Customer Discovery Interview Questions That Work
Short answer
Good customer discovery interview questions ask about specific things that already happened: when the problem last occurred, what the person did about it, and what it cost them. Avoid questions about your idea, about future intentions, and anything answerable with yes or no. Close by asking for a small commitment such as an introduction.
Founders tend to treat customer interviews as a chance to get feedback. That framing is the problem. Feedback is about your idea, and the moment your idea is on the table, people start being kind to you.
Customer discovery works the other way round. You want stories and numbers from the other person’s recent past. The questions below are built to get those and nothing else.
Three types of question, and one you should never ask
Every useful question in a discovery interview falls into one of three buckets. Context questions tell you who the person is and how their week runs. Incident questions reconstruct the last time the problem happened. Spend questions find out what the problem already costs in time and money.
The question you should never ask is any version of “Would you use this?” It invites a guess about the future, and people are generous guessers. Rob Fitzpatrick’s book The Mom Test is named after this trap: ask questions that even your mum could not answer with a white lie, because they are not about your idea at all.
A 16-question customer discovery script
You will not use all sixteen in one call. Pick eight to ten and let the conversation decide the order.
Context
- Walk me through a typical week in your role.
- What are you measured on?
- Which tools do you open every single day?
The last incident
- When did [the problem] last happen? Talk me through it.
- What did you do next?
- Who else got pulled in?
- How did it end?
Spend and workarounds
- What have you tried to fix this? What happened?
- What are you paying for that today, including spreadsheets, freelancers or extra staff time?
- Have you searched for a better option? Where did you look?
- Why did you not switch?
Buying
- Who signs off on a purchase like that?
- How did the last similar purchase go, from first look to payment?
Close
- What should I have asked that I did not?
- Who else deals with this that I should talk to?
- Can I come back in a few weeks when I have something to show?
Question 9 does more work than the rest combined. Any team that already pays for a clumsy workaround has a budget line. Someone who has never spent a penny or an hour on the problem is unlikely to start because your product exists.
Rewriting the questions founders usually ask
| Criterion | Why it fails | Ask this instead |
|---|---|---|
| Would you use an app that…? | future guess, politeness | When did this last happen to you? |
| How much would you pay? | number pulled from thin air | What do you spend on this today? |
| Do you like the idea? | invites compliments | What have you already tried? |
| Would feature X help? | you supplied the answer | What annoys you most about the current process? |
| Do you often struggle with Y? | yes/no, leading | Tell me about the last time Y went wrong. |
Price deserves a note of its own. You cannot measure willingness to pay by asking about it. You measure it with a real offer, for example a priced smoke test landing page. Inside the interview, knowing current spend is enough.
Running the call: example with a property manager
Say you are exploring software for small residential lettings agencies in Manchester. Open the call with two sentences: you are researching how agencies handle maintenance requests, and you are not selling anything. Then you ask permission to take notes.
From there, context. How many properties does the agency manage? Who answers tenant calls? What gets logged where? You are listening for names of tools and for frustration in the voice.
Then the incident. “Tell me about the last boiler breakdown.” Follow the story with short prompts: and then?, who did that?, how long did it take? If the answer involves three phone calls, a WhatsApp group and a contractor who never showed up, you have something worth writing down word for word.
Plan for roughly a quarter of the airtime being yours. If you notice you have been talking for two minutes straight, you are pitching.
Video call or in person?
Video calls are fine for most B2B interviews and far easier to book. Meet in person when the problem lives in a physical place: a warehouse, a clinic, a shop floor. Watching someone do the task often reveals workarounds they would never think to mention, like the sticky note on the monitor that holds the real process, or the paper list that duplicates the software everyone claims to use.
Scoring what you heard
Write up each call within the hour, and separate what people did from what they said. Four headings are enough:
- Facts about past behaviour, with dates or frequencies.
- Problems in the person’s own words, quoted.
- Money or time already spent on workarounds.
- Commitments made at the end: intro, follow-up, pilot.
Compliments and “I would definitely use that” get no column. They are noise, however pleasant.
After ten interviews, look for repetition. Which problems came up unprompted more than once? Where has money already changed hands? If nothing repeats, your segment is probably too broad. Narrow it before booking more calls of the same kind.
Finding people to talk to
Cold outreach works better than most founders expect, provided the message is short. Who you are, which problem you are trying to understand, that you are not selling, and a request for 20 minutes. LinkedIn, trade associations, niche Slack communities and subreddits where the problem is discussed are all fair game.
Every call ends with question 15. Referrals are the cheapest source of further interviews, and they double as a test. If someone cannot name a single other person with the same problem, it may be rarer than they made it sound.
These interviews are one stage of a wider process, which our guide on how to validate a business idea lays out step by step. Once a clear pattern emerges, the next useful question is how big the pool of people with that problem really is, and the walkthrough on calculating TAM, SAM and SOM shows how to estimate it from the bottom up.
Good interview notes pay off later too. When you eventually brief a development team, whether in-house or a studio such as OnlineBuilders, real customer quotes beat a feature list every time.
Your next step: pick ten questions from the script, delete any that someone could answer without remembering a real moment, and send five interview requests today.
Questions people ask
Can I record customer interviews?
Ask first, every time. Some US states, California among them, require consent from everyone on the call, and under UK and EU GDPR a recording is personal data that needs a lawful basis and a deletion plan. Saying what the recording is for and when you will delete it usually gets a yes.
Should I pay people for customer interviews?
For consumers, a small voucher is common and helps recruitment. For B2B it tends to change the tone of the conversation, so most founders skip it. An honest reason to talk, and a promise not to pitch, matters more.
How many customer discovery interviews are enough?
Around 20 to 30 per segment is a sensible target. Griffin and Hauser's 1993 study found that this range surfaced roughly 90 percent of customer needs in the markets they examined.
Should I show my prototype during the interview?
Not in a problem interview. Once you show a solution, people start reviewing it instead of describing their lives. Keep demos for a separate, later round.
.jpg){kind=link}