Generator · customer discovery interview questions
Customer Discovery Interview Question Generator
Pick a stage and audience, get a curated list of Mom-Test-aligned interview questions designed to surface real behavior — not polite opinions. Free, deterministic, no AI.
Three Mom Test rules
- Talk about their life, not your idea. Pitching contaminates the conversation.
- Ask about specific past behavior, not hypothetical futures. “Tell me about the last time…” beats “would you…”.
- Listen more than you talk. Aim for 80% listening. Compliments are not data.
Problem discovery · B2B small business
10 questions to bring to your next conversation
Establish whether the problem exists, how often it happens, and what people already do about it. Stay out of solution-talk.
- 01
What is the hardest part of dealing with this? Why is that hard?
Open
- 02
Tell me about a time it caused you to miss something important.
Past behavior
- 03
Who else have you talked to about this in the last month?
Past behavior
- 04
What metric or signal tells you it has gotten worse?
Open
- 05
Who else cares when this goes wrong? Who has skin in the game?
Open
- 06
What did you do once you noticed it? Step by step.
Past behavior
- 07
What workarounds have you built up over time? Show me one if you can.
Past behavior
- 08
How often does this come up in a typical week? Tell me about the last three times.
Past behavior
- 09
When this happens at work, whose problem is it — yours, your team’s, or your manager’s?
Open
- 10
How do you describe this problem when you complain about it to a friend?
Open
Same inputs always produce the same questions in the same order — share this page URL with your team for a reproducible interview list.
Methodology
How this list is generated
A curated bank of Mom-Test-aligned questions, filtered by stage and audience, ordered deterministically.
The generator uses a hand-curated bank of customer-discovery questions, each tagged with the discovery stage it belongs to and the audiences it applies to. There is no LLM and no external API call — picking a stage and audience filters the bank, and a deterministic seeded shuffle (the seed is built from your inputs) orders the result.
Because the shuffle is seeded, the same inputs always produce the same list in the same order. That means you can share the URL with a teammate and you will both interview from exactly the same script.
Every question follows three Mom-Test rules: talk about their life rather than your idea, ask about specific past behavior rather than hypothetical futures, and listen more than you talk. We deliberately avoid “would you” framings, avoid pitching, and avoid leading questions.
Framework
The five stages of customer discovery
- Problem discovery
- Establish whether the problem exists, how often it happens, and what people already do about it. Stay out of solution-talk.
- Solution discovery
- Once the problem is real, find out what a good fix looks like, what would block adoption, and what people would happily switch from.
- Willingness to pay
- Pressure-test economic intent. Who pays today, what for, who approves, and what would have to be true for them to pay you.
- Pricing
- Surface price anchors and packaging preferences. Use specific numbers and past purchases — never abstract willingness questions.
- Channel discovery
- Learn how this customer actually finds and adopts new tools. Dig into the last purchase, not hypothetical channels.
Worked example
A realistic interview scenario
The setup
Founder testing an AI sales coach idea. Stage: problem discovery. Audience: B2B mid-market. The interview is with a Director of Sales at a 200-person SaaS company.
Three high-signal questions from the generator
- “Walk me through the last time this came up. What were you doing right before?”
- “How do you find out when it has happened? Who flags it first?”
- “What is the last thing you bought to help with this? What did it actually fix?”
What good signal looks like
The Director describes a specific Tuesday two weeks ago, names two reps, mentions Gong by name as the tool they bought last quarter, and complains that nobody actually uses Gong’s coaching features because reps find them generic. That is the gold: behavior, names, brands, concrete dissatisfaction.
What weak signal looks like
“Yeah, that sounds like a real issue. We probably need better tools.” Compliments and abstractions. Push back with: “Tell me about the last specific time it cost you something.” If they cannot name a specific time, the pain is not as urgent as it sounded.
Pitfalls
Common traps in each stage
Problem discovery
Avoid pitching, avoid hypotheticals, and never accept "yeah, that sounds useful." Compliments are not data. Push back on every "I would" and ask for the last time it actually happened.
Solution discovery
People will happily describe ideal products that they would never buy. Cross-check every "I would use it" with "what stopped you from solving this last time?"
Willingness to pay
Verbal commitments are worthless. Look for behavioral signal: a deposit, a pilot agreement, or at minimum a pre-order. If they will not name a number or a person who would approve it, the answer is no.
Pricing
Asking "would you pay X" returns lies. Anchor on past purchases, ranges, and named alternatives. Use the Van Westendorp brackets (too cheap / cheap / expensive / too expensive) to triangulate.
Channel discovery
Avoid leading channel questions ("would you find us on Twitter?"). Ask only how they actually discovered the last tool they bought, not how they imagine they might discover yours.
Recommended next actions
What to do after the interviews
Write up within 24 hours
Memory decays fast. Within a day, write up each interview as a one-page summary: behavior observed, brands mentioned, specific times, direct quotes.
Score against the Idea Validation rubric
Use what you heard to update your Pain and Demand sliders in the Idea Validation Score Calculator. Re-run and compare against your pre-interview score.
Look for falsification, not validation
Five interviews that all loved the idea is more likely a problem with your questions than a real signal. Look for things that disconfirm your thesis.
Move to MVP Readiness
Once Pain and Demand have moved from guesses to defensible numbers, run the MVP Readiness Calculator to see what is left before any building.
FAQ
Frequently asked questions
What is the Mom Test?
The Mom Test is a customer-discovery method named after the observation that even your mother will compliment your idea. The way to get useful answers is to ask about specific past behavior, not hypothetical future intent. The questions in this generator are written to that standard — every one points at something the customer actually did, not something they might do.
How many interviews should I run?
Five is the floor for any single hypothesis. Ten is normal for a meaningful pattern. If at five you are still hearing wildly different things, your audience definition is too broad — narrow it before going further. If at ten you are hearing the same thing, you have signal; do not run twenty for false confidence.
Should I record interviews?
Always with permission. Recording lets you focus on the conversation instead of note-taking, and lets you re-listen for the parts you missed in the moment. Tools like Otter and Fireflies handle this for under $20 per month. If recording is not possible, take notes by hand and write up immediately afterwards while memory is fresh.
What if my customer is too polite?
Politeness shows up as compliments, "interesting", "I would totally use that". Treat all of these as zero signal. The question to bring them back to ground: "Tell me about the last time you actually solved this — what did you do?" Specific past behavior cannot be polite-faked.
How is this different from a survey?
Surveys force people to choose from your options; interviews let people tell you about their actual experience. Use surveys after interviews — once you understand what the customer is actually doing, surveys can quantify how common that pattern is. Surveys before interviews almost always confirm your priors and miss the real story.
Can I use these for B2B and B2C?
Yes. The audience filter narrows the bank to questions that fit each context — B2B-only questions cover org buying, approval chains, and budgets; B2C-only questions cover personal purchase decisions. Many questions are universal because the underlying mechanic of past-behavior interviewing applies to both.
How do I follow up after an interview?
Within 24 hours, send a short thank-you email with one sentence about what you took from the conversation. If they brought up a specific tool or person, ask if you could be introduced. Most useful insights come from the chains of warm intros, not the first conversation.
What if all my answers are positive?
Positive answers usually mean you are asking hypothetical or leading questions. Re-read the questions you used; replace every "would you" with "tell me about the last time you did". Re-interview if possible. The point of customer discovery is to look for falsification, not validation.
What is the Mom Test?
How many interviews should I run?
Should I record interviews?
What if my customer is too polite?
How is this different from a survey?
Can I use these for B2B and B2C?
How do I follow up after an interview?
What if all my answers are positive?
Continue your validation
Pair this with one of these
Get early access
Turn interview answers into a validation plan.
Kavon Studio reads interview notes and produces a falsifiable validation plan with hypotheses to test, a landing page, and a kill-or-continue metric.
- Convert interview notes into a structured validation plan.
- AI-generated landing pages and prototypes for each hypothesis.
- Track every test to a decision memo.