Twenty Strangers and Their Trouble
A customer-discovery interview gathers facts about a person's past behavior around a problem and never pitches the solution. · 11 min
In the last folio you found a problem with evidence behind it. Now you go and confirm it with the people who live it. The tool is the customer-discovery interview — a conversation whose whole job is to collect facts about what a person has already done about the problem. It has one discipline that beginners find surprisingly hard: you never pitch your solution. The moment you sell, the person starts being polite, and polite people stop telling the truth.
Guess before you learn
You are interviewing a dog owner about walking their dog on busy days. Which question belongs in a discovery interview?
The middle question. It asks about a real event in the past, so the answer is a fact, not a guess. The first pitches a solution and asks about an imagined future; the third leads the witness. If you were tempted by the pitch, that is the instinct this whole folio works to retrain.
9–12
3–5
When you want to learn about someone's problem, ask what they have already done, not what they think of your idea. 'The last time this happened, what did you do?' gets a true story. 'Would you buy my thing?' only gets a guess, and people guess kindly to be nice.
Your job in the talk is mostly to listen. Ask a short question, then be quiet and let the whole answer come out before you say anything back.
6–8
A discovery interview trades in past behavior, not opinions or predictions. Ask 'when did you last face this, and what did you do?' rather than 'would you buy this?' Past behavior is a fact you can check; a prediction about buying is a guess the person makes to please you. Two rules follow: never describe your solution, and after each question, stay silent long enough for the real answer to arrive.
9–12
The interview is a search for facts a hopeful founder would otherwise miss: what the person did, when, how often, what it cost them, and what they used instead. Every question about the future ('would you', 'could you', 'do you think you'd') invites a courtesy answer worth nothing. Anchor each question to a specific past episode, and treat compliments as noise — you are collecting behavior, not approval.
K–2
Ask a friend what they did, not what they might do. 'What did you eat for lunch?' has a real answer. 'Would you like a moon sandwich?' just makes them guess.
A story about yesterday is true. A promise about tomorrow is only a maybe. So ask about yesterday.
Undergrad
Structure the interview to defeat two biases at once. Confirmation bias pushes you to hear agreement; social-desirability bias pushes the respondent to give it. Both are neutralized by anchoring to concrete past events: 'Walk me through the last time.' Facts about what already happened cannot be inflated by politeness the way predictions can. Keep the sample honest, too — interview people who have the problem, not only the friends who like you.
Postgrad
This is qualitative field research with a strong prior for retrospective, episodic elicitation over prospective, hypothetical questioning. Hypotheticals measure imagined utility under demand characteristics; recalled episodes approximate revealed behavior. The founder is a compromised instrument, so the protocol does the work: open prompts, no leading, no solution disclosure, and disciplined silence to avoid contaminating the response. Analysis codes for behaviors and their costs, not for expressed enthusiasm, which is confound, not signal.
discovery interview
A conversation that collects facts about what a person has already done about a problem, with no pitch of your solution.
Why is this true?
Why does pitching your solution ruin an interview?
Once you sell, the person shifts into being polite about your idea, and polite answers hide the truth. You want their honest history, which only comes out when they are not protecting your feelings.
The most common failure is a question that hides a pitch or leads to a yes. Fixing it is a skill you can practice: strip out the solution, drop the imagined future, and anchor the question to a real past episode. The worked example below turns one bad question into a clean one, step by step. Notice that each step removes something — the sale, the hypothetical, the nudge — until only a fact-finding question is left.
Turn a leading pitch into a discovery question — the steps fade as you master them
"Would you pay $10 a week for someone to walk your dog?"
"How do you handle walking your dog on busy days?"
"Tell me about the last busy week — what did you actually do about the dog?"
(ask, then say nothing; write down what they did)
Do this twenty times and a pattern appears that no single conversation could show you: the same workaround named again and again, the same hour lost, the same frustration in nearly the same words. That repeated language is a gift — it is how real customers describe their own problem, and you will reuse it. In the next folio you compress everything you heard into a single sentence: the value proposition.
Practice — new ink and old, interleaved
1.Which of these is a leading question in disguise?
2.You want to know if a problem is frequent. Which question finds out best?
3.In one sentence, why do you ask about the past instead of the future?
4.What are you listening for across many interviews?
The same workaround, cost, and words repeating — the pattern that reveals the real problem and how customers describe it.
How close were you? Grade yourself honestly — it sets your review date.