How to write a belief, find people, ask questions that get stories, and decide what you learned. Written for founders doing it for the first time.
Customer discovery goes wrong when the belief is vague. "Small businesses struggle with scheduling" cannot be tested. Write it as five parts: who, what they do about it today, the problem, when it happens, and why. If you cannot fill in a part, that is your first interview question.
We believe practice managers at small physio clinics, who ring tomorrow's list by hand, lose their first hour to no-shows when patients book weeks ahead, because reminders reach the wrong number.
Each part is a claim you can hear evidence for or against. Narrera labels these parts as you type and asks for the missing ones; you can do the same thing on paper.
Your first round needs five interviews with people who match the "who". Friends and family do not count unless they are practice managers. Find them where they already talk: trade forums, LinkedIn groups, the comments under a competitor's review, the receptionist's desk.
Ask for twenty minutes. Say what you are trying to learn. Do not pitch. People say yes to being asked about their day far more often than to being sold something.
The single most useful rule: ask about something that happened, not something that might. "Would you use an app for reminders?" produces politeness. "Tell me about the last Monday you had a no-show" produces a story, with numbers and names in it.
Shut up after each question. Silence is where the story comes out.
Not every sentence is evidence. Score each claim on a five-point scale: 5 for something that happened, dated and specific; 4 for a concrete claim about their world today; 3 for a general statement; 2 for a hypothetical; 1 for a compliment. Only 4s and 5s count. The evidence scale is explained in full here.
Then count people, not quotes. If three of five people told you a dated story about the same problem, you have something. If one person told you ten things, you have one person.
After five interviews, look at each part of the belief. Which parts have 5s behind them from different people? Which have nothing? Rewrite the belief, keep the people, and run a second round. Stop when two interviews in a row add no new theme.
Write the belief, get ten questions, and let the notetaker keep the calls.