How to Test Clarity With Five Strangers and No Budget

The most expensive question in web design has a nearly free answer. The question is: does anyone understand what this site is about? The answer costs you five strangers and an afternoon.

I say strangers deliberately. Your friends won’t work. Your cofounder definitely won’t work. Anyone who has heard you explain the product even once is contaminated. They’ll read your headline, and instead of processing the words on the screen, their brain will helpfully substitute the explanation they already have. This is why founders are the worst judges of their own homepages. It’s not that they lack taste. It’s that they physically cannot see the page the way a first-time visitor sees it, any more than you can hear your own accent.

So you need people who owe you nothing and know nothing. Where do you find them? Anywhere people are mildly bored. A coffee shop. A coworking space you don’t belong to. A relative’s coworker. The person next to you on a train. You don’t need a usability lab or a recruiting panel. You need someone willing to look at a screen for thirty seconds.

The test itself is almost stupidly simple. Show them your homepage for five seconds. Then take it away and ask three questions. What does this company do? Who is it for? What were you supposed to do next? That’s the whole test. No eye tracking, no think-aloud protocol, no consent forms. Five seconds, three questions.

Why five seconds? Because that’s roughly the budget a real visitor gives you. Someone who clicks through from a search result or a link a friend sent isn’t going to study your page. They glance, they form a guess, and the guess decides whether they stay. If your page requires careful reading to be understood, it will not be understood, because nobody reads carefully on the first visit.

The answers you get will hurt, and that’s the point. When I first ran this test on a project of mine years ago, the answers to “what does this company do?” included two shrugs, one “some kind of software?”, and one confident answer that was wrong. Not vague. Wrong. The person had latched onto a word in my headline and constructed an entirely different product from it. I had written “insights” when I meant “reports,” and she thought we were a consulting firm.

That’s the second thing the test teaches you: the failures are specific. You don’t just learn that the page is confusing. You learn which word did the damage. When four out of five people misread the same phrase, you don’t need a statistician to tell you the phrase is broken. This is why five people is enough. You’re not measuring a conversion rate, where small effects need large samples. You’re hunting for comprehension failures, and comprehension failures are big and repeatable. If your headline confuses strangers, it confuses most of them, and five is plenty to see it.

There are a few ways to ruin the test, and they’re all forms of cheating. The most common is talking. The moment your stranger hesitates, you will feel an overwhelming urge to help. “Well, it’s basically a tool that…” Stop. Every word you say is a word your website failed to say. The hesitation is the data. Sit on your hands and let the silence be awkward. The second way to cheat is to show more than the first screen. Real visitors decide before they scroll, so your test should too. The third is to explain the product before the test “just for context.” Then you’ve recruited another contaminated friend, and you’ve wasted a perfectly good stranger.

What do you do with the results? Rewrite and retest. This is the part people skip, and it’s where the compounding happens. The first round tells you your headline is fog. You rewrite it in plain words, find five more strangers, and run it again. Usually the second round produces a new, smaller failure — the headline lands but the button label confuses, or people understand what you do but not who it’s for. Each round costs an afternoon and fixes one layer. Three rounds is usually enough to get a page that strangers can play back to you accurately, which is a bar most homepages on the internet do not clear.

I’ll admit the reason I think about this so much. I built GazeSite, which does a version of this test with software: it looks at a page the way a first-time visitor would and asks whether the message survives a cold read. But the software exists because the manual version taught me something first, which is that clarity is not a matter of opinion. People talk about homepage copy as if it were taste, like choosing a paint color. It isn’t. Either a stranger can tell you what the page says or she can’t, and you can find out which by asking her. The test is cheap, the answer is binary, and the only real cost is the discomfort of hearing it.

Most founders never pay that cost. They polish the page they can already understand, show it to people who already understand it, and then wonder why traffic doesn’t convert. Five strangers would have told them. Five strangers will tell you. The hard part isn’t the budget. The hard part is asking.