most advertisers turned this on with a checkbox at some point and have never looked at it since.

I pulled ad-level insights broken down by product for every product-extension ad across five accounts we have data access to. 122 ads, roughly $270k in spend.

The grid takes a lot of your traffic

Depending on the brand, somewhere between 12% and more than half of all link clicks went through the product grid instead of the primary link. One ad sent 99% of its clicks that way.

So half your paid traffic can be taking a route you have never tested, onto pages you probably haven't looked at in a year.

And it converts better

For the two multi-product brands in the set, the grid clicks were the strongest clicks the ads produced.

At the apparel brand, grid clicks converted at 5.8% against 2.1% for the primary link, with revenue per visitor of $8.26 against $2.46. Put differently: 53% of the clicks produced 79% of the revenue. At the menswear brand the gap was wider still, 14.2% against 2.3%.

Average order value came out nearly identical on both paths, which tells you the difference is conversion rate and not basket size. That makes sense when you think about who is clicking. Someone who taps your primary link liked your ad. Someone who taps a product tile liked your ad and then pointed at the specific thing they want. The second person is much further along.

Unless you sell one thing

The third brand sells a single high-ticket product, and for them the grid slightly underperformed the primary link. On their creator videos it was worse than that. One video pushed 89% of its clicks into the grid, where they converted at 0.2%, while the landing page was converting at 6.6%. All that warm traffic went window shopping.

So the honest answer to "do product extensions work" depends on your catalog. Real assortment at accessible price points, and the grid is your best-performing surface. One hero product, and it may be costing you.

Is it the product or the ad?

Mostly the product. Across one brand's ads, product identity explained 57% of the variance in which tiles got clicked, and the same few caps and accessories won almost everywhere regardless of what the ad was about.

The ad still steers it, though. Ads built around a specific accessory produced grids that converted at 11.7%. Brand-story ads produced grids converting at 3.8%. Same catalog, same tiles, three times the outcome.

The result I keep coming back to: at the menswear brand, grid clicks on products that didn't match the ad converted better than clicks on the product being advertised, 17.6% against 12.7%. People watched an ad for one shirt and bought a different one. The grid is doing cross-sell work inside the ad unit.

What I'd do with this

Start by pulling your own split. A product breakdown on your ad insights will tell you in ten minutes what share of your clicks are going through the grid, and for most people that number is going to be a surprise.

Then treat it as a decision. Big catalog, mid price points: lean in. One considered purchase: test turning it off, especially on video.

And look at where those clicks are landing. Every grid click skips the landing page you built for the campaign and drops someone onto a product page that has no idea which ad they came from. That is a lot of high-intent traffic arriving at your least prepared pages.

Standard caveats: this is Meta-attributed conversion data and not a holdout test, and a couple of the cells are small enough that I wouldn't bet the quarter on the exact figures. The traffic share is the part I would act on. That one is hard to argue with, and almost nobody is looking at it.

Mend The Gap: The Hierarchy of Message Match

2026-08-19T00:00:00Z

Ask ten ecommerce teams whether they do message match and ten will say yes. Ask what they actually did and you'll hear the same answer: they put the ad's headline in the H1.

That definition has been circulating since about 2012, and it has done more damage than ignoring the idea entirely would have, because it's small enough to finish. You copy the headline into the hero, you tick the box, you move on, and you now believe the problem is solved. The visitor who clicked your ad is still confused, and you've lost the vocabulary to say why.

A click is a promise. Somebody watched fifteen seconds of your creative, believed a specific thing about what was on the other side, and paid you with their attention. The landing page either continues that promise or changes the subject. Headline-matching is one way to continue it, and not the most important one.

We've written twice before about the message-match gap and about scoring 21 DTC brands on it. Both pieces broke an ad into the promises it makes and asked whether the page carries each one. What neither said is that those promises are ranked, and that the ranking is the most useful thing about them.

Mind the gap.

The ladder

Read it as a stack, not a list. Each rung only counts when everything under it is true. A page with perfect tonal continuity and no promise is not eighty percent of the way there. It's a page that sounds like your ad while selling something else, which is worse than a mismatch a visitor can name.

Rungs 4 and 5 each bundle several things, so don't try to add the ladder up. And the score is a ratio of what a given ad actually carries, so the denominator moves from ad to ad. There's no hundred-point total to reconcile.

The weights, per component, are these. Primary promise is worth 25. The offer is worth 20. A specific claim is worth 15. Everything left over — constraints, audience signal, CTA continuity, tone, proof — is worth 10.

Every team I talk to is sure their problem is trust. They want more reviews above the fold, another press logo bar, a bigger guarantee badge. Proof is worth 10. The promise is worth 25. The thing they're certain about is the thing that matters least, and I've never once had that conversation go smoothly.

These are the weights we run in production, and we audited them against 298 real ad-to-page messages to find out where the rubric was lying to us. It was. There was an entire class of message we couldn't represent, social proof and endorsements, and it had been getting force-fit into "specific claim" 45 times. So social proof became its own component with its own weight. That's what an audit is for.

Three brands, three rungs. Every score below comes from a live capture of the page on 19 August 2026.

Rung 1: The Promise

Dropped

The ad

The page it links to

drinkolipop.com

A returning fan-favorite flavor, landing on a homepage about digestive health. The word "back" appears nowhere.

OLIPOP started running this ad on 1 May, headlined Flavor in full bloom. Blackberry Vanilla is back after years away. The creative earns that headline literally: an illustrated can sitting in vanilla blossoms and blackberries, purple and cream, closer to a botanical print than a soda ad. If you're an OLIPOP drinker who remembers that flavor, this is a reunion, and every part of the ad plays it that way.

The click landed on the homepage, whose hero reads "Real ingredients. Real refreshment. Real digestive health support."

The promise scored 8 out of 25. Blackberry Vanilla is on the page, purchasable, sitting in a product grid alongside twenty other flavors. The claim that it was gone for years scored a flat 0 out of 15: nowhere does the page acknowledge that it went away, that it came back, or that anyone missed it. The audience signal, the returning fan the ad is speaking to, scored 4 out of 10.

The page does carry a "Limited" badge, and it's on Pineapple Paradise. Scarcity framing exists on that page. It just isn't pointed at the flavor the ad is about.

The ad sold a reunion in vanilla blossoms. The page sold fiber.

This is the most common failure I see and the least discussed, because it doesn't look like a mistake in a spreadsheet. The traffic arrived. The product was available for purchase. Every dashboard says this worked.

Rung 2: The Offer

Dropped

The ad

The page it links to

magicspoon.com/…/custom-mixed-bundle-6-box

The ad's whole headline is a promo code. The page has never heard of it, and offers a subscription discount instead.

Magic Spoon has been running this ad since 30 April, and its headline is the offer: Code TREATSVIP to save. Five dollars off any order, ten dollars off orders over sixty-three.

The landing page did not mention the code when we captured it, nearly four months later. Not five dollars off, not ten dollars off, not the sixty-three dollar threshold. It offered a subscription discount and a free bowl set.

Offer: 0 out of 20. Offer constraint: 0 out of 10.

Notice what the creative shows, too: a man at a kitchen counter. No product, no text, no code. The offer lives entirely in the ad's copy, which is worth remembering the next time someone audits message match by looking at the images.

I like this example because you can feel it without knowing anything about optimization. Your visitor is standing at the register holding a coupon, and the store has never heard of it. Whatever they do next, they're doing it while quietly downgrading their estimate of how much attention you pay.

Rung 2 is also the cheapest rung on the ladder to hold. Nobody has to rebuild a hero to put a promo code on a page. It fails constantly anyway, because the ad was made by the team running the promotion and the page was made by the team running the site, and the promotion has an end date nobody told the page about.

Rung 3: The Claim

Claims dropped

The page it links to

hiyahealth.com/products/kids-daily-probiotic

One ad for four products, landing on one of them. The ad's creative is no longer retrievable — it stopped running, and Meta keeps no inactive ads for this page — so its wording is quoted below verbatim.

Hiya ran this ad from 17 March, and its copy covers the whole product line. I'll print the full readout for this one so you can check our work:

Multivitamin: 15+ essential vitamins and minerals for complete nutrition

Probiotics: Nurture their gut health and strengthen their immune system

Bedtime Essentials: Support restful sleep without melatonin

The click landed on the Kids Daily Probiotic page.

Rung	Score	On the page
1 · promise	10/25	One probiotic product page, not the whole line
3 · claim	5/15	"15+ vitamins," mid-page, about the multivitamin
3 · claim	0/15	Greens powder's 55 ingredients: absent
3 · claim	11/15	No sugar, no gummy additives, non-GMO: above fold
3 · claim	0/15	Sleep without melatonin: absent
3 · claim	15/15	Pediatrician-backed, delivered fresh: carried
4 · audience	9/10	Parents of kids two to twelve, throughout
5 · proof	6/10	6,861 reviews; the ad claims hundreds of thousands
5 · CTA	10/10	"Try Now" / "Add To Cart" matches "Shop now"

Overall: 51.

Two flat zeros, both of them absences. A parent who clicked that ad because their kid won't sleep arrives on a probiotic page, and the word melatonin never appears.

Look at what the page does well, though, because it's the more useful lesson. Three components score 11 or better. The page is good. It's answering a different ad.

The fix isn't subtle either. If one ad sets up four products, either the ad points somewhere that holds four products or the page answers all four. Picking one and hoping is a coin flip you're paying Meta to run.

The message that never arrived

Not every message in an ad reaches a human being. A claim spoken out loud in a video reaches them. A claim printed across the creative reaches them. A claim sitting in the fourth line of Instagram body copy, under a "See more" fold, in a feed being thumbed at speed, does not.

So we floor text-only components to a fifth of their weight. You can watch it happen in the data: a specific claim carried by the creative is worth 15, and the same claim living only in text is worth 3. That has happened to 41 real claims in our corpus. A primary promise drops from 25 to 5. An offer drops from 20 to 4.

Which means you have been auditing the wrong artifact. Your ad copy document is not your ad. Nobody read it. If you want the list of messages your page has to answer, open the ad, watch the video with the sound on, and write down what it claims out loud and what it shows on screen. That list is shorter than your copy doc and it's the only one that binds.

Here's what that exercise turns up.

The ad

The page it links to

cutsclothing.com/products/icon-pull-on-pant

The urgency in this ad is spoken, not written. It exists in no copy document, so nothing on the page answers it.

Cuts runs a video ad for a pant, and the last thing the presenter says is better grab them now before they sell out going into spring. That's a scarcity claim and a seasonal deadline, and it exists nowhere except in the audio. It isn't in the ad's text, so it isn't in any copy doc, so nobody put it on the page. The page has no stock counter, no low-stock notice, no seasonal availability, nothing. That component scored 1 out of 10.

When we first scored that ad from its text alone, the urgency claim wasn't in the list at all, because as far as the text was concerned it was never made. Transcribing the audio didn't change the page's overall score. It changed which rungs were failing.

Said, but buried

A message can also fail to arrive by being on the page in a place nobody who clicked that ad will reach.

Take that same Cuts ad. Its case for the product is a fabric called Enduratech, and in the video the presenter says it looks like denim but is better. On the product page, that claim scored 9 out of 15. Enduratech is named, mid-page, next to badges for four-way stretch and softness and a fibre breakdown of 52% poly, 38% cotton, 10% spandex. The comparison the ad actually makes, that this replaces your jeans, is not there.

Hiya's fifteen essential vitamins: 5 out of 15. Present, in a mid-page comparison table, describing a product that isn't the one you landed on.

Both are technically covered. If you sent either page to the brand team and asked whether they mention it, they'd point at it and be right, and the visitor still wouldn't have seen it.

Which gives a message two ways to fail to arrive. It was never said, or it was said where the arriving visitor won't get to it. Both produce a person who came for one specific thing and now has to hunt for evidence that they're in the right place. Most of them don't hunt.

This is why "we already say that on the page" isn't a defense. Position is part of the claim.

Where to actually start

Severity and sequence are different questions, and the ladder only answers the first.

The promise is worth 25, and fixing it usually means rebuilding a hero, which means design time and somebody's sign-off. CTA continuity is worth 10 and takes an afternoon. So the first thing you fix is often not the most damaging thing, which is fine as long as you know which one you did. Rank by weight over effort, ship the cheap ones this week, and don't let three quick top-rung wins convince you the base is sound.

Fixed in the right order, it moves. For the client currently running Throughline (which automates message matching), we're showing a 31% conversion lift.

What mending it looks like

A golf equipment brand, scored on the same rubric. The ad promises premium equipment from a legendary player's line, designed specifically for women golfers, and it lands on a partnership page.

Rung	Before	After
1 · promise	12/25	24/25
3 · claim	7/15	14/15
4 · terms	4/10	8/10
5 · voice	7/10	8/10

54 to 90, with every gap the diagnosis identified closed.

That's a scored rebuild, not a served one. The page moved from telling a general partnership story to answering the specific promise that ad made to a specific woman who clicked it. Whether it produces revenue is a question I'll answer when a test of ours actually reaches power.

Mend the gap

We scored this ladder by hand for two years before we automated it. The automation is called Throughline. It reads your ads, video and creative included, works out which messages your landing page misses, and rebuilds the page to close them.

If you want to see where your own pages sit on the ladder, email me your five highest-spend ads and I'll send back the readout, rung by rung, with the gaps named. You can also see how the biggest DTC brands rank on the Message Match Index.

How these were scored

Every number here comes from one scoring pass per ad-page pair, run on 19 August 2026. Each ad's messages were extracted from its creative, copy and CTA, then a page captured live that day was scored against them, component by component, blind — the scorer is never told whether a page is an original or a rebuild.

Ad start dates are the Ad Library's, and the gaps described here had all been in place for months by the time we captured the pages: Cuts since January, Hiya since March, Magic Spoon and OLIPOP since the end of April. Whether those ads are still in rotation today we don't know, and it doesn't change what the pairs show.

The Cuts ad's audio was transcribed and scored, which is where the spoken urgency claim comes from. One limitation: the Magic Spoon ad also carries video, and its file is no longer retrievable, so that pair was scored on text and imagery alone. A spoken claim would not appear in its component list, which can only understate its gaps, never invent one. Its two flat zeros are an absent promo code, which no amount of audio would put on the page. Hiya's ad, the one whose full scorecard is printed above, is static, so nothing was missed there.

Two Ads, One Landing Page, and a Rebuild That Only Worked for One of Them

2026-08-27T00:00:00Z

A golf-apparel brand we work with dropped a collection this summer. Two ads went out for it in the same week, both pointing at the same collection page.

The first is a still image: a clean product grid on a cream background. Hats, polos, anoraks, headcovers, a couple of bags, a Nalgene. The caption reads: "It's just a collection that brings us a lot of joy. We've worked really hard on this for a long time. And now it's time to share it with you all."

Click it and you land on a collection page: a grid of hats, polos, anoraks, headcovers, bags. The ad is a grid of the products. The page is a grid of the products. Whatever that ad promised, the page delivers it about a second after the click.

The second ad is a minute of video, and one of the founders narrates it.

Ad one · the product grid

Ad two · the founder's video

Same collection, same week, same landing page. One of these is about products. The other is about a decade-old community project.

Here's what he says:

More than a decade ago, we started the Hidden Gem project. We made a map of all of our favorites. Affordable, accessible, interesting golf courses. Bonus points if there's a good sandwich shop nearby. You guys have even submitted your favorites, and we added them to the map over the years. It's one of my favorite parts about the community we have here.

Then, at the turn:

And this collection is an homage to all of it. Colorways that remind us of an adventure to find a secret golf course.

On screen while he talks: the map itself.

A search box for course, city, or architect. Filters for muni, 9-hole, any fee. A counter reading 150 gems, and a prompt to nominate one that's missing.

Click that ad and you land on the same collection page. The grid of hats and polos.

The page didn't get worse between the two clicks

Nobody redesigned anything. The same page, the same week, the same collection, and it is completely right for one ad and completely wrong for the other.

The grid ad promises here is a collection of stuff. The page is a collection of stuff. Nothing to fix.

The video promises something else entirely. A decade-old community project. A map of 150 courses that real people submitted. An aesthetic built around hunting down places nobody has heard of. It says, explicitly, that the clothes are an homage to that. Someone who watches all sixty seconds and clicks is not arriving to browse outerwear. They're arriving because they got told a story and want to be part of it.

Against the hierarchy, here is what he promised and what the page said back:

Rung	What the ad promised	What the page said
1 · promise	"This collection is an homage to all of it"	The project is never mentioned
3 · claim	"We made a map of all of our favorites"	No map, no link to one
3 · claim	"150 gems, you guys submitted your favorites"	No count, no community, no nomination
4 · audience	"Affordable, accessible, interesting golf courses"	Nothing addressed to that shopper
5 · voice	"Bonus points if there's a good sandwich shop nearby"	Nothing on the page sounds like him
5 · proof	"More than a decade ago…"	No history, no reason to believe it

So mismatch isn't a property of a page. It's a property of the pair. Audit your landing pages in isolation and this is invisible by construction, because there is nothing wrong with the page.

What we changed

For the video ad's traffic only, and for nobody else's:

The hero stops being a tall photo and starts naming the thing he named: colorways that remind us of an adventure to find a secret golf course. The products come up almost immediately, because someone who watched a minute of video doesn't need another minute of preamble.

Inside the grid, a strip inventories the kit the way he inventoried it out loud: anoraks with matching shorts, vests, windshirts, leather and knit headcovers, the bags, the Nalgenes.

Below the products, the parts of his story the page was missing: where the project came from, the map, the community that filled it in, and his sign-off.

The copy is his. It's lifted from what he actually says, not written fresh in a brand voice. He already made the promise in language his customers responded to well enough to click; there's no version we'd invent that beats it.

Someone arriving from the grid ad sees none of this. They get the page as it was, which was already right for them.

What happened

Both ads ran as 50/50 split tests against the same collection page. Two weeks, about 6,300 visitors between them.

The video ad, the one with the gaps:

	Control	Rebuilt
Visitors	1,431	1,390
Conversion rate	0.42%	0.94%
Revenue per visitor	$0.52	$1.54

Conversion rate up 123%. Revenue per visitor up 197%.

The grid ad, the one that already matched:

	Control	Rebuilt
Visitors	1,742	1,770
Conversion rate	0.69%	0.62%
Revenue per visitor	$0.91	$0.88

Flat. Slightly down, inside the range you'd expect from a coin flip.

The flat one is the interesting one

If we had only run the video ad's test, the honest read would be "a fresh page lifted conversion," and you'd be right to shrug. New pages lift conversion all the time. People click on things that look different. That effect is real, it's temporary, and the industry has been selling it as a win for twenty years.

But the same rebuild engine ran on the same page for the grid ad's traffic, in the same two weeks, with the same brand and the same products. And it did nothing.

That's the whole argument. If novelty were doing the work, both would have moved. If "our AI writes better pages than yours" were doing the work, both would have moved. Only the one with something to fix moved, which means what moved it was closing the gap between what the ad promised and what the page said.

The grid ad never had a gap. There was nothing there to close.

Try it on five ads this week

Take your five highest-spend ads. For each one, open the ad and the landing page side by side.

Watch the ad all the way through with the sound on and write down every promise it makes, spoken or on screen or in the caption. Not the theme. The specific claims. A map of 150 courses. The community submitted them. More than a decade old.

Then read the page and check them off.

The ones you can't check off are the gaps. Most of the time you'll find the page keeps the caption's promise and drops everything the video said, which tracks: the caption is the part someone wrote in the same tool that made the landing page.

Do that five times and you'll know whether this is a problem in your account. It took us longer to explain than it'll take you to run.

Doing it by hand works. It stops working around ad six. Every ad checked against the page it points at, re-checked as new creative ships, is the job we built Throughline to do, for email campaigns and the pages their links land on as much as for ads. Everything ships as a split test, and nothing serves until you approve it.

If you want help setting this up, let me know.

Your AI Can Write the Copy. It Can't Build the Tribe.

2026-09-02T12:00:00Z

We went to Opticon NYC this year as an Optimizely partner, which meant two days in the room instead of a recap read afterward. This year's event is branded "Camp Opticon," with sessions sorted into tracks called The Trailhead, The Bonfire, The Outpost, and Tent Talks, like someone crossed an enterprise roadmap with a Boy Scout jamboree. Every core product got the word "Agentic" stapled to the front of it too: Agentic CMS, Agentic EXP, Agentic Commerce, Agentic Feature Management. At some point a roadmap stops being a roadmap and starts being a find-and-replace.

Underneath the camp cosplay and the buzzword bingo, though, session after session made the same real point: the busywork is over. Agents will structure your content, tag it, publish it, run your tests, flag your features, and adjust in real time. One roadmap talk promised, "agents that test, learn, and adjust in real time." Another pitched agents that handle the busywork of structuring, tagging, and publishing content so teams can focus on strategy instead. A third framed the whole shift in six words: "AI Runs the Tests. You Run the Strategy."

Sit through enough of that back-to-back, and a pattern falls out that's bigger than any one feature. Optimizely is making a bet about what a marketer's job is actually for. A few sessions later, on the same day, I sat in on a talk called "Humanity Is the Premium Layer." Its thesis, straight from the abstract: AI creates abundance. Human judgment creates value.

The busywork was never the job.

Nobody built a career in marketing because they loved tagging content or babysitting a test until it hit significance. Nobody's LinkedIn headline says "Passionate About QA-ing UTM Parameters." That was always the toll you paid to get to the actual work: understanding a customer well enough to say something true to them, at the right moment, in a way that made them trust you a little more than they did yesterday.

Optimizely's own numbers make the point better than we could. One session referenced 154,000 experiments run through their platform and asked what "winning" even looks like once AI is choosing which ones to run. Another, "The Last Click Lies," made the case that the conversion you can measure was never where the decision actually happened. It happened earlier, somewhere nobody was tracking, built out of trust that accumulated over time.

AI is about to absorb the mechanical half of the job: the structuring, the tagging, the test-running, the fortieth landing page tweak of the quarter. Good. That's the plan working, and it's the same bet we've made in our own work: strip out the grunt work between a good idea and a live experience, so the time goes to the parts that actually move revenue instead of the parts that just prove you were busy.

Table stakes isn't a strategy.

When every brand has access to the same agents, the same test velocity, the same content-at-scale, none of that is a differentiator anymore. It's table stakes. Optimizely's own speakers seemed to know this. A talk titled "The Human Is Obsolete. Long Live the Human" was built to argue itself down. "He's Just Not That Into You" made the case that over-personalization is a trust killer and restraint is what actually wins. And a Bonfire discussion just asked flatly: who are you really marketing to, humans or AI?

There's something a little unhinged about an enterprise software company hosting a debate, on its own stage, over whether enterprise software is making people obsolete. Still, you don't program a room full of talks arguing both sides of the same question unless the industry genuinely doesn't know the answer yet. The shape of it is clear enough, though. When the mechanical work gets cheap, judgment gets expensive, and so does taste, restraint, and knowing your customer well enough to leave something out.

The thing that doesn't scale

We'd push further than the sessions did. If judgment is the premium layer, its highest form is finding the actual people who are your people, the specific, non-average humans who'll buy from you five times and tell three friends, and building something around them that a competitor with the same AI stack can't copy in a weekend.

Agents can run more ad variants and tighten a bid faster than any person could. None of that gets you the customer who showed up because a friend already mentioned you in a group text. That's not a targeting problem. A referral code doesn't buy the same thing as a real recommendation, no matter how well the automation behind it runs. The brand people bring their friends to without being asked wasn't built by an agent, and it's the one acquisition channel that doesn't get more expensive every time a competitor turns on the same AI.

This pattern shows up everywhere the mechanical work is getting automated out of marketing right now, including in the tools we build. Taking manual editing off someone's plate is just the mechanism. What matters is where that freed-up time goes: better creative, a sharper point of view, an actual relationship with the people you're trying to reach.

Do something real with the time back.

Three moves, if you actually run marketing instead of just attending conferences about it:

Audit where your team's hours actually go this month: not what's on the roadmap, what people are doing right now. If a meaningful chunk is mechanical (formatting, tagging, manual QA, repetitive test setup), that's the first thing to hand off, not the last.
Put the freed-up time somewhere AI can't follow: a customer advisory group, a community you moderate yourself, a segment you actually talk to instead of just watching in a dashboard.
Get suspicious of anything that scales too easily. If a competitor with the same tools could copy it in a weekend, that's overhead you haven't automated yet, not a differentiator.

None of that needs a lanyard or a free Hydro Flask.

AI will keep getting better at the parts of this job that were always closer to administration than marketing. Good. Let it. The value was never in that work; it was in what you did with the time it freed up. Opticon didn't quite ask that question out loud, but every other session was circling it. Your AI can write the copy. It can even test forty versions of it before lunch. It still can't find your tribe for you, and it definitely can't build one.

AI Is My Second Desk

2026-09-09T00:00:00Z

In second grade, I was given two desks because I couldn't keep my things organized enough in one. I was the only kid in the class with two desks. Looking back, my parents probably should've gotten me help. Instead, the teachers just gave me two desks.

My whole life, I've been finding crutches for my inability to complete executive function tasks. Now, with AI, I've outsourced as much of my executive function as I can, because, well, I'm crappy at getting things done, but I'm pretty good at figuring out what needs to get done.

This quarter I've been leaning into every place AI can take over tasks that take me and my team longer than necessary. Here's what we're using the most.

1. A chief of staff that never sleeps

I built a set of Claude skills that run on a schedule, checking my calendar, email, Slack, and Linear so I don't have to. It preps me for meetings, flags what needs a reply, and catches the stuff that would otherwise fall through the cracks between my desks.

2. Product photography without the photoshoot

We're using FASHN to put clothing on AI-generated models instead of booking a shoot every time a client needs new product imagery. What used to be a multi-week production is now something we can turn around same day.

3. Bullet points at catalog scale

We generate PDP summary bullets across catalogs running into the thousands of SKUs, a job that used to cost a copywriter weeks per client. Now it's a pass we run and review. We're testing them at the top of the product page, and seeing real CVR gains.

4. Closing the relevance gap automatically

Throughline, the product we've been building at Mobile1st, dynamically matches landing page content to the ad that actually drove the click. Instead of every visitor landing on the same generic page regardless of which ad brought them there, the page adjusts to match.

5. Building bigger experiments in less time

Using Opal and Claude, we're able to build more complex, multi-change experiments in the time it used to take to ship one simple variant, and have the AI verify the implementation actually works before it goes to real traffic.

6. Ad creative that copies what's already working

We're testing ways to generate ad variations modeled on our own top performers, so instead of guessing at the next creative, we're iterating off what's already proven to convert. Still hunting for the right tool here. If you've found one that actually works, tell me.

7. A daily check that catches broken tests before they cost us

A scheduled check pulls our GA4 experiment data every day and flags in Slack the moment a live test's event counts drop or stop firing. Before this, a broken test could quietly burn two weeks of traffic before anyone noticed.

I got two desks in second grade because I couldn't manage one. This quarter, AI has basically become desk number three. What did I miss?

Why Doesn't Every A/B Testing Tool Tell You When Your Test Stopped Running?

2026-09-16T00:00:00Z

We didn't set out to build an A/B testing platform. Throughline CR exists to fix landing pages: pull in your ad campaigns, rebuild the page to actually match what the ad promised, let you approve the change. The testing layer was a means to an end. We needed to prove our rebuilds worked, and setting that up by hand in an existing testing tool, for every ad, every day, while the underlying pages kept changing on us, wasn't realistic. So we built our own.

Along the way we ran into a problem that had nothing to do with statistics and everything to do with plumbing. Our changes kept silently breaking.

The idea wasn't wrong. The page had moved out from under it: a theme update rotated an element's ID, a new app changed how a section rendered, a developer pushed something unrelated to a totally different part of the site. Whatever the cause, the code we'd written to find an element and change it could no longer find what it was looking for. And nothing about that showed up anywhere. The test kept running, the dashboard kept showing a clean split, and zero percent of visitors were actually seeing the variant.

Every tool assumes the variant keeps landing

Open any A/B testing platform and you'll see the same report: visitors, conversions, lift, confidence. What you won't see, anywhere, is the answer to a more basic question: of the people you intended to show this variant to, how many actually saw it?

That number is assumed to be 100%. It almost never is, and nothing tells you when it isn't.

Sites change constantly: a platform update, a new app installed, a developer's unrelated commit, a redesign of one section. Any of those can quietly sever the connection between the change you defined and the change actually rendering, and when it happens, the experiment doesn't fail loudly, it fails by going quiet. The test keeps collecting data, the data keeps looking normal, and everyone downstream (the analyst, the growth lead, the exec reading the readout) treats a null result as proof the idea didn't work. The honest answer is that nobody knows, because the test stopped serving three weeks ago.

We measured this on our own production traffic. Across a sample of live rebuilds, roughly a third of everything actually serving on real pages had lost the labels needed to tell which specific change was doing what, so our first attempt at a detector was flying blind on two-thirds of what mattered. Once we could see clearly, a separate check found that on a typical account, over half of all serving selectors were pinned to something the underlying page platform rotates on every publish. One theme republish, one afternoon, and a dozen live experiments go dark at once, with the reporting dashboard none the wiser.

That's the mechanism by which any client-side experiment can rot, on any platform, with any tool. We happened to be forced to look directly at it, because proving the change is real is the entire business.

What we built instead

We built a health monitor into the testing layer itself: for every individual change inside a test, what fraction of intended visitors actually got it.

Every change we ship gets tagged. When it runs in a real visitor's browser, it reports back one of a few outcomes: applied cleanly, applied but late, found and then overwritten by something else on the page, or never found at all. We aggregate that per change, per test, continuously. If a change that should be reaching 95%+ of eligible visitors drops below a threshold, we know before anyone has to notice a strange-looking readout. The repair loop matches that logic: a first, deterministic pass that tries to re-locate the element using something sturdier than whatever broke, and only escalates to a full regeneration of the change if that fails. Retries are capped, so a permanently broken anchor doesn't burn effort every hour forever.

The part that actually took the work was refusing to let silence read as success. It would have been easy to ship a dashboard that shows 0% when there's no data and call it finished. We had to decide, explicitly, that no signal is not the same as no problem. A test with zero visitors this week and a test serving to zero percent of the visitors it has needs to render differently on the page, or the whole exercise is pointless.

Why this isn't already standard

Most testing tools are built by people solving the statistics problem: sample size, significance, sequential testing, priors. That's genuinely hard, and it's worth solving. But it all assumes the treatment reached the person you meant to treat. Almost nobody instruments the boring layer underneath: did the DOM query still find the thing? Did the change apply before the visitor scrolled past it? Did some other script on the page overwrite it half a second later?

Testing vendors know pages change. What they haven't built is the machinery to check, for each individual change, whether it survived contact with a real page today. That's a different engineering problem than measuring whether a delta is statistically significant, and most tools were built to answer the second question.

We built it because we had no choice. If a client asks why conversion didn't move on a change we shipped, "the test ran cleanly" isn't an acceptable answer when the honest one is "the test stopped applying to anyone eleven days ago." The tool had to be able to tell on itself.

How to build this into what you're doing

You don't need a testing platform to apply the idea. It generalizes to almost any system where you deploy something and then trust, rather than verify, that it kept working.

The first question to ask is whether something running and something working are actually the same thing in your system, because most monitoring treats them as identical when they aren't. A job that didn't throw an error gets marked healthy by default. That's an inference, not a measurement, and the fix is to build an independent check for the actual outcome you care about.

Then find the seam most likely to break without telling you, which usually isn't the one that's easiest to watch. For us that's a CSS selector reaching into someone else's DOM: the single most fragile point in the whole pipeline, and one that fails silently by design, since a missing element just does nothing rather than throwing. Your equivalent might be an API field a vendor can quietly remove, a config value someone forgets to set, or a scheduled job that's been logging "success" for weeks while its input sits empty.

Whatever you build, silence needs its own visible state. Zero isn't the same as unmeasured, and no errors isn't the same as healthy. A dashboard that can't tell "nothing happened because there's nothing to report" apart from "nothing happened because it's broken" will eventually get read the wrong way, at the worst possible time.

If you can fix drift before you alert on it, do that first and page a human second. Most breakage in a system like this has a mechanical fix, and burning someone's attention on a problem the system could have solved itself is a real cost: the next alert gets a little less trust. Just cap the retries. A self-healing loop with no ceiling on attempts is only a quieter way to waste resources on something it can never actually fix.

None of this is exotic, it's the same instinct behind health checks, synthetic monitoring, and canary deployments, pointed at a narrower and less glamorous claim: that the thing you shipped is still doing what you built it to do, today, for the person actually using it. Most tools stop at confirming the deploy went through. We only built ours because we couldn't afford the gap between that and the truth.

Scrambled Tofu Is the Longest Experiment I've Ever Run

2026-09-23T12:00:00Z

Random photo from 2024, so pardon the mess on the table.

For seven years, my wife and I have eaten scrambled tofu for breakfast, Monday through Thursday. That's somewhere around 1,300 bowls.

If you haven't had it: scrambled tofu is the vegan answer to scrambled eggs, crumbled and sautéed with spices to land somewhere close to the real thing. My wife's vegan, and we like sharing breakfast and coffee before the day splits us up: work for me, school drop-off for the kids now. Tofu turned out to be the breakfast that actually stuck. Everything we tried before it was either not filling enough or not good enough to eat five days a week for years running.

What's kept it interesting is that there's no single way to make it. Over those seven years we've changed jobs, had two kids, bought a house, renovated it, lived in a tiny apartment for a stretch, and moved into a new place. The tofu changed right along with all of it.

We started simple: nutritional yeast, salt, sautéed on the stove. That worked for a while. Then we'd see a recipe online and bolt something on, paprika one week, garlic the next, and run with it for a few days to see if it stuck.

After watching Salt Fat Acid Heat, we decided the tofu needed more acid, so we tried za'atar. That one stayed. Then we wanted more of an eggy flavor, so we added black salt, the sulfurous stuff that actually smells like eggs. That worked too, until keeping a jar of it stocked every week turned out to be more effort than it was worth, and we dropped it.

Then our second kid showed up, and we needed the whole morning to run hands-off. So we started cooking the tofu the night before and reheating it in the toaster oven. That held for a while, but we didn't want a second pan dirty every night, so we switched to baking it from raw in the toaster oven instead of sautéing it at all. That's roughly where we've landed for now.

Every one of those changes got a real trial. And every one eventually got knocked loose by something outside the recipe: a new kid, a new kitchen, a new schedule. We were chasing the best tofu for right now. Right now kept moving.

The same rules apply to revenue

The basis is the same whether you're making breakfast or moving revenue for a business: stay open to trying something new, and keep looking for where you can improve. Experiments come in a lot of shapes. Sometimes it's an A/B test, sometimes it's just making a change and reading the tea leaves, but the work behind it is the same either way: talk to customers, look at the business intelligence you've got, form a hypothesis, and make a change big enough that people actually notice it.

A few things held up on both the stove and the website:

You need more than one try to know if you love or hate something. A single morning tells you almost nothing. Run it for a week or two before you trust your own reaction.
Make the change big enough to notice. A pinch more paprika won't tell you anything. Add enough that you can taste the difference, then dial it back if it's too much.
Don't be afraid to change more than one thing at once. The goal isn't to isolate which variable moved the needle, it's to make the best tofu, or the best site. Change things together, read what happened, and decide what to touch next. You can always pare back the changes later if you need to know which one mattered.
Use inspiration, don't copy it outright. Their za'atar ratio, their homepage, their onboarding flow: none of it was built for your kitchen or your customers. It's a fine starting point. It's never the finish line.

Seven years in, we're still not done, and we don't expect to be. The tofu we make tonight is built for the four of us, in this kitchen, this month. In another year, something will change, and we'll change the tofu with it.

The (current) Aronstein tofu recipe

For speed and convenience on a weekday morning.

Ingredients:

1 block of tofu, lightly pressed and hand-torn
Oil (don't be shy)
Nutritional yeast, eyeballed until every piece is coated
A large splash of paprika
A half splash of za'atar
A half splash of turmeric
Optional: sugar

Steps:

Tear the tofu onto a baking sheet lined with parchment (we reuse the same sheet of parchment the next morning)
Pour the oil over the tofu and mix
In a separate bowl, mix the dry ingredients
Pour the dry mix over the tofu and mix again
Bake at 400°F for 29 minutes
The next morning, reheat at 300°F for 10 minutes while you make coffee

Your 20% Off Ad Lands on a Page Where Two Products Are 20% Off

2026-09-30T12:00:00Z

A shopper sees an ad for up to 20% off appliances. She's been meaning to replace the dishwasher, so she taps it. The page loads: the appliance category, 48 products in a grid, sorted the way the merchandising team sorted it last quarter. The first row is full price. So is the second. She scrolls a little, sees nothing marked down, decides the ad was bait, and leaves.

The ad was true. There were qualifying products on that page. Two of them, in row six.

When I was at LivingDirect, before Build.com bought us, we fought this constantly. The design team would make a great sale banner for a real promotion, and the banner would point at a page where only some of the products applied. Keeping the two in sync was close to impossible. The sale changed on marketing's calendar, the grid got re-sorted on the site team's, and the page drifted away from what the banner said.

Nobody was careless. The creative team owns the banner, the site team owns the page, and merchandising owns the sort order. Each group did good work, and the shopper still landed on a page that made the ad look dishonest. I've since seen the same thing on retailer sites, distributor sites, and DTC brands with fifteen products. The bigger the catalog, the worse it gets.

"Up to" is a promise about the first screen

Retailers run compliance checks on sale claims. Is there a minimum number of qualifying products? Is the discount real? Those checks are good. None of them ask where the qualifying products appear on the page.

"Up to 20% off" is technically true if one product on page three qualifies. For the shopper it's only true if she sees it in the first few seconds, because she clicked expecting the thing you told her about. When the first screen shows full-price dishwashers, she assumes you misled her. She doesn't go looking for row six. You paid for that click and the page wasted it.

Why nobody fixes it

The category page serves more than the sale ad. It gets organic search traffic, email traffic, and returning customers who never saw the promotion. Re-sorting it for the sale changes it for all of them, and the merchant who owns the page has good reasons to say no.

Timing makes it worse. Promotions are planned weeks ahead and change often, while category pages get updated on the site team's schedule. By the time someone files a ticket to move the sale products up, the sale is half over.

And the gap between the banner and the page belongs to no one. The banner team thinks the page is the site team's job, the site team thinks the promotion is marketing's job, and it stays broken.

Let the page rearrange itself for the click

The fix is letting the page respond to where the visitor came from. A better ad won't do it, and neither will a new landing page for every promotion.

When someone clicks the 20% off ad, the products that are 20% off belong in the first row, badged so she can tell at a glance they match what she clicked. The headline should say what the ad said. The rest of the page stays as it is, and anyone who didn't click that ad sees the page exactly the way the merchant built it. The merchant keeps their page, and the site team never gets a ticket.

Why the first screen matters this much

We looked at a year of Meta product-extension ads for three of our clients. These are the ads that show a row of products under the main image, so a shopper can tap a specific product instead of the ad itself. The two paths land on different pages. Tapping a product lands on that product. Tapping the ad lands wherever the ad points.

For a golf apparel brand, shoppers who tapped a product converted at 5.8%. Shoppers who tapped the ad converted at 2.1%. For a men's apparel brand the gap was wider: 14.2% against 2.3%. Order values were the same on both paths, so the difference was entirely whether people bought. The third client sells one hero product, and there the product row didn't help, which makes sense: when there's only one thing to buy, every page already shows it.

Same ads, same audiences, same days. The shoppers who landed on exactly what they tapped bought roughly three to six times as often. That's the size of the gap a sale ad opens when its first screen shows the wrong products.

How Throughline closes it

We built Throughline for this problem. It starts by reading the ad the way a shopper would: the headline, the body copy, the text on the image, and what's said in the video. From that it pulls out what the ad promised, such as the discount, the products it showed, and who it was talking to.

Then it reads the landing page and scores how well the page delivers each of those promises. A sale ad that lands on a full-price first row scores low on the offer. The changes it proposes go after the promises the page misses and leave the rest alone. On a product grid that usually means moving the products the ad featured or discounted into the first row, badging them so they match the ad, and putting the ad's own headline above the grid. The copy comes from the ad. Throughline doesn't write new claims, so a merchant never finds a promise on their page that marketing didn't already approve.

A person on your team approves every change before it goes live. Once it's live, the change reaches only visitors whose click carries that ad's ID. Someone arriving from search, email, or a bookmark sees the page the merchant built. Nothing in your site's code changes, so there's no ticket for the site team. A single line of script on the page asks our edge server whether this visitor came from an ad with an approved change, and applies it if so.

Pages change underneath you, which was the part we could never keep up with at LivingDirect. Throughline rechecks every live change against the real page every hour. When the site team re-sorts the grid or ships a new layout, the change is repaired, or flagged for review if it can't be.

When Throughline mends the gaps between the ad and the landing page, we increase conversion rate and RPV by over 25%. Reach out to see what Throughline CR can do for you.

40% of What's in Your Email Is Missing From Your Landing Page

2026-10-14T12:00:00Z

Pull up the last email your team sent. Read it. Then click the button and read the page it takes you to.

Now count what the email said that the page doesn't. The offer. The product angle. The person the email was written for. The reason to buy this week instead of next month.

We went through our clients' emails and the pages they link to, claim by claim. On average, about 40% of what the email said was nowhere on the landing page. Some pages were missing 20%. Some were missing 70%. The customer clicked because of a specific promise and landed somewhere that had never heard of it.

Why it goes missing

Nobody did anything wrong here. The two teams just work on different clocks.

Your email team writes a new story every week. Monday it's a gift guide for dads, Thursday it's a restock on the product that sold out, and next week it's a pitch to people who bought once and never came back. Each email is written for a specific person with a specific reason to click.

The landing page was built once, for everyone, and it's been sitting there ever since. Updating it means a ticket, a designer, a developer, and a sprint. That's fine for a homepage redesign and useless for an email that goes out Thursday and is old news by Monday.

So the email keeps moving and the page stays put.

What it looks like

An auto parts brand sent an email titled "Is Your Transmission Cooler Trying to Tell You Something?" The pitch was simple: know the warning signs and replace your cooler before it leaves you on the side of the road. The email also promised a lifetime warranty with accident protection on every product, and 0% APR financing through PayPal or Klarna.

The link went to the transmission cooler category page. The page had the right products and none of the rest. There was nothing about warning signs or breakdowns, no warranty, and no financing. The headline just said "Transmission & Oil Coolers."

Someone who clicked because they were worried about a breakdown, and who wanted to know they could spread the cost out, landed on a product grid and had to take it on faith.

The fix

The page has to finish the story the email started. It doesn't take a new page, just a few additions for the people who came from that email:

List every claim in the email: the headline promise, the offer, the product benefits, who it's talking to, the deadline.
Check which ones the page already makes. Usually a handful are there.
Add the missing ones, only for visitors from that email. A few lines of JavaScript keyed off the email's UTM tag does it:

const params = new URLSearchParams(location.search);
if (params.get('utm_campaign') === 'transmission-cooler-warning') {
  const title = document.querySelector('.category__title');
  title.insertAdjacentHTML('afterend', `
    <div class="email-match">
      <strong>Don't wait for the breakdown.</strong>
      Every cooler comes with a lifetime warranty and accident protection.
      0% APR financing with PayPal or Klarna at checkout.
    </div>`);
}

Someone arriving from Google sees the page as it was. Someone arriving from the email sees the page plus the things the email promised them.

Does it work?

Early results say yes. We've been running this same approach on paid social traffic, where the ad plays the role of the email. For one brand, matched pages converted 22% better than the originals over two and a half months. The ad or email already did the persuading. The page just needs to reassure the visitor they're in the right place.

The catch

For one email this takes an afternoon. Most teams send several emails a week, and each one has four or five links going to different pages. That's dozens of pages a month, every one needing its claims listed, gaps found, code written, tested, and switched off when the campaign ends. Nobody on the email team has that kind of time, and the dev team has a backlog of their own.

So we built a tool for it

Throughline does this for every email you send. Forward the email to your Throughline address. It pulls out every link, reads what the email promises, compares that to each landing page, and builds the matched version. Your team reviews it and approves it. From then on, email clicks see the matched page, everyone else sees the original, and a split test runs automatically so you can see the lift. Every link in one email puts visitors in the same test group, so the numbers stay clean.

Forward us your last campaign email. We'll send back a list of what your landing page is missing.

Your Black Friday Ad Is Ready. The Page It Opens Is From September.

2026-10-21T12:00:00Z

Some night in the next four weeks you're going to approve a BOGO ad at 11pm, push it live, tap it on your phone, and land on a collection page still selling fall layers. The ad says buy one, get one. The page doesn't mention it. The shopper scrolls, goes looking for the deal, can't find it, and leaves.

You'll know because you'll have done the tap. Then you'll open a ticket.

That ticket is the whole problem. Demand gen can launch forty ad variants in an afternoon. The page they point at changes when the web team gets to it, and in November the web team is in code freeze, because nobody wants a developer touching the site during the week that pays for the year. So the ads move, the page stays put, and every new launch widens the gap.

Everyone in demand gen knows this. Most of us have just decided to live with it.

The usual fixes run on the wrong clock

You can build a landing page per ad. Every page builder on earth will tell you it takes minutes. Sure, minutes per page, times forty ads, times the creative refresh you're doing the Monday after Thanksgiving, all built by the one person who knows the builder and is also running the ads.

The ticket and the builder both move at the speed of a project, and your ads move whenever someone on your team has a good idea at 4pm. The page has to follow the ad by itself or it won't happen.

What we built

Throughline watches your ad accounts. When a new ad starts spending, it reads the ad (copy, image, video), reads the page the ad sends people to, and rebuilds that page so it says what the ad said, using the site's own fonts, colors and layout.

It edits the page you already have, at the URL you already have. The destination link in Ads Manager stays exactly where it is, and so does every link in your emails. A line of code on your site swaps in the new version for the people who clicked that ad, and everyone else sees the page as it was. Same with email: forward us the campaign, and the pages it links to get rebuilt for whoever clicks from it.

You get a link, look at it on your phone, and approve it or toss it. Then you choose: split traffic 50/50 to prove it works, or send everyone to the new page.

Nobody has to tell us a new ad exists. Emma, who runs e-commerce at a DTC apparel brand we work with, asked on a call whether she should be letting us know when new ads go live. She doesn't need to.

"This is the easiest collab ever," she told us. "You guys just send in ideas, and we're like, that looks good."

What she checks for is whether it looks like her brand. "The style seems to be very consistent with the site," she said. "All those updates just naturally are blending in well."

The guy who didn't buy it

Jonah, her partner on the e-commerce side, was less charmed, and he had a point. When a SaaS product "claims a mondo benefit, especially when it feels like it's relatively light touch," he said, "it just tends to not feel real, and often in my experience, it often isn't real."

I've bought those products too. I have the annual contracts to prove it.

He also called out our early work, which was "literally almost lifting a snippet of text from the ad and throwing it in the subhead." Yep. That was version one. The pages do a lot more now. On one of their collection pages, the rugby shirt from the ad sits at the top and comes back again at the bottom.

Revenue per visitor on their rebuilt pages is up 45% over the originals. Which is exactly the kind of number Jonah just told you not to trust, so he gets the next word too.

Jonah's theory for why the small stuff works: "Simple can be really good, and not f***ing with things too much, but making slight tweaks, I think, is where a lot of the margin is."

What they're doing with Black Friday

Emma's plan for their BOGO sale is to skip the 50/50. "Instead of doing a 50/50 A/B test, maybe lean more on what Throughline can update for us…" she said, "especially that sale."

That's the right call. Once the rebuilt page has beaten the original, splitting traffic during the biggest week of the year means sending half of your most expensive clicks to the worse page on purpose.

Jonah went one further. Around planned events, he wants the post-click experience worked out "before the fact," so the page is ready the moment the ads go live. Your creative calendar for the next four weeks is already written. The pages those ads land on are the part nobody has scheduled.

Go tap your own ads

Tonight, pull up every Black Friday ad you've got queued and tap through each one on your phone. If every page tells the shopper the same thing the ad did, you don't need us, and I'd like to know how you pulled it off.

If you find one that doesn't, send me the screenshot. I'll show you what the page should have said.

How to Write Ad Creative Meta Will Actually Show

2026-10-28T12:00:00Z

One of our clients loaded 20 versions of primary text into a single Meta ad. Over the next three days the top version got 61% of the impressions and the second got 24%. The other 18 split what was left, and 15 of them got less than 1% each.

Somebody wrote those 15. Probably in a doc with a comment thread about tone. Meta showed them to almost nobody.

It's the usual pattern. We read the asset-level delivery for every dynamic and flexible-format ad our clients run, because our product needs to know which version of an ad a person actually saw. On ads with more than one image, the most-shown image gets a median of 80% of impressions. On a typical ad with several headlines, more than half the headlines get under 5%. One ad had ten images: two of them took 91% and the other eight shared 9%. Another had six images across 36,000 impressions, and one image got 80% of them.

How Meta picks

Every time someone opens Facebook or Instagram, Meta runs an auction for that ad slot. It estimates how likely this specific person is to do the thing you're optimizing for, and it makes that estimate for each version of your ad it could show. Then it shows the version it's most confident about.

The estimate comes from what it has already seen. A version that got shown early and picked up a few clicks has data behind it. A version that barely ran has almost none, so the system has little reason to believe in it and keeps passing it over. The early leader gets more impressions, which gets it more data, which makes the system more sure of it. That loop is how you end up with the numbers above.

It's a sensible way to spend your budget if the goal is conversions this week. As a way to learn anything, it's garbage. Meta is optimizing for its own prediction, and it has zero interest in giving your sixth headline a fair trial.

No campaign structure or toggle gives you a vote in which version shows. You decide what goes on the shelf, and it decides what gets picked off it, per person, for its own reasons. So the work is all in what you put on the shelf.

How to find and replace the versions Meta isn't showing

Find out what actually ran. Open your highest-spend flexible or dynamic ad and pull delivery by asset: it's in Ads Manager's breakdown menu, and in the API as image_asset, title_asset and body_asset. Use at least a week of data and sort each list by impressions.
Draw a line at 5%. Any version under it never really played. Ignore its CTR and ROAS, because a few hundred impressions can't tell you anything, and you've learned nothing about whether it was any good.
Read the starved versions next to the winner. Usually they make the same promise in different words. That's why they lost: Meta picks the version it thinks fits each person, and five rewrites of one promise give it nothing different to match against a different buyer. (Here's how to tell if your 100 ads are really one ad.)
Replace them with a different reason to buy. Write each new version around a different problem the product solves or a different offer, so it has a shot at a buyer the winner doesn't reach. Leave the winner alone. Editing it can send the ad back through Meta's learning phase, and it's the version doing the work.
Re-read after a week. A new version still under 5% is Meta's answer for now. Swap it for the next idea instead of rewording it.
When you need an actual answer, take it out of the pile. If you want to know which of two angles sells better, put them in separate ads and run Meta's A/B test tool, which forces an even split. Inside one flexible ad you'll only ever learn which one Meta preferred.
Brief and report on the version that's running. If one image gets 80% of the impressions, that image is your ad. It's the one your landing page should match and the one your report should show, even if it wasn't your favorite in the review meeting.

Our part in this

We started pulling asset delivery because our pages need to match what a visitor actually saw. "The ad" turned out to be a stack of versions where one did nearly all the talking, and we'd been treating every version as if it ran.

A caveat on our numbers: each read covers a recent window of three to fourteen days for a live ad. It's a snapshot of what delivery was doing that week, and a starved version might get its turn later.

Go open your highest-spend flexible ad and pull the asset breakdown. If your impressions are spread evenly across your versions, send it to me. I haven't seen one yet.