In short: the five-second test asks whether a stranger who sees only your first screen, for five seconds, can say what the product is, what it does for them, and what to do next. Run it on one person who has never seen the page, then grade the answers against the five Clarity tests of the rubric: the kind of thing it is, the outcome, plain words, one promise, and the next step. Fix the first screen before you touch anything below it.
The time is not a figure of speech. Jakob Nielsen found that "users often leave Web pages in 10–20 seconds", and that to keep someone for minutes "you must clearly communicate your value proposition within 10 seconds" (Nielsen Norman Group, *How Long Do Users Stay on Web Pages?*). Five seconds is a strict version of that window, and a useful one: if the first screen survives it, it survives the real visit.
How do you run the test yourself?
You need one person who has never seen the page, and nothing else.
- Show only the first screen. On a laptop, the viewport as it loads, before any scroll. Then do it again on a phone, where the first screen is smaller and usually worse.
- Count five seconds, then hide it. Close the tab or turn the screen away. Do not explain, and do not let them scroll.
- Ask three questions, in this order: - What is it? (a tool, a service, a course, a template…) - What would it do for you? - What would you click, and what do you think happens then?
- Write down their words, not yours. If they say "some kind of AI thing", that is the answer, even if your headline says something else.
A friend who already knows the product cannot do this test. Neither can you. The point is a reader with no context, which is what launch-day traffic is.
What does each answer tell you?
The three questions map onto the five Clarity tests the rubric checks on every first screen. Clarity carries 35 of the 100 points, the largest share, because nothing else on the page gets read if this fails.
| If they could not say… | The test it fails | What it looks like |
|---|---|---|
| what kind of thing it is | C1 Category | The headline is the product name plus a slogan; the category is below the fold |
| what it does for them | C2 Outcome | The headline lists features or describes the product, not a result the buyer would recognise |
| it in words they would use | C3 Plain words | Invented nouns, category jargon, "seamless", "powerful", "AI-driven", "next-gen" |
| one thing it promises | C4 One thing | Three promises in the hero, or three audiences at once |
| what to click and what happens | C5 Next step | Several buttons, or a button whose label does not say what comes after |
Which part fails most often?
On the public pages graded so far, the outcome. As of 21 October 2026, of 32 public launches judged on it, C2 Outcome was passed by 18: 8 passed it only in part and 6 failed it. The category is rarely the problem: C1 was passed by 28 of the same 32. Founders tend to say clearly what they built and leave the reader to work out why it matters to them.
The rest sit in between. C3 Plain words was failed by 4, C4 One thing by 3 with 5 more partial, and C5 Next step was passed in part by 8. These are small samples from the public record, not a survey of the web, and each test page shows the real pages behind the numbers.
Or let the grader read it
Run all 18 tests on your page.
Paste your address. You get a score, a grade and the fixes in two to five minutes. Free, no card.
How do you fix a first screen that fails?
Fix it in the order the stranger reads it.
Name the kind of thing. Put the category in the headline or the line straight under it, in the words a buyer would search for. "Invoicing for freelance designers" passes; a brand name and a slogan do not, however good the slogan.
Say what changes for them. Swap the description of the product for the result: the hours saved, the mistake avoided, the thing they can now do. Write it in the buyer's own words, the ones they use when they complain about the problem, not the ones you use to describe the solution.
Strip the words that do no work. Read the headline aloud and delete every adjective that a competitor could also claim. If a word needs a definition, it is the wrong word for the first screen.
Keep one promise. If the hero makes three claims, pick the one your buyer cares about most and move the others down the page. A first screen that promises one thing gets remembered; one that promises three gets skimmed.
Make the next step obvious. One primary button, with a label that says what happens: "Start free trial", "See pricing", "Book a 15-minute call". Not "Get started", and not three buttons of equal weight.
Then run the test again, on a different person. The second reading is the one that tells you whether the change worked.
What can the five-second test not tell you?
It tells you whether the page is understood, not whether it persuades. A visitor can pass all three questions and still leave because the price is hidden, the proof is thin, or nothing says why to act today. Those are the other three axes of the rubric: Audience, Offer and Proof. The five-second test is the first gate, not the whole check.
It also cannot see your tracking. If people understand the page and you still see no sign-ups, check that the sign-up is recorded before you change a word: launched and got zero sign-ups walks through that order.
The checklist
- Shown to one stranger, first screen only, five seconds, on a laptop and on a phone
- They could say what kind of thing it is (C1)
- They could say what it would do for them, in a result they recognise (C2)
- They repeated it without your jargon (C3)
- They remembered one promise, not three (C4)
- They knew what to click and what would happen (C5)
- Fixed in that order, then tested again on someone new