The Landing Page Teardown: From 1% to 3%

When founders hear that some landing pages convert at three percent and theirs converts at one, they assume the difference is a secret. A better headline formula, some psychological trigger, a color. So they go hunting for the one big trick. But when I actually tear down underperforming pages, I almost never find one big problem. I find eight small ones. And the arithmetic of small problems is brutal, because they multiply. If five separate points on your page each silently turn away twenty percent of the people who reach them, together they’ve cut your conversions to about a third of what they could be. That’s the whole gap between one percent and three. No secrets, just leaks.

This is actually good news, because leaks can be found methodically. A teardown isn’t inspiration; it’s a walk-through. You go through the page in the order a stranger experiences it, and at each stage you ask one question: what fraction of reasonable people would stop here, and why? Let me walk through the stages, because they’re the same on nearly every page.

The first stage is the first second. Before anyone reads your headline, they take in the gestalt: does this look finished, current, and legitimate? A layout that shifts around while loading, a stretched logo, text overlapping an image on a phone. None of these are arguments against your product, but they’re arguments against your competence, and visitors treat the page as a sample of the product. This stage is invisible to owners because they’ve seen the page hundreds of times and their eyes correct for it. It’s the main reason I built GazeSite around screenshots rather than just code analysis: you have to see the page as a stranger’s browser renders it, on desktop and on a phone, because that rendering is what’s actually leaking.

The second stage is the headline, and the test is simple: could someone read only the headline and correctly tell a friend what you sell? Most headlines fail this. They describe a feeling, an outcome, a mission. “Work smarter, together.” A visitor who can’t categorize you can’t want you, and categorization has to happen before persuasion. The fix is usually to demote the clever line to a subtitle and promote a plain statement of what the thing is.

The third stage is the middle of the page, and its failure mode is claims without evidence. Every assertion a visitor can’t verify gets discounted, and a page made entirely of unverifiable assertions gets discounted entirely. Evidence here means specifics: a screenshot of the actual product rather than an illustration, a concrete description of what happens in the first five minutes of use, numbers that could conceivably be checked. The middle of the page is where you stop telling and start showing, and pages that never make that transition bleed the skeptics, who are most visitors.

The fourth stage is the ask itself. Look at your button and the three inches around it. Does the label say what happens next, or does it say something abstract like “Get Started”? Is the price anywhere nearby, or does the visitor have to click to find out what they’re agreeing to? Is there a sentence handling the obvious fear, like whether a card is required? Ambiguity at the moment of commitment is the most expensive ambiguity on the page, because it strikes exactly the people who were about to convert.

The fifth stage is the form, if there is one. Every field is a toll. Some tolls are worth charging, but most forms charge tolls out of habit: a phone number that will never be called, a company-size dropdown that exists for a dashboard someone stopped reading a year ago. Cut a form to the fields you will actually use this week and the leak narrows.

Now, the discipline that makes a teardown honest: you have to rank the leaks before fixing anything, because effort is finite and the stages don’t leak equally. A broken mobile layout upstream of everything else matters more than a suboptimal button label at the bottom, since everyone passes through the top and only survivors reach the bottom. Fix in page order, top to bottom, upstream before downstream. And fix with humility about measurement. Most small sites don’t have enough traffic to A/B test their way through eight changes; at a few hundred visitors a week, the statistics will tell you nothing for months. That’s fine. When a page has obvious defects, you don’t need a controlled experiment to justify repairing them, any more than you need one to fix a leaking pipe. Test the big rewrites if you ever get the traffic. Fix the defects now.

The uncomfortable part of all this is doing it to your own page, because familiarity makes you blind to exactly the things a stranger trips on. That’s why teardowns of other people’s pages are easy and fun, and teardowns of your own are neither. Borrow eyes wherever you can get them: a friend outside your industry, a stranger, an automated audit. What they lack in context is precisely what makes them useful. Going from one percent to three isn’t a leap. It’s eight patched leaks, found in order, by someone willing to look.