Customer Discovery Interviews: How Founders Get Honest Answers

Ask a friend whether they would use your startup idea and they will almost certainly say yes. They like you, they do not want to hurt your feelings, and saying yes costs them nothing. Founders walk away from these chats feeling validated, spend months building, and then discover that enthusiasm in a conversation is not the same as willingness to change behaviour or pay money.

Customer discovery interviews exist to fix exactly this. Done well, they replace flattering opinions with useful facts about how people actually behave, what they currently do about a problem, and what that problem costs them. This guide explains how to prepare, run, and learn from discovery interviews, using methods that work for B2B and consumer products alike. It builds on the ideas in our article on the cost of skipping product discovery, with a focus on the conversation itself.

What discovery interviews are (and are not)

A discovery interview is a structured conversation designed to learn about a person's world: their goals, workflows, frustrations, and the ways they currently solve a problem. It is not a sales call, a demo, a focus group, or a survey.

• It is not a pitch. If you spend the call describing your solution, you are learning nothing new.
• It is not a feature poll. People are poor at predicting what they will use, and good at describing what they did yesterday.
• It is not a one-off. The value comes from patterns across many conversations in the same segment.

The core principle: past behaviour beats future promises

The most reliable information in an interview is about things that already happened. Compare these questions:

• Weak: "Would you pay for an app that automates your invoicing?"
• Strong: "Tell me about the last time you had to chase a late invoice. What happened, step by step?"

The weak question asks for a prediction about a hypothetical. The strong question asks for a story, and stories contain facts: which tools were used, who was involved, how long it took, what went wrong, and whether the person ever tried to fix it. If someone has never tried to solve a problem, spent money on it, or even complained about it, that is useful evidence that it may not be painful enough to build a business around.

A real-world example: a founder building scheduling software for clinics

Imagine a hypothetical founder who believes small physiotherapy clinics need better appointment scheduling. Her first round of conversations goes badly: she describes her idea, and every clinic owner says "that sounds great". She has no idea what to build.

In round two she changes her approach. She asks each owner to walk her through the last week of bookings. She learns that the real pain is not the calendar at all. It is no-shows, and the awkward process of chasing patients by phone to confirm. Several owners mention they already use a shared WhatsApp number for reminders, which is clumsy but proves the need. One has paid a freelancer to build a spreadsheet tracker.

Those workarounds are gold. They show that the problem is real, that people already spend time and money on it, and that the language to use in marketing is "fewer empty slots", not "smarter scheduling". The founder's first version becomes an automated reminder and confirmation flow, a far narrower and more valuable product than the one she started with. This is the kind of outcome discovery should produce: a sharper problem, not just a confirmation of the original idea.

Step-by-step: how to run discovery interviews

1. Define your learning goals. Write down the three to five assumptions that would sink your idea if they were wrong. For example, "Clinics lose meaningful revenue to no-shows" or "Owners, not receptionists, make tooling decisions."
2. Choose a narrow segment. Interview people who share a context, such as "owners of independent clinics with two to five practitioners". Mixing very different customers muddies the patterns.
3. Recruit thoughtfully. Ask for warm introductions, post in relevant communities, and reach out directly. Offer a short, specific ask (twenty to thirty minutes) and be clear it is research, not a sales call. A small thank-you gift is fine if appropriate.
4. Prepare a short guide, not a script. Draft six to eight open questions around past behaviour, and leave room to follow interesting threads.
5. Open the conversation well. Explain the purpose, ask permission to record, and reassure them there are no wrong answers and that you are not selling anything.
6. Ask for stories. Use prompts like "Walk me through the last time...", "What happened next?", and "How did you handle that?"
7. Dig into specifics. Follow up with "Can you show me?", "How often does that happen?", "What did it cost you in time or money?", and "What have you tried so far?"
8. Listen more than you talk. Aim to speak for a small fraction of the call. Embrace silence, because people often add the most revealing detail after a pause.
9. Close with commitment signals. Ask for an introduction to someone else with the same problem, or whether they would be willing to join a pilot or a follow-up call. Real interest shows up in time, introductions, or money, not compliments.
10. Debrief immediately. Within an hour, write down the key facts, direct quotes, surprises, and what changed in your understanding.

Questions worth asking

To understand context

• "Tell me about your role and what a typical week looks like."
• "What are the most important outcomes you are measured on right now?"

To understand the problem

• "When did this last come up? What happened?"
• "What is the most frustrating part of that process?"
• "How often does it happen, and what does it cost you when it does?"

To understand current solutions

• "What do you use today to handle this?"
• "What have you tried and abandoned? Why?"
• "Have you ever paid for something to solve this?"

To understand priority and buying

• "Where does this rank among your top problems this quarter?"
• "Who else is involved when a tool like this is chosen?"
• "What would need to be true for you to switch from your current approach?"

Questions and habits to avoid

• Leading questions. "Don't you think it would be great if..." tells the person what answer you want.
• Hypotheticals. "Would you use..." and "How much would you pay..." produce unreliable answers.
• Pitching mid-interview. Save the idea for a later round, and only after you understand the problem.
• Treating compliments as data. "That's a cool idea" carries almost no information. Look for facts, commitments, and workarounds.
• Interviewing only friendly people. Seek out sceptics and people who chose a competitor. They teach you the most.

Turning interviews into decisions

After ten or fifteen conversations, patterns should emerge. Gather your notes and look for:

• Repeated problems mentioned independently by several people.
• Common workarounds, such as spreadsheets, manual follow-ups, or paid freelancers.
• Language: the exact words customers use to describe their pain, which you can later use in copy.
• Segments: groups who feel the problem more strongly than others, who may be your best early customers.
• Disconfirming evidence: interviews that contradict your assumptions. Do not explain these away.

Then decide what to do next. Common outcomes include narrowing the problem, changing the target customer, pivoting to a more painful problem that surfaced, or gaining enough confidence to build a small prototype. To test demand more cheaply than building, consider the approaches in our guide to AI-assisted market research for validating startup ideas, which complements interviews with desk research.

Using AI tools responsibly

Transcription tools save hours, and AI summarisation can help you cluster themes across many interviews. Use them as assistants, not oracles. Keep the original recordings, verify any quote you plan to use, and watch for models that smooth contradictory opinions into a tidy consensus. Remember also to handle participant data with care: get consent, store recordings securely, and remove personal details from shared notes.

From insight to prototype

Interviews tell you what problem to solve, not how to build the solution. Once you have a clear problem, a defined segment, and a few people who have asked to see what you build, you can scope a focused first version. Resist the urge to build everything you heard. Choose the one painful workflow that people already spend time on, and build the smallest product that improves it. If you need help scoping that first version, our SaaS development team works with founders to turn interview findings into a realistic build plan, and our app cost calculator gives a quick starting estimate.

Key benefits of disciplined discovery

• Less wasted build time. You learn that an idea is weak before investing heavily in it.
• A sharper product. You build around real workflows and pain points.
• Better messaging. Customers' own words become your headlines and sales scripts.
• Early advocates. People who help shape the product often become your first customers and referrers.
• Stronger fundraising stories. Investors respond to founders who can describe customer problems in concrete, evidence-backed detail.

Conclusion

Discovery interviews are cheap, fast, and uncomfortable in the right way. They force you to listen to evidence that may contradict your favourite idea. Focus on past behaviour instead of future promises, interview a narrow segment until patterns emerge, record and debrief carefully, and treat workarounds and commitments as stronger signals than compliments.

Block out a few hours this week, book your first five conversations, and ask for a story instead of an opinion. What you learn may save you months of building the wrong thing.

Frequently Asked Questions

How many customer discovery interviews should a founder do?

There is no magic number, but many founders aim for fifteen to thirty conversations within a single customer segment before drawing conclusions. The real signal is when new interviews stop surprising you and the same problems, words, and workarounds keep repeating.

What is the biggest mistake in discovery interviews?

Pitching your idea and asking whether people would use it. Hypothetical questions invite polite, unreliable answers. Ask instead about specific past behaviour, such as the last time the problem happened, what they did, and what it cost them.

Should I record the interviews?

Yes, with permission. Recording lets you stay present in the conversation instead of typing, and lets you review exact wording later. Always tell the participant that you are recording and how the notes will be used.

How do I find people to interview?

Start with your network and ask for introductions, then go where your target customers already gather: communities, forums, LinkedIn, industry events, and customer lists of adjacent products. Be clear that you are researching a problem and not selling anything.

Can AI tools help with customer discovery?

AI can help transcribe calls, cluster themes across interviews, and draft summaries. It should not replace the conversations themselves, and you should always check summaries against the original transcripts, since models can smooth over nuance or misattribute quotes.