The Five Second Test for a SaaS Homepage

A five second test for a homepage is simple. Show a stranger the first screen for five seconds, hide it, and ask three questions: what is this, who is it for, and what would you click. If most people cannot name the kind of product it is, the homepage is failing, and nothing further down the page will win that visitor back.

You can run it this week with five people, a screen share and a notes file. No panel, no tool, no budget. It matters most to an early-stage founder because he is the one person who can never take the test. He knows the product too well to see the page the way a stranger does.

Four of five people said some kind of analytics thing

Picture an AI tool for support teams. It sits inside Zendesk, reads each new ticket, drafts a reply from the help docs and past answers, and tags the ticket. A support agent reads the draft, edits a word or two, and clicks send. That is the whole product, and it is a good one.

Now picture its homepage. The headline reads Know what your customers mean. The subhead reads AI-native insight for support and product teams. On the right sits the only image above the fold: a dashboard with a purple line chart, a tile that says Sentiment 72, and a donut chart labelled Top themes. Two buttons sit side by side at the same size, Get started and Watch the film. The words Zendesk, reply and ticket first appear in the third section, under a heading that says How it works.

Five people see that screen for five seconds each. Asked what it is, four of them say something close to "some kind of analytics thing?" The fifth guesses a customer feedback survey tool. Asked who it is for, three say marketing teams. Asked what they would click, three pick Watch the film, because nobody wants to get started on something they cannot name. Not one of the five says the product writes replies. The page described a chart, so the strangers described a chart.

Why the founder is the one person who cannot run it

The founder reads Know what your customers mean and his head fills in the rest: the Zendesk sidebar, the drafted reply, the agent clicking send. He cannot read the page without that picture, any more than you can look at a word in your own language and not read it. That blind spot is the same one the curse of knowledge in product design describes for screens, and it applies to every line of a homepage too.

So he checks the page the way founders do. He posts the link in the team Slack and gets six thumbs up. He shows it to two angel investors who have heard the pitch three times. He asks his cofounder whether the headline is clear, and his cofounder, who wrote half the product, says yes. Every one of those people already knows the answer to the first question. None of them is a test. When someone outside calls the site vague, he rewrites the headline again, more specific about the feeling and no clearer about the thing.

How to run the test this week with five people

Most guides send you to a testing panel and a sample of thirty people. That is useful when you want a percentage. You want something cruder: does a stranger file you in the right drawer, yes or no. Five people answer that, and you can find them by Thursday.

• Take two screenshots of the first screen only, exactly as it loads, nothing below the fold. One at laptop width, one on a phone, and use the phone version with two of the five.

• Find five people who have never heard the pitch. Not your team, not your investors, not a friend you told about the company at a wedding. People who roughly match your buyer are best. A founder from a different field in your coworking space is fine. Your partner is not.

• Get on a video call with each one. Say you are going to show a page for five seconds and then ask three short questions, and that there are no wrong answers.

• Share the screenshot, count five seconds in your head, and stop sharing. Do not let them scroll, and do not show the live site, where they can click around.

• Ask the three questions in the next section, in the same order, word for word. Type each answer exactly as they say it, including the hesitation.

• Say nothing about what the product really is until all three answers are written down. The moment you explain, the test is over for that person.

Each call takes about four minutes, so the test fits into one afternoon. Once someone has heard you explain the product, they are no longer a stranger, so keep fresh names for the rerun.

The three questions, and the ones to leave out

Ask exactly these. First: what kind of product was that? Second: who do you think it is for? Third: if you were interested, what would you click first? The first question is the one that decides the test. The other two tell you what to fix once the first one passes.

Leave out the questions that teach you nothing. Did you like it and was it clear both invite politeness, because saying no feels rude. What do you remember is also weaker than it looks. A visitor can remember the purple chart perfectly and still have no idea what the product does. You are testing whether they can name it, not whether they noticed it.

Scoring the answers so the result is not a vibe

Score the first answer from each person on a three point scale, and decide the scale before the first call so you cannot bend it afterwards.

• Two points: they name your category, or a noun a buyer would type into a search bar to find you. For the support tool, AI replies for support tickets or a helpdesk assistant both count.

• One point: they name a neighbouring category. Customer feedback tool is a near miss. It is the right department and the wrong job.

• Zero points: they answer with a benefit, a vague shape, or a shrug. Some kind of AI thing, an analytics thing, a platform for teams, and no idea all score zero.

The page passes when four of five people score two. Three of five is a fail, not a near pass, because the people on your calls were paying attention and your real visitors are not. For the second question, a pass is a role and a rough company size that match your buyer. For the third, a pass is your primary button. Score those as right or wrong. The support tool above scores one point out of ten on the first question. Four zeros and one near miss is not a headline problem. The page has never said what the product is.

What each wrong answer tells you to change

Each question points at a different part of the first screen, so the fix follows from which one failed.

• Wrong category, or a benefit instead of a noun. The noun is missing from the headline and subhead, or the image is telling a different story. Put a noun your buyers already use above the fold, and our guide to naming your product category on the homepage covers how to pick it. Then check the image. If the only picture is a dashboard, people will say dashboard whatever the words say.

• Right category, wrong audience. The subhead is too broad. Support and product teams covers almost every company. Name the role and the tool they live in, such as support leads on Zendesk.

• Right category and audience, wrong click. Two buttons of equal weight, or a primary button that asks too much. Make one button clearly the main one and name the first step, not the relationship. The anatomy of the first screen is set out in SaaS hero section design.

For the support tool, the rewrite does not need new design. The headline becomes AI replies for your Zendesk tickets. The subhead becomes It drafts the answer from your help docs, and your agent clicks send. The dashboard picture is replaced by one Zendesk ticket with a drafted reply under it and a green Send button. The main button becomes Try it on 20 of your tickets, and Watch the film shrinks to a text link. Then run the test again with five new people. If four of them say something about replying to support tickets, you are done with the top of the page.

Change one thing per round where you can, so you know which change worked. For the rest of the page, the SaaS homepage audit checklist runs the same stranger logic section by section.

Why nothing below the fold wins the visit back

A common answer to a vague hero is that the page explains everything further down. The support tool really does say Zendesk and reply in its third section. It does not matter. A visitor who has filed the product as analytics in the first few seconds reads the rest of the page as an analytics page, or does not read it at all. Nielsen Norman Group's research on page visit length found that the first 10 seconds are critical to the decision to stay or leave, and that users often leave pages within 10 to 20 seconds. The careful explanation in section three is written for people who have already gone.

Most homepage advice treats the page as an argument that builds, with the hero as a hook and the proof lower down. For an early-stage company with no brand to lean on, the hero is not a hook. It is the whole decision for most visitors. Proof, features and pricing only work on people who already know what they are looking at.

Every one of those visitors was paid for, in ad spend, in a blog post that took a week, or in the founder's evenings on LinkedIn. The cost lands whether or not anyone understood the page, and buying more traffic makes the waste bigger at the same rate. The people who do sign up came in expecting analytics, so they open the product looking for charts and find a reply drafter. Product-led benchmarks, measured on SaaS companies, put activation at 20 to 40 percent for most products. A visitor who arrives with the wrong picture is starting from behind that number before the first screen loads. With the Series A bar for AI startups around 3.5 million in ARR (Carta, Q1 2026), that is a slow leak a seed company cannot carry for long.

Studio Maydit is a web and product design studio for AI founders in the US, UK and Europe. We rebuild the first screen so a stranger can name the product in five seconds, then carry that clarity through the rest of the site in Framer, Webflow or custom code, usually on a fixed scope of three to four weeks. If your homepage passes every review inside the company and still loses strangers, book a 30 minute call and we will look at your first screen with you.