Founders

How to do customer discovery: talk to customers early

Customer discovery is the work of talking to the people you think you will serve, before you build much, to find out whether the problem is real, who has it and what they do about it today. It is a set of interviews and observations that replaces guesses about your idea with evidence from customers.

What customer discovery is

Customer discovery means learning from potential customers how they live and work, so that a founder can decide what is worth building. The term comes from Steve Blank’s customer development method. In his account, there are “no facts inside the building”, so founders have to get outside and test their assumptions with customers (steveblank.com). The Startups.com glossary says Blank introduced the approach in The Four Steps to the Epiphany in 2005, and that Eric Ries later folded it into the Lean Startup.

Blank’s method has four phases: customer discovery, customer validation, customer creation and company building. This page covers only the first one, which is the stage where you still do not know whether anyone needs what you are about to make.

The output is not a product. It is a short list of statements you can defend: who has the problem, how often it happens, how they cope now, what that costs them and who pays for a better way. If you cannot write those down after a round of interviews, you are not ready to build.

Why founders skip it, and why that is costly

Most founders love the idea first and look for customers afterward. That feels efficient, since building is fun and conversations with strangers are awkward. It is also how a team ends up with a polished product and an empty pipeline.

The Startup Project, which cites CB Insights research on shut-down venture-backed startups, reports that 43% of the failures in that data cited poor product-market fit. CB Insights’ own report was not reachable when this page was written, so treat that figure as a pointer to the source and check the original before quoting it. The direction is not in dispute: many startups fail because nobody wanted the thing, not because the code was bad.

Customer discovery is the cheapest insurance against that. A week of conversations costs time and a few coffees. A failed six-month build costs salaries, runway and morale. For a young company watching its cash, this is also a money question, and it is one reason finance people care about early validation. If you are weighing when the numbers need an owner, see why a start-up needs a CFO.

Start with a hypothesis, not a pitch

Before you book a single call, write down what you believe. A simple form works:

  • Customer: who exactly has the problem (for example, owners of small clinics who send invoices by hand).
  • Problem: what goes wrong, in their words.
  • Current solution: what they use instead, including spreadsheets, interns and doing nothing.
  • Your guess: why this is painful enough that they would pay or switch.

Every line is an assumption that the interviews should try to break. Founders who skip this step tend to hear what they hoped to hear. A written hypothesis makes it much easier to notice when reality disagrees with you, and that is the entire point.

Keep the segment narrow. “Small businesses” is not a customer. “Bookkeepers at agencies with fewer than 20 employees” is. A hundred interviews with the wrong segment tell you less than fifteen with the right one, which is a point several founder write-ups make, such as the Startup Project and Startups.com.

Who to talk to and how to find them

Aim for people who actually have the problem now, not people who might one day. Friends and family are the worst source, because they want to be kind. Good places to find real prospects:

  • Warm introductions from people in your network, followed by the question “who else should I talk to?”
  • Professional networks and LinkedIn messages that are short and honest about what you want: a 20-minute chat about how they handle a specific task, with no sales pitch.
  • Online communities where your customers already gather, such as industry forums, Slack groups and subreddits. Read the rules first and do not spam.
  • Local meetups, trade events and customer-facing places such as stores and clinics.
  • People who already use a competing product or a workaround. They have shown that the problem is worth spending on.

Say plainly that you are doing research. Most people agree to a short call about their own work, particularly if you are not selling. Offer to share what you learn. Sometimes a small thank-you gift is appropriate, but never offer money in a way that makes the answers worth less.

How many customer discovery interviews are enough

There is no law of nature here. Sources disagree. The Startup Project describes problem validation as 15 to 20 interviews over two to four weeks, followed by 10 to 15 on the solution. Startups.com suggests roughly 30 to 50 interviews overall and says that fewer than 10 to 15 makes it hard to tell signal from noise. In Blank’s own class, the target he mentions is 10 to 15 customer interviews a week.

A workable rule for a first pass: talk to people until you stop being surprised. When the fifth person tells you the same story as the first four, you have a pattern. When every conversation sounds different, your segment is too broad or the problem is not shared, and you should narrow the group or change the problem.

The Startup Project also proposes a checkpoint before you move on to testing a solution: at least 15 interviews, the same problem showing up in more than 60% of them, and an average pain score of 7 out of 10 or higher. These numbers are one source’s rules of thumb, not an industry standard. Use them as a prompt to set your own bar in advance, so you do not move the goalposts after the fact.

How to talk to customers: the Mom Test

The most cited playbook for these conversations is The Mom Test by Rob Fitzpatrick. According to the book’s website, it is a short, practical handbook about getting more learning from customer conversations even when people are not fully honest with you, and it covers how to avoid biased feedback and how to tell whether a prospect will really buy. The title comes from the idea that even your mother will lie to you about your business idea, if you ask her the wrong way.

The core habits are simple:

  1. Talk about their life, not your idea. Ask how they handle the task today, what they tried and what went wrong.
  2. Ask about specific past events. “Tell me about the last time that happened” beats “would you use something that fixed this?”
  3. Listen far more than you talk. Several guides suggest an 80/20 split in the customer’s favor.
  4. Treat compliments as noise. “That sounds great” is not evidence. A customer who gives you time, money, an introduction or access to their data has shown commitment.
  5. Look for emotion and effort. People who have built a workaround or already pay for a poor tool care about the problem.

The first minute matters. Introduce yourself, say you are researching how people handle a certain task, promise not to sell anything and ask permission to take notes. Then move straight into questions.

Customer discovery questions that work

Good questions are open, concrete and about the past. You do not need a long script. A dozen well-chosen prompts, used flexibly, are enough.

Context

  • What does a typical week look like in your role?
  • How do you currently handle [the task]?

Problem

  • Tell me about the last time you ran into [the problem]. What happened?
  • What was the hardest part?
  • What happens when it goes wrong?

Current solutions

  • What have you tried to fix it? What did you like and dislike?
  • Are you paying for anything now? How much?

Priority

  • Where does this rank among your biggest problems this quarter?
  • Why have you not solved it yet?

Wrap-up

  • Who else deals with this and could I talk to them?
  • Is there something I should have asked and did not?

Follow-up prompts do most of the work. “Why?”, “Can you show me?” and “What did you do next?” pull out the detail that a first answer skips. If someone says a thing is “annoying”, ask how much time or money it cost them last month.

Avoid these:

  • Leading questions. “Wouldn’t it be great if…?” invites a polite yes.
  • Hypotheticals. “Would you pay $20 for this?” is a guess about the future, and people are poor at predicting their own behavior.
  • Pitching. The moment you describe your solution, the person starts judging your idea instead of describing their situation.
  • Feature wish lists. If a customer asks for features, ask what they would do with each one. The underlying problem is what you want, not their design.

How customers evaluate a product

Interviews also teach you how people decide. Customers rarely compare products in the abstract. They weigh a new product against what they already do, and the most common rival is the status quo: a spreadsheet, a habit, an email thread, or simply living with the pain.

Listen for how the decision is made and what it takes:

  • The trigger. What event made them look for something new? A missed deadline, a lost customer, a new boss?
  • The criteria. What would a good solution have to do, in their words? Often it is speed, cost, trust or fit with existing tools, not the feature you are proudest of.
  • The risk. What could go wrong if they switch? Data migration, training time, a bad vendor?
  • The people. Who uses the product, who pays, who must approve it? In business products these can be three different people, and you need to understand each of them.
  • The alternatives. What did they look at before, and why did they pick or reject each?

This is the groundwork for pricing, positioning and your first sales conversations. A founder who knows the real buying process can write a better pitch and, just as important, avoid long sales cycles that never close.

From interviews to a decision

Conversations are only useful if you record and review them. Do these things right after each call, while memory is fresh:

  1. Write a short summary in the customer’s words, including quotes.
  2. Note the facts: role, company size, current tools, what they have spent.
  3. Mark where the conversation confirmed or contradicted your hypothesis.
  4. Record any commitments, such as a follow-up call or an introduction.

After every five or so interviews, read all the notes together and look for repeats. Group the problems, count how many people mentioned each and rank by how much pain it caused. Note who is not worth serving as well: customers with a mild problem and no budget are a poor first market.

Then decide. There are only a few real outcomes:

  • Proceed. A clear problem shows up in many interviews, people are already trying to solve it and you have a hunch about a better approach. Move to testing a solution with a prototype or a simple manual version.
  • Narrow. The problem is real but only for one sub-group. Refocus on them.
  • Pivot. A different, bigger problem keeps coming up. Follow it.
  • Stop. Nobody cares enough. Stopping at this point is a good result, since it costs weeks and not years.

Customer discovery and product management basics

The same habits are the foundation of product management. A product manager’s job is to decide what to build and why, and that requires a steady stream of customer insight, not a one-time study. Many teams keep a weekly rhythm of customer calls long after launch, so decisions about the roadmap rest on current evidence.

After discovery comes validation: showing something to customers and watching what they do. The Startup Project mentions the Sean Ellis survey, in which a product is considered to have product-market fit if more than 40% of users say they would be “very disappointed” without it. The same source, citing MeasuringU, notes that the 40% line is a rule of thumb and not a validated cutoff. Once real usage begins, numbers such as activation, retention and churn tell you whether customers keep coming back; for a worked example see the SaaS math of activation, retention and churn.

Discovery also feeds your financial thinking. What customers say they would pay, how often they would buy and how long they take to decide all become inputs to a forecast. A model built on interview data is far more credible than one built on a top-down guess; see three traits of a great financial model.

Customer discovery and fundraising

Investors at the earliest stage mostly back evidence of a real problem and a founder who understands it. A summary of what you learned, with a few direct customer quotes, is stronger than a market-size slide. If you plan to raise, it helps to know the steps to getting funded and what a seed round in Canada usually looks like. Early money tends to follow validated demand.

Be honest about what you have. Twelve strong interviews are a good start; they are not proof of a large market. Avoid presenting a small sample as a statistic.

Common mistakes

  • Talking to friends, colleagues or anyone who will be kind.
  • Interviewing a segment so wide that no pattern appears.
  • Asking about your solution before you understand the problem.
  • Hearing only what supports your plan and ignoring the rest.
  • Mistaking politeness for interest.
  • Stopping after three or four calls and calling it research.
  • Never writing anything down, so the lessons disappear.

Plenty of useful money mistakes are covered in top finance mistakes startups make, and skipping customer work belongs on the same list.

Questions readers ask

What is customer discovery in simple terms? It is talking with potential customers to learn whether a problem is real and worth solving, before you invest in building a product.

How long does customer discovery take? A first round of 15 to 20 interviews often takes two to four weeks, according to the Startup Project. Real projects vary with how fast you can find people.

Can I do it without a product? Yes, and that is the best time. You need a hypothesis and a list of people, nothing more.

Is it the same as validating my idea? Close, but discovery is open-ended learning, while validation tests a specific idea or solution. Discovery comes first.

Sources and limits

Sources used (all read in November 2026): Steve Blank’s site, the customer development category page on steveblank.com; the Startups.com lexicon entry on customer discovery; the Startup Project customer discovery guide at startupproject.org (including its references to CB Insights and MeasuringU, which we did not check at the original source); and the website of The Mom Test by Rob Fitzpatrick at momtestbook.com. Interview counts, the 60% threshold and the 40% survey rule are rules of thumb from these sources, not validated standards. This article is general information, not business or investment advice.