What is a customer discovery interview?

A short conversation with someone who might have the problem your venture addresses, held to learn how they deal with it today. It is not a sales call and not a product demo. The aim is to test an assumption about the customer or the problem against what people have actually done.

How many customer discovery interviews should I do?

Enough to see a pattern within one clearly defined type of customer, and few enough that you finish before your next decision. A practical approach is to start with a batch you can complete in two or three weeks, debrief it, then run a second batch with sharper questions. If every interview sounds different, your customer definition is probably too broad.

Should I show my product or prototype in a discovery interview?

Not in the first part of the conversation. Once people see your solution, they start reacting to it and stop describing their own situation. Learn how they handle the problem today first. If you want reactions to a prototype, ask for that at the end or in a separate meeting.

Can I interview friends and family?

Only if they genuinely match your target customer, and even then treat their answers with care, because people who like you tend to be kind about your idea. Use them to practice your questions. For evidence, look for strangers who have the problem.

What counts as evidence in a customer interview?

What people have already done, spent or used: the last time the problem happened, what they tried, what it cost them in money or hours. Compliments, opinions of your idea and promises about the future are weak evidence. Concrete offers, such as a second meeting, data, an introduction or a paid pilot, are stronger.

Customer Discovery Interviews: How to Learn What Customers Really Do

A customer discovery interview is a short, structured conversation with someone who might have the problem your venture addresses.

It is held to learn what they actually do about it today, rather than what they think of your idea.

It is one of the cheapest ways an early venture can test its riskiest assumption.

You can do it before spending months building on that assumption.

This guide covers:

who to talk to
how to ask questions that produce evidence
how to take notes
what to do with the results
how to use the interviews with your mentor

The approach draws on two practice frameworks.

Steve Blank’s customer development starts with customer discovery. That means testing business hypotheses with real customers before building and scaling.

Rob Fitzpatrick’s The Mom Test adds the most useful single rule for the conversations themselves.

Ask about past behavior and real problems, not opinions of your idea.

What are you trying to learn from customer discovery interviews?

You are trying to find out whether a specific assumption is true.

Before you book a single conversation, write the assumption down.

Good assumptions are concrete enough to be wrong:

“Owners of independent dental clinics lose enough revenue to missed appointments that they actively look for fixes.”
“Parents of secondary school students already pay for tutoring and find tutors mainly through word of mouth.”
“Small manufacturers track machine downtime on paper and would share that data to get a better picture.”

Each of these points to things you can ask about.

You can ask how often the problem happens, what people do about it, what it costs them and how they buy.

“People want a better way to manage appointments” points to nothing.

Anyone will agree with it, and agreement teaches you little.

A useful habit is to rank your assumptions by two things.

One is how much the venture depends on each one. The other is how little evidence you have for it.

Interview for the one that scores high on both.

Who should you interview?

Interview people who match a clear definition of your target customer.

They should have the problem now or have had it recently.

The tighter the definition, the faster you see a pattern.

Define the type, not the market. “Clinic owners with one to three practitioners who handle their own scheduling” is a type. “Healthcare” is a market.
Separate the user from the buyer. In many businesses the person who has the problem does not sign the purchase. You may need to interview both.
Find them where they already gather. Professional associations, online communities, trade events, former colleagues, your mentor’s network. A short, honest message works. Say you are researching how people in their role deal with a specific problem. Say you are not selling anything, and that you would value twenty minutes.
Be careful with friendly sources. Friends, family and fans are kind about ideas. Use them to rehearse, then go to strangers.

Keep the interviews within one customer type until you have a pattern.

If every conversation sounds different, narrow the definition before you add more interviews.

Which questions produce evidence rather than compliments?

Questions about the past do.

People are poor predictors of what they would do.

They can give concrete answers about what they did last week.

The core of a discovery interview is getting them to tell you about specific, recent events.

A question set for a 20 to 30 minute interview:

“Tell me about your role and a normal week.” (Context. Keep it short.)
“When did [the problem] last happen? Walk me through what happened.”
“What did you do about it?”
“What have you tried before to deal with this? What happened with that?”
“What does it cost you, in money, time or anything else, when this happens?”
“Who else is involved when this comes up? Who decides what to do about it?”
“Is solving this a priority for you right now, compared with other things on your list?”
“Is there anyone else I should talk to about this?”

Rewrites of common weak questions:

“Would you use an app that does X?” becomes “How do you handle X today?”
“How much would you pay for this?” becomes “What do you spend on this now, and on what?”
“Do you think this is a good idea?” becomes “What is the hardest part of dealing with this?”
“Would you buy it if it existed?” becomes “Have you looked for a solution? What did you find?”

When someone gives a general answer (“we always have trouble with that”), ask for the most recent example.

When someone praises your idea, thank them and steer back to their own situation.

Praise is pleasant and almost free to give.

That is why it carries so little information.

How should you take notes during the interview?

Write down what people did and said, in their words.

Keep it separate from your interpretation.

Within an hour of each interview, write a short summary:

Who: role and type of organization, no names needed in shared notes.
Situation: how the problem shows up for them, and the last time it happened.
Current solution: what they use or do today, including workarounds and doing nothing.
Cost: money, hours, stress, lost customers, in their own estimate.
Quotes: two or three exact phrases.
Signals: anything they offered or did, such as a follow-up, data, an introduction or a pilot.

Then sort your notes into three groups:

what people did, spent or use
opinions and compliments
promises about the future

Only the first group is evidence.

The second and third may still be interesting, but do not let them carry a decision.

If you can, bring a second person to take notes while you listen.

If you record, ask permission first and still write the summary the same day.

What do customer discovery interviews look like in practice?

Three illustrative situations, composites rather than case studies, each with a different lesson.

A software founder serving small clinics.

Her assumption was that missed appointments cost clinics so much that they would pay monthly for automated reminders.

In eight interviews with clinic owners, everyone agreed missed appointments were annoying.

But when she asked about the last missed appointment and what they did, most described a receptionist making a phone call.

None had looked for a paid solution in the past year.

Two owners, both running clinics with several practitioners, had tried and abandoned a tool.

They described the cost in detail.

Her verdict: the assumption was weak for the smallest clinics.

It was worth testing further for multi-practitioner clinics, and her next batch focused on those.

A consumer app for booking tutors.

The founder’s friends loved the idea.

When he interviewed parents he did not know, he asked how they had found their current tutor.

Almost every answer involved a recommendation from another parent or a teacher.

Several parents said they did not trust online reviews for something this personal.

His riskiest assumption changed from “parents want an app” to “parents will trust a recommendation that arrives through an app”.

That was a different and more useful question.

He would not have found it by showing the app.

A university team with a manufacturing idea.

The team planned a sensor product to track machine downtime.

Their first interviews were with professors and other students.

Those produced enthusiasm and no evidence.

Their mentor asked who actually decides on equipment spending in a small factory.

The mentor then introduced them to two plant managers.

Those two conversations showed that downtime was already tracked, on paper or in spreadsheets.

The real pain was getting the data out of the machines’ existing controllers.

The team kept the problem and dropped the sensor.

In each case the useful finding came from asking about something that had already happened.

It also came from talking to people who matched the customer definition.

What should you do after the interviews?

Give your riskiest assumption a verdict: supported, weakened, unclear, or replaced by a different assumption.

Write the verdict in one sentence, with the evidence that produced it.

Supported: look for stronger signals next. Will people give you time, data, an introduction, or money for a pilot?
Weakened: this is a cheap and early result. Ask whether a narrower customer type or a different problem in the same area looks stronger.
Unclear: usually means the customer type was too broad or the questions were about opinions. Tighten both and run another small batch.
Replaced: start again with the new assumption. Use what you learned to define the customer more precisely.

Then decide what to test next and by when.

Discovery is not one round of conversations.

It is worth keeping a habit of a few customer conversations every month, long after the first batch.

How do you use customer discovery interviews with your mentor?

A mentor who has built or run a business is especially useful at three points.

Before the interviews.

Bring your assumption list and your question draft.

Ask your mentor to play a target customer while you practice.

Ask them to stop you each time you ask about the future or ask for an opinion.

Ask whether anyone in their network matches your customer type and would talk honestly.

During the interviews.

Send a short note after the first two or three.

Early course corrections are cheap:

a question that is not working
a customer type that is wrong
a pattern you are already seeing

After the interviews.

This is where a mentor earns their place.

Étienne St-Jean’s research on mentors of novice entrepreneurs lists confrontation among the nine mentor functions it identifies.

In practice, that means questioning conclusions the founder is attached to.

Bring your notes sorted into the three groups.

Say what you hoped to hear before you started. Then ask your mentor:

Which of my conclusions are not supported by these notes?
What pattern do you see that I may be missing?
Is my sample too narrow, too friendly, or the wrong people?
What would you test next with what I now know?

Your mentor cannot validate the idea for you, and they should not decide whether you continue.

They can help you read the evidence honestly and plan the next test.

For preparing the debrief itself, see how to prepare for a mentoring meeting.

For more questions to bring, see questions to ask a mentor.

If you run a program for founders, designing a startup mentoring program covers how to build sessions like this into the structure.

Sources

Blank, S. (2005). The Four Steps to the Epiphany. Self-published (some printings dated 2003). Practice framework.
Fitzpatrick, R. (2013). The Mom Test. CreateSpace. Practice framework.
St-Jean, É. (2011). Mentor functions for novice entrepreneurs. Academy of Entrepreneurship Journal, 17(1), 65-84.