The Message Test: Can a Stranger Repeat What Your Product Does?

Can a stranger repeat what your product does after one read?

That single question is the cheapest positioning audit in B2B software. Show someone outside your company your homepage for thirty seconds, take it away, and ask them to tell you what you sell and who it is for. If they cannot, no amount of paid traffic will fix the page.

I run this test before I touch a word of copy, and it fails more often than founders expect. Not because the writing is bad. Usually the writing is polished. It fails because the page describes a category, a philosophy, or a feeling instead of a product.

Clarity is not a style preference. It is the thing that decides whether a visitor can form an opinion at all.

What is the message test and how do you run it?

Find five people who do not work with you, show each one your homepage for thirty seconds, then hide it. Ask three questions. What does this company sell. Who is it for. Why would someone pick it. Write down their exact words. Do not explain, correct, or help. Their confusion is the finding.

Five people is enough to see a pattern. If four of them say something close to what you meant, your message is working. If you get five different answers, you do not have a messaging problem in the small sense of word choice. You have an agreement problem inside your own company about what you sell.

The detail that makes this work is hiding the page. People are generous readers when the text is in front of them. Take it away and you measure what actually landed, which is the only thing a busy buyer ever gets.

Record the words they use. Those words are more valuable than the ones your team invented, because they are the words your future customers will type into a search box or into ChatGPT.

Why does this beat asking people if your messaging is clear?

Because nobody answers that question honestly. Ask if a page is clear and you get politeness. Ask someone to repeat what the page said and you get evidence. The first question tests their manners, and people are kind. The second tests your writing, and writing cannot be kind back.

The same trap catches internal reviews. Your team cannot fail the message test, because your team already knows the answer. Every person in the building carries context that the visitor does not have, and that context quietly fills every gap in the copy.

This is why I no longer trust a stakeholder review as a substitute for a reader test. A stakeholder review tells you whether the page offends anyone internally. It tells you nothing about whether it communicates.

What kinds of failure does the test expose?

Four failures show up again and again. Category confusion, where nobody can name what kind of thing you are. Audience drift, where the page tries to speak to everyone. Benefit fog, where the claims are true of any software. And proof absence, where the reader believes nothing because you have shown nothing.

Category confusion is the worst of the four, because the reader cannot even file you mentally. A page that says it helps teams move faster could be a project tool, a CI system, an agency, or a consultancy. Give the reader the noun first, then the difference.

Benefit fog is the most common. If your headline would still be true with a competitor's logo on it, it is not a headline, it is a mood. I make a habit of pasting a rival's name into a client's homepage claims. If nothing breaks, the copy is not doing work.

Proof absence has a design dimension too. Logos alone rarely close the gap, which is why I argued that customer logos are not proof. Readers want a specific outcome attached to a specific situation.

Why do founders write messaging that fails this test?

Because founders write from the inside out. You know the problem so deeply that the interesting part, to you, is the nuance. To a first time reader, the interesting part is the basics you find boring. The curse of knowledge is not carelessness. It is expertise pointed in the wrong direction.

Pressure makes it worse. Webflow's 2026 State of the Website report found that 96 percent of marketing leaders believe the growing pace of innovation and customer expectations has increased the pressure on their department in the last year. Teams under that kind of pressure ship more pages, not clearer ones.

There is also a fear at work. Naming your category and your buyer feels like closing doors. Founders write broadly to keep options open, and end up with a page that converts nobody in particular. Specific pages lose the wrong visitors on purpose, which is what makes them work.

How do you rewrite a message that fails?

Write one sentence with four parts. The noun for what you are, the specific buyer, the job it does for them, and the one thing that makes it different. Then read it aloud. If you run out of breath or reach for a comma more than twice, cut until you do not.

Put that sentence at the top of the page in plain language, above anything clever. The clever line can live underneath. I have watched many teams spend weeks on a tagline while the sentence that actually explains the product sits three scrolls down. That order is backwards.

Then rebuild the page around the four questions readers ask silently. What is it, who is it for, what happens when I use it, and why should I believe you. That structure is unglamorous and it converts. I unpacked the common structural mistakes in my piece on founder homepage structure, and nearly all of them come from answering question four before question one.

Finally, test the hero section on its own. The hero carries most of the weight, and it is the part most often written last and fastest. My rules for it are in designing a hero that says what you do.

Does the message test matter for AI search too?

More than ever. Systems like ChatGPT, Perplexity, Claude and Google AI Overviews summarise your page before a buyer ever sees it. If a human cannot repeat what you do after one read, a model producing a two sentence summary will not do better, and its guess becomes the version buyers hear.

The same Webflow report found that more than half of marketing leaders, 52 percent, will prioritise optimisation for AI-driven search and summaries in 2026. That priority is usually spent on schema markup and technical work. Those matter, but the raw material is the sentence. A model can only compress what your page actually says.

So the answer engine version of the message test is simple. Ask an AI assistant to describe your company from your homepage alone, then compare its answer to what you meant. I do this for every client now. It takes two minutes and it is frequently uncomfortable.

How often should you run it?

Run it whenever the product, the buyer, or the market moves, and at least twice a year regardless. Messaging decays quietly because the company changes faster than the homepage does. A page written for your first ten customers rarely fits your next hundred.

Keep the results. When you rewrite, you want yesterday's confused answers next to today's, because that comparison is the only clean measure of improvement you will get without a large test budget.

Do not wait for a redesign. Redesigns are expensive, slow, and usually triggered by taste rather than evidence. A message test costs an afternoon and tells you whether the redesign is the actual problem.

What should you do next?

Book five short conversations this week with people outside your company. Run the thirty second test, write down their exact words, and compare their answers to what you believe you sell. Then rewrite one sentence, put it at the top of your homepage, and run the test again with five new people.

The teams that win on clarity are rarely the best writers. They are the ones willing to hear that the page they are proud of did not land, and to fix it without arguing with the evidence.

If you run the test and the answers surprise you, send me what people said. I am happy to look at the page and tell you where the message is leaking.