Two posts crossed my feed within hours of each other this morning. A venture development strategist wrote that a founder can interview thirty potential customers, document every answer and still learn very little, because collecting information is not the same as using evidence to make a decision. A startup program designer, in a separate post, named the exact sentence his team has learned to distrust in grant reports: forty interviews, problem validated.
Both are describing the same quiet failure. The startup world won the battle to make founders talk to customers. It has not yet won the battle to make those conversations change anything.
Why does the interview count feel like progress?
Because it is countable. Accelerators ask for it, investors nod at it, and every completed call produces the warm sensation of work. Counting interviews is easy. Deciding because of them is painful, because a real decision closes doors. The founder who says forty interviews, problem validated is usually reporting effort, not evidence. The number proves attendance. It says nothing about what survived contact with the answers.
There is also a comfort in the pile itself. As long as the notes stay notes, every version of the idea is still alive. The moment you treat one answer as binding, some version of the product dies. Transcripts preserve optionality. Evidence kills it. That is exactly why evidence is worth more.
What should actually change after thirty interviews?
At least one of these, and usually several. The target customer gets narrower, because some segment you imagined turned out not to feel the pain. A problem you assumed was urgent gets demoted, because people described it with a shrug. You discover the person experiencing the problem is not the person who pays to solve it, which quietly rewrites your whole go-to-market. The proposed solution, the price, the channel or the business model shifts. Or you decide not to pursue the idea in its current form at all, which is not wasted effort but the cheapest exit you will ever be offered.
Notice what is on that list: decisions. Not themes, not word clouds, not a slide titled key learnings. Decisions.
How do you tell transcription from discovery?
Run the dead-belief test. Before the next batch of conversations, write down what you currently believe: who has the problem, how often, what it costs them, what they do about it today, who signs the invoice. Date the document. After the interviews, read it again and mark what you no longer believe. The marked lines are your discovery. If nothing is marked, one of two things happened. Either your beliefs were already perfectly calibrated, which would make you the first founder in history, or the interviews were built to confirm rather than to test, and they did their job.
Why do founders stop at collecting?
Partly incentive: programs reward the count, so founders optimize the count. Partly fear: an interview that changes nothing threatens nothing. But mostly because the collecting step and the deciding step feel like the same activity, and they are not. Collecting is listening. Deciding is admitting, in writing, that last month you were wrong about something specific. The second one is the research. The first one is just its raw material.
Key takeaways
- Interview counts measure effort, not evidence. Forty interviews can validate nothing.
- The deliverable of discovery is a changed decision: a narrower customer, a demoted problem, a different buyer, a killed idea.
- Write your beliefs down before interviewing, then diff. The dead beliefs are the discovery.
- A notes folder that changes no decision is not research. It is a diary with a sample size.
The founders who get the most out of discovery are not the ones with the most conversations. They are the ones who can tell you, with a date attached, which belief died and what it cost to keep believing it. Everything else is a word count.