Customer interviews that tell you the truth

Ask a friend whether your business idea is good and they will almost always say yes. They are being kind, not dishonest. The trouble is that founders, product teams and marketers often run customer interviews the same way: describe the idea, ask what the person thinks, and leave encouraged by answers that mean very little.

Rob Fitzpatrick, a founder and Y Combinator alumnus, wrote a short book about this called The Mom Test. The title comes from the idea that you should ask questions that even your mother could not lie to you about, because they are about facts in her life rather than her opinion of you. This guide draws on his approach; the book itself is well worth reading in full.

The three rules

Fitzpatrick’s method rests on three simple rules:

1. Talk about their life instead of your idea. Once you describe your product, the conversation becomes about being polite to you.
2. Ask about specifics in the past instead of generics or opinions about the future. “Would you use this?” invites a guess. “When did this last happen?” invites a fact.
3. Talk less and listen more. The more you explain, the less you learn.

Bad questions and better ones

The pattern is the same throughout: replace a question about the future or about your idea with one about what has actually happened.

Spot the noise

Fitzpatrick warns about three kinds of answers that feel good but mislead:

• Compliments. “That sounds really useful.” Pleasant, and almost worthless as evidence.
• Fluff. Generic statements (“I always”, “I usually”), future promises (“I would definitely”) and hypotheticals (“I might”). Gently bring the person back to a specific example.
• Ideas and feature requests. Useful as clues, but dig into why they want it rather than writing it straight onto the roadmap.

Look for commitment

Polite interest costs the other person nothing. Real interest costs them something. Fitzpatrick groups these costs as time, reputation and money:

• Time: agreeing to a longer follow-up session, or trying an early version and giving feedback.
• Reputation: introducing you to their manager, a colleague or another potential customer.
• Money: a pre-order, a deposit, a paid pilot or a letter of intent.

If a meeting ends with “keep me posted” and no next step, it is a polite no, however warm it felt.

Running the conversation

1. Prepare three big questions. What do you most need to learn this week? Write those down first.
2. Keep it casual. A relaxed conversation over coffee often gets more honest answers than a formal “interview”.
3. Start broad, then narrow. Ask about their work and priorities before asking about the specific problem.
4. Follow the energy. When they talk about something with feeling, ask for the last example.
5. Ask about money and time already spent. Existing spend is the strongest sign a problem is real.
6. Close with a request. Ask for a next step that involves some commitment.

After the conversation

• Write notes within an hour, while details are fresh.
• Separate what they did (facts) from what they said they think (opinions).
• Note exact phrases they used; they are useful later for marketing copy.
• Share the notes with your team so the learning is not stuck in one head.
• Look for the same problem coming up across several people before acting on it.

Common traps

• Pitching too early. Save any demo for the end, if at all.
• Talking to friends only. They are the people least able to give a disappointing answer.
• Counting interviews instead of learning. Ten conversations that taught you nothing are worth less than two that changed your plan.
• Hearing what you hoped for. If every interview confirms your idea, check whether your questions are leading people there.

When to stop interviewing

You do not need hundreds of conversations. Stop a round when you keep hearing the same problems in the same words, and when you can predict what the next person will say. Then act on what you learned, such as a change of target customer, a simpler offer or a first paid pilot, and start a new round with new questions.