Customer discovery is talking to customers to test a hypothesis about their problem before you spend money on a product, a new range or advertising. You do not ask whether people like your idea; you ask how they already deal with the problem: what they did last time, how much they paid, what did not work. Steve Blank described the method as customer development; Rob Fitzpatrick set out the rules for the questions in The Mom Test.
Most customer interviews produce a false result because of the questions: "Would you buy this?" and "What do you think of the idea?" collect polite compliments, not facts. Below: where the method comes from, how a bad question differs from a good one, how to run the interviews in four steps, a ready question guide for an online store and how to score the results.
When people search for customer discovery
Usually just before a decision that costs money: launching a new category, opening a store in a new niche, changing shipping or payment options, rebuilding the website. Someone advised "do customer discovery first", and the person searches for what it means in plain words, how to run it, which questions to ask and how it differs from a survey.
The key one is which questions to ask. That is where customer interviews break most often: the conversation happened, but there is nothing in it to test the hypothesis against.
What customer development is and where it comes from
Customer development is an approach in which a new business first tests its assumptions in conversations with customers and only then builds the product and the sales process. Steve Blank proposed the term and the method in his book "The Four Steps to the Epiphany"; his main point is that there are no facts inside your office, so get out of the building and talk to customers.
Blank divides the path into four steps:
- Customer discovery. You write down hypotheses: who the customer is, what problem they have, how they solve it now, and test them in conversations.
- Customer validation. You check that you can sell repeatedly: people actually buy, not just agree. If it does not hold, you go back to step one.
- Customer creation. You scale demand: advertising, sales channels.
- Company building. You rebuild the team and processes around the proven model.
What people usually mean by customer interviews is mostly the first step: conversations to test hypotheses. What sets them apart from an in-depth interview in market research is the goal: customer discovery tests a specific assumption and ends with a decision to do it, not do it, or change it.
Which questions to ask: the rules of The Mom Test
Ask about specific things that happened in the past, not opinions about the future. This is the first rule in Rob Fitzpatrick's book "The Mom Test"; the name means a question should be one that even your mom, who wants to support you, cannot lie in answer to.
There are three rules:
- Talk about their life, not your idea. As soon as you describe the idea, the other person starts rating and praising it.
- Ask about specifics in the past. "What did you do last time" gives you a fact; "would you" gives you a forecast that costs nothing.
- Talk less, listen more. The pause after an answer often brings the most valuable part.
| Instead of | Ask | Why |
|---|---|---|
| "Would you buy from us if shipping were free?" | "When did you last give up on an online purchase? What happened?" | You see whether shipping is really the reason |
| "What do you think of our website?" | "Show me how you looked for this product last time" | You see the real path, not a rating |
| "How much would you pay?" | "How much did you pay last time, and where?" | A price from a receipt, not from imagination |
| "Is this important to you?" | "What have you already tried to solve it?" | People have already tried to solve problems that matter |
Compliments ("great idea!") and promises ("I'd definitely buy it") are noise. The signal is facts: money and time already spent, workarounds the person has already invented, and willingness to take a next step: leave a contact, place a pre-order, introduce a colleague.
How to run customer interviews in four steps
Customer discovery has four steps: state a hypothesis, find people to talk to, hold the conversations, draw conclusions.
- Hypothesis. One sentence that can be proven wrong: "Parents return children's shoes because of the size, since they cannot measure the child's foot at home." Not "people need a convenient store".
- Who to talk to. People who have the problem now: your customers from recent months, those who returned an item or abandoned a cart, your competitors' customers. Not friends and relatives: they will be polite.
- Interview. A call or meeting of up to half an hour, recorded with permission. Keep the question guide in front of you but do not read it out in order: follow the answers and ask "and what happened next?". Do not pitch your idea until the end, and preferably not at all.
- Conclusions. Right after the conversation, write down facts, not impressions: what happened, what it cost, how the person solved the problem. Compare the conversations with each other, not with what you hoped to hear.
The method does not set a number of interviews. A practical rule: keep going until new conversations stop bringing anything new. If the answers keep diverging after several interviews, you may be talking to different customer groups, and it is worth separating them.
A customer interview question guide for an online store
The example is the shoe-returns hypothesis from step one. The questions fit most hypotheses about buying: replace the product and the situation with your own.
- Tell me how you bought [product] last time. How did it start?
- Where did you look? Which options did you compare, and on what?
- Why did you pick that particular seller?
- What was the hardest or most annoying part?
- Have you returned anything in the past year? How did it go?
- What did you do to get the size right? How long did it take?
- Who else was involved in the decision?
- What should I have asked but didn't?
The last question often produces the most unexpected fact. If the person starts describing a solution ("it would be great if…"), do not agree and do not argue; ask "and how do you handle it now?"
How to score customer discovery interviews
Put the conversations into a table: one conversation per row, with columns for the facts that relate to the hypothesis. Count the result by facts, not by how many people said "yes".
Illustrative example for the arithmetic; the numbers are not results of a real study. You talked to 10 buyers of children's shoes:
- 7 of 10 had returned shoes because of the size, and 5 of them had already measured the foot at home with a ruler or against the insole: the problem exists, and people are already trying to solve it;
- 2 of 10 buy only in physical stores because they want to try shoes on: a different group with a different problem;
- 1 had never returned anything: the hypothesis does not apply to them.
Conclusion: the hypothesis held for most people, so the next step is not another round of interviews but a test by action (customer validation in Blank's terms). For example, add a size table with insole length in centimetres and inches to the product pages and compare the share of size-related returns a month later. You can build one with the shoe and clothing size chart.
Customer discovery does not replace numbers: it explains why people behave the way they do. Whether a change pays off and what it costs to win a customer are calculated separately; see customer acquisition cost: formula and example. And what customers tell you about service is worth turning into rules for working with them; see ecommerce customer service rules.
Mistakes that make customer interviews useless
- Pitching the idea at the start. From then on, the person is assessing you instead of recalling their experience.
- Asking about the future. "Would you buy" is the cheapest answer in the world; it commits to nothing.
- Talking to people who want to support you. Friends, family, loyal customers who are afraid to offend you.
- Selling during the conversation. The moment you start persuading, the other person starts defending.
- Writing down impressions instead of facts. "They liked it" cannot be compared across conversations; "returned something three times this year" can.
Where to find people to interview when you already have a store
In your own customer base: people who buy regularly, people who stopped buying, and people who returned items. Before the call, know what the person bought and when; then "tell me about your last purchase" becomes a specific question.
In Elbuz, customers and orders sit in one database: the customer stage (lead, regular, dormant) is calculated from purchases, and the customer record shows the full order history. So a list for your interviews, for example dormant customers who bought children's shoes, comes from a filter rather than from memory. More on the ecommerce CRM page.
Who it will not help: if you are not selling yet and are testing a niche from scratch, your customer base is still empty; look for people to interview in topic communities and among your competitors' customers.
Customers and orders in ElbuzPurchase history and customer stages in one database, 14 days free
Frequently asked questions
What is customer discovery?
Talking to customers to test an assumption about their problem before you spend money. You ask about the person's real experience, not their opinion of your idea.
What is the purpose of customer discovery?
To find out whether the problem you want to solve really exists, who has it and how they solve it today. It ends with a decision: go ahead, drop the idea or change it.
What are the four steps of customer development?
Customer discovery, customer validation, customer creation and company building. Steve Blank described them in "The Four Steps to the Epiphany"; customer interviews belong to the first step.
What questions should you ask in a customer discovery interview?
Questions about specific past events: "how did you buy this last time", "what was the hardest part", "how much did you pay". Do not ask "would you buy" or "what do you think of the idea": they collect compliments, not facts.
What is The Mom Test?
A book by Rob Fitzpatrick on how to talk to customers so that even people who want to support you cannot mislead you. Its rules: talk about their life, ask about specifics in the past, listen more than you talk.
How many customer interviews do you need?
The method does not set a fixed number. Keep going until new conversations stop bringing new facts; if the answers differ widely, split the people into groups.
Sources
- Steve Blank, "The Four Steps to the Epiphany" — about the book on the author's site: steveblank.com.
- Customer development: the four steps of the method — en.wikipedia.org.
- Rob Fitzpatrick, "The Mom Test" — momtestbook.com.
Checked on October 31, 2026.
John Smith
This article is very interesting! I never thought that interviewing clients could be so useful for business development. Do you have any tips for conducting effective interviews?