The Most Common Issue We Flag (and It's Not What You Think)

When I started building GazeSite, I assumed the reports would be full of technical problems. Missing alt text, slow scripts, broken meta tags. Those show up, certainly. But the finding that appears more often than any other is something much more embarrassing: the page never says what the site does.

I don’t mean this loosely. I mean that a reviewer looks at the screenshot of the homepage, reads the visible text, and cannot answer the question “what would I get if I gave these people money?” The headline says something like “Empowering Tomorrow’s Solutions.” The subheadline says “We help you do more.” There’s a button that says “Get Started,” which is a strange thing to say to someone who doesn’t know what they’d be starting.

The way GazeSite works makes this failure very visible. You give it one URL. It loads the page in a real browser, takes a screenshot on desktop and on mobile, saves the HTML, and then six reviewers each grade the page from one angle: accessibility, clarity, conversion, readability, SEO, and technical health. Each finding comes back with a score, a headline, a description of the issue, and a prescribed fix. The clarity reviewer’s job is essentially to be a stranger. And strangers, it turns out, are confused by most websites.

Why does this happen? I think it’s a curse-of-knowledge problem, and it’s nearly universal. If you spend all day inside a business, the thing the business does becomes so obvious that stating it feels beneath you. Saying “we sell accounting software for restaurants” feels crude when you’ve spent two years thinking about the nuances of your product. So you reach for something that captures the nuance, and you end up with a sentence that captures nothing. The founder reads the vague headline and mentally fills in everything it leaves out. The visitor reads the same headline and fills in nothing, because they have nothing to fill it in with.

The reason I find this failure interesting is that it’s invisible to every tool that came before. A performance scanner will tell you your page loads in under a second. An HTML validator will tell you your markup is clean. An SEO crawler will tell you your title tag is the right length. All of these can be true of a page that communicates nothing. The page is technically perfect and practically mute. You need something that actually reads the page the way a person would, and until recently software couldn’t do that.

There’s a simple test I’d suggest even if you never run an audit. Show your homepage to someone who has never heard of your company, for ten seconds, then take it away and ask them what you sell. Not what impression they got. What you sell, concretely, in a sentence. Most sites fail this test. The fix is usually not a redesign. It’s rewriting one headline so that it names the customer and the thing: who it’s for, and what it does for them. “Accounting software for restaurants” beats “Empowering Tomorrow’s Solutions” every single time, and it’s shorter.

What surprised me more is who fails. You’d expect small businesses with homemade sites to have this problem, and they do, but they fail differently. The plumber’s site says “Plumbing Services, Call Now,” which is unglamorous but perfectly clear. It’s the funded startups and the mid-sized companies with design budgets that produce the fog. There seems to be an inverse relationship between how much a company spent on its website and how quickly the website says what the company does. Money buys polish, and polish, unattended, tends to smooth away meaning. The words that survive a committee are the words nobody objected to, and nobody objects to words that don’t say anything.

This is also why the mobile screenshot matters. A headline that works on desktop often gets truncated or pushed below a cookie banner and an app-install prompt on a phone. So even sites that pass the ten-second test on a laptop can fail it on the device most of their visitors actually use. When GazeSite grades a page it looks at both, because “clear on the designer’s monitor” and “clear on a phone in sunlight” are different properties, and only the second one pays the bills.

I’ll admit there’s something funny about building software with screenshot capture, network traces, and a panel of specialized reviewers, and having its most frequent output be “please say what you do.” But I’ve come to think that’s exactly right. The most valuable thing an outside reviewer gives you is not expertise you lack. It’s distance you can’t have. You cannot un-know your own business, so you cannot read your own homepage the way a stranger does. A tool can. And the most common thing the stranger says, looking at page after page after page, is: I still don’t know what this is.

If you fix only one thing on your website this month, fix that. Everything else on the report is downstream of it. A fast, accessible, well-structured page that doesn’t say what you sell is a beautifully engineered empty room.