Most founders I speak with have done some version of customer discovery. They set up calls, asked questions, took notes. Then they went back and built what they had already planned to build anyway.
That is not discovery. That is confirmation bias with extra steps.
The difference between discovery that changes your roadmap and discovery that wastes your time comes down to how you structure the conversation. Here is the approach that actually works.
What discovery is not
Customer discovery is not a sales call. It is not a product demo. It is not a chance to explain your idea and watch for nodding heads. The moment you start describing your solution, you have ended the discovery and started the pitch.
Most founders make this mistake within the first five minutes. They outline the problem they are solving, ask whether it resonates, hear the person say yes, and walk away thinking they validated something. They did not. People are polite. Nodding is easy. Buying is hard.
The goal of discovery is to understand the person's world, not to introduce yours.
Who to talk to
Your first instinct will be to talk to people who are warm to you: friends, former colleagues, investors. The signal here is weak. People who like you will be supportive regardless of whether your idea has merit.
You want to talk to people who have the problem you are solving — ideally people you have no prior relationship with, who have no reason to be kind. The honest ones are found in communities where they already complain about the problem: Slack groups, Reddit threads, LinkedIn posts, industry forums. Reach out cold. Offer fifteen minutes and nothing in return.
Aim for a sample that covers different roles, company sizes, and contexts where the problem occurs. Homogenous interview pools produce homogenous — and often misleading — results.
The questions that matter
The mistake is asking what people want. The right approach is asking what people have already done.
Future intent is a terrible predictor of actual behavior. "Would you use a tool that did X?" almost always produces a yes. Past behavior is the only reliable signal you have.
Build your interview around questions like these: Walk me through the last time you dealt with this problem. What did you actually do to solve it? How much time did that take? What did you try before settling on that approach? What frustrated you most about the way you handled it?
Notice what these questions have in common: they are all about what already happened. You are reconstructing lived experience, not eliciting hypothetical preferences.
Follow-up questions matter more than your prepared list. When someone mentions a workaround, go deeper. Ask why they chose that workaround over the obvious alternative. Ask what it would take for them to switch. Ask what they would lose if the workaround disappeared.
The signals worth paying attention to
After twenty or thirty interviews, you develop an ear for what matters. Here is what to listen for.
Frequency and recency. If someone brings up the problem unprompted and can point to the last time it bit them within the past week or month, you are looking at an active pain point. If they have to think hard to remember when it last happened, the pain is not acute enough to drive purchasing behavior.
The workaround. Every manual workaround is an opportunity. If someone is spending hours in spreadsheets to do something that should take minutes, they are already willing to invest time in solving the problem. Willingness to invest time is a proxy for willingness to pay.
Emotional language. The word "nightmare" tells you more than the word "inconvenient." Listen for the moments where tone shifts. Frustration, resignation, and surprise are signals that the problem has weight. Mild annoyance is not worth building a company around.
Budget ownership. Somewhere in the conversation, try to understand who controls the budget for solving this kind of problem. If your champion has no budget authority and no influence over it, you will always be fighting uphill.
How many interviews do you need
The number most people cite is twenty to thirty. That is a reasonable floor, but the real answer is: until the conversations stop surprising you.
When you finish an interview and you already know what the person is going to say before they say it — when the problems, the workarounds, and the frustrations are predictable — you have reached saturation. That is when to stop and synthesize what you have learned.
Before that point, resist the urge to conclude anything. One compelling interview is an anecdote. Ten interviews with a consistent pattern is a signal. Twenty with that same pattern is something you can act on.
What to do with what you learn
Document every interview the same way: the person's context, the problems they named, the workarounds they use, the exact phrases they used to describe the pain. Pattern-matching across interviews is how insight emerges.
Look for the customers where the problem is sharpest, the workaround is most painful, and the context matches the customer you can actually reach and serve. That overlap is your early wedge.
Discovery does not end after your first twenty calls. The best founders treat it as a permanent practice. As your product evolves, as your market shifts, as new competitors appear, the conversations keep the strategy honest in a way no analytics dashboard can.
The founders who skip this step, or who do it once and declare it done, are the ones who build carefully crafted solutions to problems nobody has urgently enough to pay for. Discovery is the cheapest form of insurance you have against that outcome.