The team has spent three meetings debating the homepage headline. One person wants “solutions.” Another wants “transformation.” A third suggests the phrase customers used in a sales call last week. Each sentence sounds plausible in the conference room, where everyone already knows what the company does.
Before rewriting the whole site, find out what an unfamiliar person understands now. A small comprehension test is not a market survey and five people cannot represent every buyer. It is a fast way to catch obvious ambiguity and decide what to investigate next.
Choose the right five people
Recruit people who resemble the audience enough to interpret the offer, but who do not work at the company or know the internal shorthand. If the service is for finance leaders at midsize manufacturers, asking five friends who run unrelated consumer businesses may expose confusing language but not validate positioning. Note how close each participant is to the intended audience.
Avoid recruiting only supportive customers. Existing customers already understand the business and may fill gaps from memory. If you have access to recent prospects who declined, their confusion can be especially useful, though their feedback may reflect a specific context. Ask for candid reactions and make it easy to say “I don’t know.”
Prepare the page and the questions
Use the page or sales material a new prospect actually sees. Do not narrate it or explain what the company meant. If you are testing a landing page, capture the page on a phone and desktop when the experiences differ. Decide in advance what you want to learn: the offer, the intended customer, the outcome, and the next step.
Ask participants to view the page for five to ten seconds, then hide it. Use neutral questions:
- What do you think this business sells or does?
- Who do you think it is for?
- What problem does it help with?
- What would you expect to happen if you clicked the main button?
- What, if anything, feels unclear or surprising?
Do not ask, “Was the headline clear?” That invites politeness and leads participants toward your preferred answer. GOV.UK’s user-needs guidance recommends learning what people need to do in language they recognize; this kind of test is one modest way to listen before settling on internal terminology. User-needs guidance.
Record the words, not just a score
Write the answer as close to verbatim as possible. Note hesitation, mistaken assumptions, and what attracted attention. A score such as “four out of five got it” hides the valuable part: perhaps all four described a different audience, or one person understood the category but missed the differentiating service.
After each session, ask one follow-up: “What on the page led you to that interpretation?” This helps separate a wording issue from a proof or hierarchy issue. If participants guessed based on a photo, the text might be invisible. If they understood the service but not the target customer, the page may be too broad. If the offer made sense but the button did not, the next action needs attention.
Treat repeated patterns as clues, not statistical conclusions. Five people are enough to uncover some usability problems; they do not establish the percentage of a market that misunderstands the offer. People also behave differently when interviewed than when they arrive under time pressure from an ad.
Sort the confusion before changing copy
Group observations into four buckets:
Category confusion: People cannot tell whether the business is software, consulting, a marketplace, or something else. Clarify the basic category and service.
Audience confusion: They understand what is provided but cannot tell whether it is meant for their situation. Name the relevant buyer, use case, or constraint.
Outcome confusion: They can name the product but cannot explain why they would want it. State the practical job or result without promising what the business cannot control.
Action confusion: They understand the offer but cannot predict what happens after clicking. Rename the action and explain the immediate next step.
Choose the earliest confusion that appears across several sessions and blocks a likely buyer. If people cannot identify the category, polishing the CTA will not solve the more basic problem. If they already understand the offer but hesitate because of price, scope, proof, or timing, rewriting the headline may be avoidance disguised as progress.
Turn the finding into one useful rewrite
Draft a plain-language explanation containing four elements: who it serves, what it provides, the problem or job, and evidence or context that makes the claim credible. It is a working sentence, not necessarily final homepage copy. For example, “We help [specific kind of team] do [specific job] through [offering], with [verifiable evidence or process].” Remove any bracketed claim the business cannot support.
Test the revision with a fresh group where possible. Reusing the same five people can make the result look better because they remember the earlier version. Compare what participants understand, what they still ask, and whether the next step is clearer. If the revision sounds simpler internally but produces the same confusion externally, keep learning.
Also check actual customer language. Read inquiry forms, support tickets, sales notes, and interview transcripts. The words buyers use to describe the problem may differ from how the company labels its category. Do not paste raw phrases mechanically; understand what they mean and whether they apply to the intended audience.
When the problem is not the sentence
Sometimes a five-second explanation is hard because the offer itself is unsettled. Different customers buy different services for unrelated reasons. The business may be serving several audiences, or the product may still be changing. A concise page cannot resolve a strategic choice that leadership has not made.
A founder on Indie Hackers described visits from multiple channels but no signups and wondered whether visitors could understand the product quickly. That discussion illustrates a useful diagnostic question, not proof that a clearer sentence would have created signups; the same founder also noted weak subsequent usage, which points to a separate retention question. Founder discussion.
Keep comprehension, desire, trust, and action separate. Someone can understand an offer and not want it. They can want it but distrust the seller. They can trust the seller but not be ready to act. The test tells you what people understood in a short encounter; interviews and behavior help explain the rest.
The first change should answer a real question
At the end of five sessions, you should be able to write a specific observation: “Three of five people thought our audit was software because the page leads with product screenshots,” or “People understood the service but could not tell whether we serve their region.” That is more actionable than “the website feels vague.”
Change the part connected to the observation, then observe again. Sometimes the fix is a sentence. Sometimes it is the order of information, an example, the service structure, or a clearer explanation of what happens next. Sometimes the offer needs work before the page does. The point of the exercise is not to protect the website from a redesign; it is to make sure the first edit addresses the confusion people actually had.