One URL, Six Perspectives: How We Split the Analysis

When you ask one person to review a web page for “everything,” you get a review of nothing in particular. I learned this the slow way, by doing it. Early on, before GazeSite was a product, I audited sites by hand, and my first instinct was to open the page and write down everything wrong with it. The reviews came back lopsided every time. Whatever I noticed first colored the whole pass. If the first thing I saw was a confusing headline, I’d produce a review that was mostly about messaging, and I’d walk right past the form whose labels a screen reader couldn’t find.

This isn’t a flaw in any particular reviewer. It’s how attention works. Looking for something and looking at something are different acts. Radiologists reading a scan for tumors can miss an obvious fracture, not because they’re careless but because search is directed: you find what you’re looking for, at the cost of what you aren’t. The fix isn’t to concentrate harder. The fix is to make multiple passes, each with a single question in mind.

So GazeSite takes one URL and reviews it six times, from six deliberately narrow perspectives: accessibility, clarity, conversion, readability, SEO, and technical. Each reviewer sees the same evidence, the rendered page and the HTML behind it, captured on both a desktop and a mobile viewport, but each is only allowed to care about one thing. The accessibility pass asks whether someone using a keyboard or a screen reader can actually operate the page. The clarity pass asks whether a stranger can tell, within seconds, what this business does and for whom. Conversion asks whether the page makes the next step obvious and easy. Readability asks whether the words themselves are doing work or getting in the way. SEO asks what a search engine can understand about the page from its titles, descriptions, and structure. Technical asks whether the plumbing, the things a visitor never sees directly but always feels, is sound.

The boundaries matter more than they look. Consider a single line of text in light gray on a white background. The readability reviewer flags it because the sentence is hard to read. The accessibility reviewer flags it because the contrast ratio fails the WCAG threshold, which is a different claim, checkable with arithmetic, with different stakes. The conversion reviewer might flag it too, if that gray text happens to be the price. Same pixels, three findings, three distinct remedies with three distinct justifications. A generalist would have written “text is hard to see” and moved on, and you’d have fixed one-third of the problem.

There’s a second benefit to the split that I didn’t anticipate: it makes disagreement visible. The six perspectives don’t always want the same thing. SEO wants text that machines can parse; clarity sometimes wants fewer words than SEO would prefer. Conversion wants the call to action loud; readability wants the page calm. When one reviewer covers everything, these tensions get resolved silently, inside one head, according to whatever that reviewer happens to weight. When six reviewers report separately, the tension shows up in the report as two findings pointing in different directions, and the decision goes where it belongs: to you, the person who knows what the business is actually trying to do. A good audit shouldn’t hide trade-offs. It should surface them.

The obvious question is why six, and not four, or twelve. The honest answer is that six is where the categories stopped overlapping and stopped leaving gaps, at least for the kind of small-business pages GazeSite mostly looks at. Every finding I’d ever written by hand fit naturally into exactly one of the six. When a category is too broad, findings inside it compete for attention and the rare ones get crowded out. When categories are too narrow, you get bureaucracy: three reviewers filing the same finding under different names. Six lenses cover the page the way a good test suite covers a program, with intent, not with volume.

I should say plainly that splitting the analysis costs something. Six passes are more expensive than one, and the findings need to be assembled back into a single report ordered by severity, because no reader wants six separate documents. But this is the same bargain every field makes when it specializes. You don’t want your general practitioner reading your MRI. You want the specialist to read it and the generalist to tell you what it means for your life. In GazeSite, the six reviewers are the specialists and the report is the generalist: each finding arrives with a score, a headline, the concrete issue, and the prescribed fix, sorted so the thing most worth doing is at the top.

The deeper principle, and the reason I’d argue for this structure even to someone building a completely different kind of review tool, is that thoroughness is not a personality trait. It’s an architecture. You don’t get complete coverage by asking a reviewer to be more careful. You get it by designing the process so that every important question gets asked on purpose, exactly once, by someone whose only job is to ask it.