How many customer discovery interviews should you do?

Plan for 15 to 20 interviews with one narrow customer type. If the last five interviews teach you nothing new about the problem, stop. If answers still contradict each other, your segment is probably too broad.

How long should a customer discovery interview be?

Aim for 25 to 30 minutes. That is long enough to get past surface answers and short enough that busy people say yes. Reserve the last five minutes for asking about a next step.

Should you pitch your idea in a discovery interview?

No. A pitch turns the conversation into a review of your idea, and people will be kind. Learn how they handle the problem today, then show a solution only in a separate follow-up.

How do you find people to interview?

Start with people who recently dealt with the problem, since their memory is fresh. Use your network for introductions, niche communities, and cold outreach that asks for advice on a specific task rather than a sales call.

What counts as a good signal in customer discovery?

Commitment beats compliments. Good signals are time, money, or reputation put at risk: a paid workaround, a request for a follow-up, an introduction to a colleague, or a pre-order.

Customer Discovery Interview Questions That Get Honest Answers

The best customer discovery interview questions ask about the past, not the future. You ask what people did the last time they hit the problem, what it cost them, and what they tried, because behavior is hard to fake and opinions about your idea are easy to inflate. Pair that with a few rules that keep you from leading the witness, and you get real problems instead of polite praise.

Why people tell you what you want to hear

Most interviewees are not lying. They are being kind. When a founder describes an idea with visible excitement, the easiest social move is to agree. Saying "that sounds useful" costs nothing and ends the awkward moment.

There are three forces working against you:

Politeness. People do not want to hurt your feelings, especially if you are a friend, a friend of a friend, or someone who asked for a favor.
Hypotheticals. "Would you use this?" asks someone to predict their own behavior. People are poor at that, and they predict generously.
Your own bias. You want the idea to be good, so you hear encouragement in neutral answers. This is classic confirmation bias, and it is one of the cognitive biases that hurt business decisions most often in early-stage work.

The fix is structural. You design the conversation so that a polite answer is not possible, because you never ask a question that has a polite answer.

Six rules for avoiding leading questions

1. Ask about the last time, not in general

"How do you handle invoicing?" invites a tidy summary. "Tell me about the last time you sent an invoice" invites a story, and stories contain the mess: the workaround, the delay, the person they had to chase.

2. Do not mention your idea until the end, or ever

Once you describe a solution, the interviewee starts evaluating it. Everything after that point is contaminated. Keep the first 20 minutes entirely about their world.

3. Ban "would you" and "do you think"

Questions that start with "would you," "could you see yourself," or "do you think" ask for predictions and opinions. Replace them with "did you," "how did you," and "what happened when."

4. Never offer the answer inside the question

"Is scheduling a big pain point for you?" tells them what you hope to hear. "What was the most annoying part of your week last week?" lets them choose the topic. If scheduling does not come up unprompted, that tells you something.

5. Follow the emotion and the money

When someone sighs, laughs, or raises their voice, stop and ask about it. When they mention a cost, a tool, or a person they pay, dig into it. Pain that costs money or time is the only pain worth building for.

6. Treat compliments as noise

"That's a great idea" is not data. Respond with "Thanks. What are you doing about this today?" Then write down only what they say next.

Question set: weak versus strong

The table below shows the swaps that matter most. Print it and keep it next to you during calls.

Weak question	Why it fails	Strong replacement
Would you pay for a tool that does this?	Hypothetical, easy to say yes	What do you spend on this problem today, in money or hours?
Is this a big problem for you?	Leads with your assumption	What was the hardest part of your last week?
Do you like this feature?	Invites a polite opinion	Walk me through the last time you needed something like this.
How often do you have this problem?	They guess and round	When did it last happen? And before that?
What features would you want?	Customers design poor solutions	What have you already tried, and why did you stop?
Would you recommend this to others?	Social-proof fishing	Who else on your team deals with this, and how do they handle it?

Template: the 30-minute interview outline

Copy this outline into your notes. The order matters.

Opening (2 minutes)

"Thanks for the time. I'm not selling anything. I'm trying to understand how people in your role handle [area], and I'll mostly be listening."
"There are no right answers, and bad news is more useful to me than good news."

Context (5 minutes)

"Tell me about your role and what a typical week looks like."
"Who else is involved when [area] comes up?"

Specific story (10 minutes)

"Tell me about the last time you [did the task or hit the problem]."
"What happened first? And then?"
"What was the hardest part?"
"How long did that take, and who had to get involved?"

Current solutions (6 minutes)

"What do you use today to deal with that?"
"What have you tried that did not work?"
"How much do you spend on it, counting tools, contractors, and your own time?"
"If it disappeared tomorrow, what would change?"

Priority check (3 minutes)

"Where does this rank among the problems you're dealing with this quarter?"
"What is stopping you from fixing it already?"

Close (4 minutes)

"Is there anything I should have asked and didn't?"
"Who else should I talk to?"
"Can I follow up in a few weeks if I have something to show?" (Note whether they agree, hesitate, or offer a specific time.)

Step-by-step: running the whole process

Pick one narrow segment. Mixing customer types makes every answer ambiguous. If you have not defined who you are targeting, use a guide on how to create an ideal customer profile first.
Recruit 15 to 20 people who recently had the problem. Recent experience beats imagination. Ask for advice, not a sales demo, and offer 25 minutes.
Write your three learning goals. For example: how often does the problem happen, what does it cost, and what do people do about it now. If a question does not serve a goal, cut it.
Record or take verbatim notes. Summaries drift toward what you hoped to hear. Quote exact phrases and mark them "said" or "did."
Debrief within an hour. Write three facts you learned, one surprise, and one thing you are tempted to over-read.
Review after every five interviews. Look for repeated behavior, not repeated opinions. Adjust the script if a question keeps producing vague answers.
Test commitment. In the final stage, ask for something small and real: a follow-up call, an intro, a signed letter of intent, or a deposit.

Worked example: a scheduling tool that was solving the wrong problem

A three-person startup was building scheduling software for physical therapy clinics. The founders believed no-shows were the core pain. Their first 8 interviews used questions like "Would a smarter reminder system help you?" and "Is no-show management a big issue?"

The results looked wonderful: 7 of 8 said yes, several called it "a no-brainer." But when the founders asked for a paid pilot at $149 a month, none of the 7 committed. Two never replied.

The team rewrote the script around the template above and ran 12 more interviews over three weeks, all with clinics of 2 to 5 therapists. They asked for the last bad week, not opinions on reminders. Here is what surfaced:

9 of 12 mentioned no-shows, but 8 of those already used a free text-reminder feature in their existing software and considered it good enough.
7 of 12 raised, unprompted, claim denials caused by insurance eligibility errors. One clinic with roughly 400 claims a month saw about 11% denied, which meant around 44 claims reworked at 30 minutes each. That is about 22 hours of front-desk time monthly, plus payment delays of several weeks.
5 of those 7 had hired a part-time billing contractor at roughly $1,800 a month to clean it up.

The no-show problem had a free fix. The denial problem had a paid one, which proved people already treated it as worth money. The team pivoted to eligibility verification at intake, and within six weeks 4 of the 12 clinics had agreed to a $249 monthly pilot. The signal was not the compliments in the first round. It was the contractor invoices in the second.

The most common mistake: pitching in disguise

Founders think they are asking neutral questions when they are actually describing their product in the form of a question. "Wouldn't it be great if your calendar could auto-resolve conflicts?" is a pitch with a question mark.

How to avoid it:

Write your questions in advance and check each one for a hidden solution.
Remove adjectives like "easy," "smart," "better," and "automatic" from every question.
Ask a teammate to listen to one recording and flag any moment where you steered.
If you catch yourself explaining the product, stop and ask, "What are you doing about this now?"

A second common mistake is counting interviews instead of insights. Twenty friendly chats that end in compliments are worth less than five uncomfortable ones that end in a specific cost and a specific workaround.

How to read what you heard

After 15 to 20 interviews, sort your notes into two columns: what people said, and what people did. Weight the second column heavily.

Strong evidence of a real problem looks like this:

They describe a specific recent incident without prompting.
They have tried at least one workaround, paid or unpaid.
They can state a cost in dollars, hours, or lost customers.
They bring up the problem before you do.
They agree to a concrete next step.

Weak evidence looks like praise, vague frustration, or interest that disappears when you ask for time or money.

If you want to go deeper on why something matters, the 5 Whys technique works well on your interview notes. And once you have paying customers, a win/loss analysis applies the same behavior-first logic to deals you won and lost.

Do not skip the follow-up. The founders in the example above learned more from the second round of interviews because the first round exposed the flaw in their script. Expect your first five calls to be rough, and treat them as practice.

Key takeaways

Ask about specific past events, not future intentions. "Tell me about the last time" beats "would you ever."
Keep your idea out of the room until you understand the problem, and treat compliments as noise.
Replace opinion questions with behavior questions: what they tried, what it cost, and who else is involved.
Interview 15 to 20 people from one narrow segment, and review your script after every five calls.
Judge the evidence by commitment: money, time, or reputation put at risk, not enthusiasm.
Check your own questions for hidden pitches, because that is the mistake nearly every founder makes first.