How do you interview customers without leading them?

A customer conversation is not an interview about your idea. It is a conversation about their past: what happened last time, what they did about it, what it cost. Five rules hold it together: past behavior over opinion, specific events over hypotheticals, listening over talking, silence as a finding, and commitment over compliment. The shape is simple and the follow-up questions do the work.

What is the Mom Test and how do you apply it?

A customer conversation is not an interview about your idea. It is a conversation about their past: what happened last time, what they did about it, what it cost. Five rules hold it together: past behavior over opinion, specific events over hypotheticals, listening over talking, silence as a finding, and commitment over compliment. The shape is simple and the follow-up questions do the work.

What questions should a founder ask in a customer discovery interview?

A customer conversation is not an interview about your idea. It is a conversation about their past: what happened last time, what they did about it, what it cost. Five rules hold it together: past behavior over opinion, specific events over hypotheticals, listening over talking, silence as a finding, and commitment over compliment. The shape is simple and the follow-up questions do the work.

How many customer interviews does a startup need?

A customer conversation is not an interview about your idea. It is a conversation about their past: what happened last time, what they did about it, what it cost. Five rules hold it together: past behavior over opinion, specific events over hypotheticals, listening over talking, silence as a finding, and commitment over compliment. The shape is simple and the follow-up questions do the work.

Customer conversations · Product Sprint

§The question this chapter answers: what actually happened, the last time this person had the problem?

§Everything before this point was preparation for this conversation. You know the market, you have heard the experts, you speak the community's language, and you have a shortlist of people who have the problem worst. Now you have to sit with one of them and not ruin it.

§ 5.1Five rules#

§Rob Fitzpatrick's short book gave the rules their best-known form, and every good interviewer since has arrived at the same five.

§Past behavior over opinion. "Would you use a tool that manages inventory automatically?" gets a yes from everyone, because yes is free. "When did you last run out of something?" gets a Tuesday, an ingredient, and what happened next. Only the second is .

§Specific events over hypotheticals. People are bad at predicting themselves and good at remembering. Ask about the last time, the worst time, the time they almost fixed it. Teresa Torres builds the whole interview around one story for this reason: a story has a beginning, a decision and a cost, and a hypothetical has none of them.

§Listening over talking. You should speak for a fifth of the conversation, at most. If you are explaining your idea, you have stopped learning. If you are filling a silence, you have just interrupted the most useful thing they were about to say.

§Silence as a finding. When someone pauses before answering, the pause is data. They are either searching for a memory that does not exist, which tells you the problem is not frequent, or deciding how honest to be, which tells you the honest answer is interesting. Wait.

§Commitment over compliment. "That sounds really useful" is a compliment and worth nothing. "Can you send me that when it is ready?" is worth a little. "Can I introduce you to my partner, she has it worse" is worth a lot. "Can we pilot it in one location next month" is the only thing worth anything at all. Fitzpatrick's rule: if they are not giving up something, time, reputation or money, they have not told you anything.

§ 5.2The shape of a conversation#

§You do not need a script. You need a shape, and the shape is the same every time.

§Open with their world, not your idea. Who they are, what a normal day looks like, where the problem sits in it. Ten minutes of this is not small talk; it is the context every later answer will be read against.

§Go to the last time. "Walk me through the last time that happened." Then follow the story: what did you do, what did you try, what did it cost, who else was involved, how did it end. Stay in the specific event. Every time they generalize, bring them back: "and that particular time, what did you actually do?"

§Explore what they have tried. Every workaround they mention is a competitor and a price signal. "What have you tried?" followed by "why did you stop?" tells you what the market has already rejected.

§Only then, if they bring it up, the solution. If they ask what you are building, tell them in one sentence and immediately return to them: "does that fit what you were describing, or did I get it wrong?" Focus on what, not how. Notice whether they light up or nod politely. The difference is visible.

§Close with a commitment. An introduction to the one person they think you must talk to. Permission to come back when there is something to see. A place in a pilot. Take what they offer and note what they dodge.

§ 5.3The follow-up questions do the work#

§The first answer to any question is the rehearsed one. The is in the second and third layers, and you reach them with a handful of follow-ups you will use in every conversation you ever run:

§"Tell me more about that." "What did you do next?" "How did that affect you?" "Why was that important?" "What else did you try?" "What happened then?"

§These are not techniques. They are the questions a curious friend would ask, and they work because they signal that you want the story and not the summary.

§ 5.4Bad exchange, good exchange#

§The difference is easiest to hear.

§Three questions, three compliments, zero facts. The founder leaves believing they validated something.

§One story, a cost in money and in a lost table, a workaround and its precise failure. That last line, "closing is too late," is a in the customer's words, and it is the kind of sentence chapter 7 is built from.

§ 5.5How many#

§Enough that you can predict the next one. Somewhere between eight and fifteen people from the same tightly described segment, you will notice that you know what the answer to each question will be before it is given. That is when to stop interviewing that segment and start synthesizing. Fewer than five and you are still hearing anecdotes. Many more than twenty and you have probably widened the segment without noticing.

§ 5.6What you leave with#

§One page per conversation, written before you do anything else that day: three verbatim quotes, the specific event they described and its cost, every workaround mentioned, what they agreed to at the end and how readily. And, after each, one line answering the question you are actually there to answer: did this person change what I believe, and how?

5. Customer conversations