Validate the Problem in Five Messages

The most common form of premature scope is also the most invisible: building against a problem you haven't actually validated. The team has a problem statement. The PRD references it. The deck talks about it. But the problem statement was written in a conference room with no users in the loop.

The five-message test is the cheapest validation procedure I know. Take your problem statement. Write it as one sentence. Send it to five users in your target segment. Ask one question: 'Does this sound like a real problem you have? Reply yes/no with one line of context.'

If three of five reply 'yes' with context that matches your framing — the problem is real. If three of five rephrase the problem back in their own words and the rephrase matches — even better. The problem is not just real, it's articulable.

Three of five does not mean 'yes, build it.' It means the problem is real. The solution is a separate question. The market size is a separate question. The willingness to pay is a separate question. The technical feasibility is a separate question.

But all four of those questions become irrelevant if the problem isn't real. So the five-message test is a gate, not a green light. Pass the gate, then you have permission to spend real product time on the next questions.

The five-message test isn't a green light. It's a gate. Pass it before you spend any more time.

The instinct is to expand the sample. More respondents = more signal. This is wrong at the validation stage. Five is the right number because five is the cohort where you can read every reply, hold every conversation in your head, and notice the rephrase pattern.

At fifty respondents, you start summarizing. Summaries lose the rephrases. The rephrases are the signal. The exact words a user uses to describe a problem are more diagnostic than any aggregate stat. You want to catch them. You can only catch them at small N.

Once the problem is validated at N=5, you can scale to N=50 to size the segment. But sizing is a different question with different methodology. Don't conflate them.

There are two ways the five-message test fails — and both teach you something useful.

First, nobody replies. This means your problem statement is not legible enough to provoke a response. Even if the problem is real, the statement isn't articulating it well enough for a user to recognize. Rewrite and try again with five different users.

Second, three of five reply 'no, but here's what I actually struggle with.' Now you have a different problem. The user is in your segment, but the framing you brought to them is wrong. The signal here is gold — they just told you what to validate next. Rewrite around what they said and run the test again.

Both failure modes are productive. The only unproductive outcome is running the test, getting ambiguous responses, and deciding to 'just start building.' That's the failure mode that destroys quarters.