Customer Discovery: How to Run Interviews That Teach You Something

Customer discovery is the work of talking to people who have the problem you want to solve, before you build, to learn how they deal with it today and whether they would pay to fix it. Hold at least 10 one-to-one conversations, ask about things people have already done, and look for workarounds that already cost them time or money.

I co-founded Hackhouse, a builders' hub in Nairobi. This guide sets out a simple interview method built on Steve Blank's customer development and Rob Fitzpatrick's book The Mom Test, with notes for founders selling in Kenya and across Africa.

What is customer discovery?

Customer discovery is the first of four steps in Steve Blank's customer development method. It turns the founder's vision into a set of guesses about the business, then tests those guesses with customers until they become facts. The method starts from the idea that there are no facts inside your building, so you have to go outside to test them (Wikipedia, Customer development). Blank's own reading list points beginners to Talking to Humans by Giff Constable as a simple primer on customer discovery (Steve Blank).

Kenya has a well-documented example of a product learning from its first users. M-Pesa was first designed so microfinance borrowers could receive and repay loans through Safaricom's airtime resellers. The pilot that began in October 2005 mainly handled loan repayments for Faulu customers, and it also tested airtime purchase and sending money within Kenya. The full commercial launch followed in March 2007, as a payment and money transfer service (Wikipedia, M-Pesa). The lesson for founders: watch what early users do with your product, and let that shape what you build next.

Who should I interview for customer discovery?

Talk to people who have the problem today and already spend effort on it. Your family, friends and fellow founders will be kind about your idea, and kindness tells you nothing about demand.

• Write one sentence first. For example: "Shop owners in Eastlands lose stock because they cannot track daily sales." If you cannot name who has the problem and what it is, you are not ready to interview.
• Go where those people already are. As examples: a matatu SACCO office for transport operators, shops on a slow weekday morning for retailers, LinkedIn or industry events for business buyers.
• Aim for at least 10 conversations before you draw conclusions. A pattern rarely shows in two or three.
• Ask every person for a referral. "Who else do you know who struggles with this?" reaches people a survey link never will.

What questions should I ask in a customer interview?

The Mom Test is subtitled "How to talk to customers and learn if your business is a good idea when everyone is lying to you" (The Mom Test). People are polite. Ask them whether they would use your product and most will say yes, so ask about things that have already happened. These questions work well:

1. "Tell me about the last time this problem came up."
2. "What did you do about it?"
3. "How much time or money did that cost you?"
4. "What have you already tried? Why did you stop?"
5. "Have you ever paid for something to fix this?"

Leave out "Would you use this?", "Would you pay for this?" and "Don't you think this would help?" Each one invites a guess or writes the answer for the other person. If someone has never tried to solve the problem, treat that as a sign the pain is mild.

How do I run the interview itself?

• Leave the pitch at home. No slides and no demo. Once you start selling, the person starts being polite.
• Bring a teammate if you can. One of you asks questions and the other writes everything down.
• Ask before you record. Kenya's Data Protection Act 2019 describes consent as an express, unequivocal, free, specific and informed indication of a person's wishes (CIPIT, Strathmore University). Tell people what you will do with the recording, and keep personal data no longer than you need it (DLA Piper).
• Ask to see the task. "Can you show me how you record today's sales?" What people do in front of you is better evidence than what they describe.
• Talk for less than a third of the time. Ask, then wait. The useful answer often comes after a pause.
• Listen for the workaround. A sales book, a WhatsApp group or a spreadsheet shows you the problem exists, that the person already spends effort on it, and what you are competing with.

How do I make sense of what I heard?

Write up each interview within 24 hours, while you still remember tone and detail. For each one, note the problem in the person's own words, what they do about it now, what it costs them, and anything that surprised you. A shared spreadsheet with one row per interview is enough.

Then put the interviews side by side and look for what repeats across at least half of them. Three outcomes are common:

• Strong, repeated pain with active workarounds. Build the smallest test you can and see whether people use it or pay for it. Our guide on validating a startup idea before writing code covers the next steps.
• Mild interest and no workarounds. The problem is too small for this group. Change the customer segment or the problem.
• One enthusiastic person and nine lukewarm ones. Follow the pattern from the nine.

Keep interviewing after you launch. Customers change, and the questions above work just as well on paying users. When retention and referrals start to pick up, read our piece on the signs of product-market fit.

What mistakes should I avoid?

• Pitching during the interview.
• Counting compliments as validation. A compliment costs the person nothing.
• Stopping after three conversations.
• Dismissing a workaround because it makes your idea look less original.
• Forgetting context such as data costs, shared phones or cash habits that shape what your customer can use.

Where can I practise customer discovery with other founders?

The Spark track of the Hackhouse Residency is for idea-stage founders: 12 weeks, KES 25,000 and no equity, with HH Gen 27.1 applications open now. You can also meet other builders at Hackhouse events and swap interview notes.

Key facts

• Customer discovery is the first step of Steve Blank's four-step customer development method.
• Ask about past behaviour and past spending, and avoid "would you" questions.
• Plan at least 10 interviews before drawing conclusions.
• Kenya's Data Protection Act 2019 requires consent that is express, free, specific and informed, so ask before you record.
• M-Pesa's pilot began in October 2005 and its full commercial launch came in March 2007.

Frequently asked questions

How many customer discovery interviews do I need?

Start with at least 10 conversations with people who have the problem. Keep going until new interviews stop telling you anything new about the problem or the workaround.

Can I do customer discovery over the phone or WhatsApp?

Yes. Phone and video calls work, and they reach people outside your city. In person is better when you want to watch someone do the task, so mix both if you can.

Should I pay people for customer interviews?

A small thank-you such as airtime or a cup of tea is fine for busy people. Large payments can attract people who say what they think you want to hear, so keep it modest.

What is the difference between customer discovery and market research?

Market research often measures the size of a market with surveys and reports. Customer discovery tests your specific guesses about a problem through direct conversations with the people who have it.

Sources

• Customer development, Wikipedia, checked 5 Nov 2026
• Books for Startups, Steve Blank, checked 5 Nov 2026
• M-Pesa, Wikipedia, checked 5 Nov 2026
• The Mom Test, Rob Fitzpatrick, checked 5 Nov 2026
• An Overview of the Kenyan Law on Commercial Use of Personal Data, CIPIT, Strathmore University, checked 5 Nov 2026
• Data protection laws in Kenya, DLA Piper, checked 5 Nov 2026

Go deeper with us. Join the Hackhouse community for conversations that go beyond the surface, where builders share the hard-won lessons that never make it into press releases.