What We Learned Auditing 1,000 Small-Business Homepages

When you audit one website, you learn about that website. When you audit a thousand, the individual sites blur and something else comes into focus: the patterns. GazeSite has now run its six-reviewer analysis, the same procedure every time, one URL, screenshot and HTML, desktop and mobile, across enough small-business homepages that I’ve stopped being surprised by almost anything. What follows isn’t a statistical study and I won’t pretend it is; I haven’t tallied percentages and I distrust people who conjure them. It’s something different: the shapes that repeat, and my best guess at why.

The first pattern is that homepages are written from the inside. The single most common clarity finding, across industries, across quality levels, is a headline that assumes you already know what the company does. “Solutions that move you forward.” “Excellence, delivered.” The founder reads this and nods, because the founder supplies the missing context automatically. The stranger reads it and learns nothing. This isn’t a writing-skill problem; many of these sites contain, three scrolls down, a perfectly plain sentence explaining the business. The problem is that the plain sentence embarrassed someone. It felt too simple to lead with. The pattern taught me that vagueness on homepages is usually not incompetence. It’s a status move, the verbal equivalent of a lobby with marble and no signage, and it costs these businesses their most valuable seconds.

The second pattern is that mobile is where sites go to break quietly. Nearly every serious layout problem we find appears on the phone-sized capture and not the desktop one. This makes sense once you picture how sites get built and reviewed: on desktops, by people at desks, showing other people at desks. The desktop rendering is inspected constantly. The mobile rendering is inspected at launch and then, in many cases, apparently never again, while content edits accumulate that were only ever previewed wide. That’s precisely why GazeSite refuses to audit a page without capturing both viewports. If we looked only at desktop, we’d be reviewing the version of the site its owner sees, which is the version least likely to be broken.

The third pattern concerns accessibility, and it upended a prejudice I didn’t know I had. I expected accessibility problems to correlate with generally sloppy sites. They don’t, particularly. Some of the most polished pages we audit, clearly expensive, clearly designed with care, are among the worst offenders: gray-on-white text chosen for elegance that fails contrast thresholds, icon buttons with no accessible name, custom controls a keyboard can’t reach. Meanwhile some plain, dated-looking sites sail through, because plain HTML, real links, real buttons, real labels, is accessible almost by default. The pattern here is that accessibility failures are largely a byproduct of sophistication. The further a site strays from the browser’s native elements, the more work it takes to get back what the browser gave away for free, and most projects never budget for that work because nobody on the team experiences the problem.

The fourth pattern is the graveyard of good intentions in the HTML. Empty meta descriptions with placeholder text still in them. Analytics snippets for tools the business stopped using years ago. Two copies of the same tracking script. Title tags that still say the name of the template. None of these is individually damning, but together they tell a story that repeats constantly: the site was built in a burst of attention, usually by an outside party, and then attention left. There is no one whose job is the website. The technical reviewer’s findings, expired certificates approaching, redirect chains, pages loading resources over insecure connections, are usually just this same story told in a different dialect. The sites aren’t badly made. They’re unmaintained, which over enough time becomes indistinguishable.

The fifth pattern is the one I find genuinely moving, and it’s why I keep doing this. The severity of a site’s problems has almost no relationship to the quality of the business behind it. We audit sites for businesses with decades of history and obvious craft, whose homepages bury the phone number, hide behind stock photography, and fail on the devices their own customers hold. The web rewards a skill, self-presentation in a browser, that has nothing to do with the skills that make a business good. An audit can’t close that gap by itself. But it can convert the gap from a vague sense that “we should redo the website” into a finite, ordered list: this finding, this severity, this fix. Finite lists get done. Vague senses don’t.

If I compress a thousand homepages into one sentence of advice, it’s this: your website’s problems are probably not the ones you’d guess, because the ones you’d guess are the ones you’d have fixed. The expired banner you’ve scrolled past two hundred times, the mobile layout you never open, the label only a screen reader notices, these survive precisely because they’re invisible from where you sit. The whole value of an audit, human or automated, is that it sits somewhere else. What we learned from a thousand homepages is mostly that: every site owner has blind spots, they’re remarkably consistent across businesses, and they are, almost without exception, cheaper to fix than to keep.