Three times in one week, in three different startup forums, the same post appeared with different names on it. A founder, early and earnest, publishing a numbered list of questions. One had five, one had ten, one had fourteen. Each opened with the same disclaimer: not selling anything, just trying to learn.
The instinct deserves respect. These are founders trying to do research before building, which already puts them ahead of most. The instrument is the problem. Every question on every list was hypothetical, generic, and aimed at the future. Would you use a tool that did this. How much would you pay for that. What features would matter to you.
Why do hypothetical questions fail?
Because people answer them as the person they would like to be. Asked whether they would use a budgeting app, almost everyone is a diligent saver. Asked whether they would pay for a tool that saves an hour a day, almost everyone is a rational buyer of time. None of this costs the respondent anything, and answers that cost nothing carry no information. The forum setting makes it worse: strangers being asked for an opinion default to generosity, because generosity is free and arguing is work.
So the fourteen answers come back warm, encouraging, and interchangeable. The founder reads them as a green light. What they actually measured is the politeness of passers-by.
What does a real demand question sound like?
It is retrospective and specific. Not would you, but when did you last. Not how much would you pay, but what did you pay, to whom, and what made you stop. The last time this problem showed up, what did you actually do about it. Past behavior is the only part of a stranger's testimony that was expensive for them to produce, which is exactly why it is the only part worth collecting.
One of the week's posts came from a student who asked a forum what he should change before writing a single line of code. Right instinct, and a better question than most. What it still cannot buy him is the answer that matters, because the people replying are whoever happened to scroll past, not the people who live inside the problem.
Isn't some feedback better than none?
No, and this is the uncomfortable part. Zero data leaves you cautious. Compliant data makes you confident, and confidence is what signs the next six months of your life over to a roadmap. The founders who got burned rarely skipped research. They ran research that was incapable of saying no, counted the yeses, and built. By the time real behavior finally spoke, through silent signups and empty dashboards, the verdict was expensive.
The question list was never the asset. The asset is who is answering, whether they match your actual buyer, and whether their answer describes something they did rather than something they might do. Fourteen questions to the wrong people about the future is noise with a sample size. One question to the right person about last month is evidence.
Key takeaways
- Hypothetical, future-tense questions harvest politeness: people answer as the person they would like to be.
- Forum respondents are whoever scrolled past, not your market. Who answers matters more than how many questions you ask.
- Compliant data is worse than no data, because it buys confidence without buying truth.
- Evidence is retrospective and specific: what matched people actually did and paid, the last time the problem showed up.
Why this today: the three questionnaire posts appeared independently within days of each other, from founders in different industries, which makes the pattern the story.
People will happily tell you who they want to be. Build for what they did.