To test positioning before you launch, run two tests, not one. A comprehension test proves a stranger can repeat your claim back accurately. A trade-off test proves that same stranger would choose it over what they already do. Most founders run the first, skip the second, and launch on evidence that only proves they were understood.
Understood is not the same as chosen. That gap is where quiet launch failures live. The message was clear. The deck was tight. Everyone nodded. Nobody moved.
This is a testable problem, and the methods are not new. They are standard practice in consumer research, where a claim has to survive a forced choice before anyone spends money on it. They have just never been translated properly for founders. Search for how to test positioning and you get five-second tests and lists of warning signs. Useful. Incomplete.
Why does positioning that tests well still fail?
Because the usual tests measure legibility, and legibility is a low bar.
Show someone your headline for five seconds and ask what you do. Ask a past client how they describe you to a friend. Read your one-liner aloud and check whether it sounds like everyone else. These are all worth doing. All three answer the same question: is this message readable?
Readable is a language problem. Chosen is an economics problem. It asks what a person would give up to get what you are offering, and whether the thing they already do is good enough to keep doing.
There is a second reason, and it is harder to see. The question you ask determines the answer you get. Ask “does this make sense?” and almost everyone says yes, because saying no is socially expensive and costs them nothing to avoid. Ask “which of these three would you actually pick, and why?” and the answers separate immediately. That is not people being dishonest. It is a badly designed question producing clean data about the wrong thing.
What does a comprehension test actually prove?
It proves your words work. Run it first, because everything downstream depends on it, and keep it to three checks.
- The repeat-back. Show the claim once. Take it away. Ask them to say what you do and who it is for, in their own words. If they reach for your phrasing, they memorised it. If they use their own words and stay accurate, they understood it.
- The exclusion check. Ask who this is clearly not for. If they cannot answer, your positioning has not been decided yet — it has only been described. Positioning that excludes nobody is a brochure.
- The category check. Ask what they would compare you to. If they name a category you did not intend, your claim is landing somewhere you were not aiming. You will be judged by that category’s standards and priced against its benchmarks, whether or not you agree.
- Force the choice. Put your positioning next to the real alternatives and ask which one they would pick. No “all of the above”. No rating scales. One pick, then ask why. The reasoning is the finding; the tally is not.
- Attach a cost. Ask for something that costs the respondent a little: a calendar slot, a named budget range, a paid pilot, an introduction to the colleague who owns the problem. Enthusiasm that costs nothing is the cheapest data you can collect, and you will collect a lot of it.
- Test for substitution. Ask what they would stop doing if they worked with you. If the honest answer is nothing, you are an addition rather than a replacement. Additions are the first line cut when a budget tightens.
- Listen for the hesitation, not the verdict. Where someone pauses, re-reads, or asks a clarifying question tells you more than their final answer. Write down the pause.
- Not understood. A language problem. The strategy may be sound and the sentence is doing it damage. Rewrite the claim. Do not change the business.
- Understood, not wanted. A positioning problem. You are being perfectly clear about something people do not need enough to act on. No amount of rewriting fixes this. Go back to the market.
- Wanted, but not from you. A credibility problem. The claim is right and the proof is missing. This is fixed with evidence — named work, specifics, results someone else can verify — not with stronger adjectives.
The exclusion check is the one founders skip, and it is the one that exposes whether a real decision has been made. If you have not yet made that decision, the problem is upstream of testing. Start with what founder positioning is meant to settle, and with whether you are testing the company’s claim, the founder’s claim, or both — they fail in different ways and need different evidence.
How do you test whether positioning is chosen, not just understood?
You make people choose, and you make choosing cost something.
None of this requires a research budget or a panel provider. It requires asking a harder question than the one that produces a comfortable answer.
What are you actually competing against?
Usually the status quo, and founders almost never put it in the test.
The comparison set a founder builds is made of companies that look like theirs. The comparison set a buyer builds is made of anything that resolves the same problem: the spreadsheet that already works, the agency on retainer, the junior hire, the decision to wait another quarter. Those are the real competitors. Leave them out and you get a clean result on a question nobody was asking.
Build the set from the buyer’s side. The simplest way to find it is to ask what they tried before, and what happened.
How many people do you need to test with?
It depends entirely on what you intend to claim afterwards, and this is where founders overreach.
A small round of conversations finds patterns, not percentages. It tells you which words caused confusion, which alternative kept coming up, where people hesitated. That is real, usable evidence. It is not a measurement.
The moment you write a percentage on a slide, you have implied a designed sample: a defined population, a recruitment method, enough respondents for the number to mean anything. A percentage taken from twelve friendly conversations is decoration. It looks like rigour and it is the opposite, because it hides how thin the base is.
So say which one you ran. “Twelve conversations, and the same objection came up in nine of them” is honest and persuasive. “75% prefer our positioning” is neither. The same discipline applies to discovery interviews generally: the goal is evidence, not encouragement.
What do you do when the test fails?
Diagnose which of three things failed, because each one has a different fix and they are routinely confused.
Founders treat all three as writing problems. Only the first one is. The second and third get harder the longer you spend rewording your way around them.
When should positioning testing happen?
After you have decided what you are claiming. Before you build a launch on top of it.
It is a distinct step, and it gets skipped because it looks like two other things it is not. It is not product validation — that asks whether the problem and the solution are real, and it answers a narrower question than most founders think. It is not brand work either; nothing here is about how anything looks.
Positioning testing sits in the gap between the decision and the launch, which is exactly where the go-to-market sequence tends to lose its footing. Everything downstream — the website, the deck, the outbound, the content — inherits whatever the positioning got right or wrong. Testing it costs a week. Discovering the problem after launch costs a quarter, and by then the evidence arrives as silence, which is the hardest kind to interpret.
Clear is the minimum. Chosen is the point.
Written by Neha Vyas, founder of You Creatives, a Stockholm-based brand strategy, positioning and go-to-market consultancy. Ten years in strategic consumer and market research, a BSc in Computer Science Engineering and an MBA in Marketing.
