How to Do Customer Interviews for a Startup Idea

What a customer interview is and why it comes first

A customer interview is a short, structured conversation with someone who might have the problem your startup wants to solve. You are not selling. You are collecting stories about how they handle the problem today. This habit is often called customer discovery, and a single conversation of this kind is a customer interview.

It comes before building because the least costly mistake is the one you make on paper. If you spend three months on a product and then learn nobody has the problem often enough to pay for a fix, the cost is large. If you learn it after five conversations, the cost is a few hours. The goal is to reach what founders call problem-solution fit: evidence that a real group of people has a real problem and that your idea addresses it.

Interviews do not replace other checks such as demand tests or market sizing. They answer one question well: what is actually going on in this person's life or work?

Step by step: how to run customer interviews

1. Write down your assumptions. Finish three sentences: who has the problem, what the problem is and what they do about it today. These are guesses. Interviews exist to test them.
2. Define who to talk to. Pick one narrow group, for example "freelance designers who invoice clients themselves", not "small business owners". A narrow group gives answers you can compare.
3. Find 10 to 15 people. Use your own network for introductions, communities where the group gathers, and direct messages to people who posted about the problem. Ask for 20 minutes and say plainly that you are researching, not selling.
4. Prepare five to seven questions. Focus on the past, not the future. The next section gives examples.
5. Run the conversation. Start by asking about their work or life, then move to the last time the problem came up. Let silences sit. Ask "why" and "what happened next" more than anything else.
6. Take notes and write a summary within an hour. Record the call only if the person agrees. Write what they did, what it cost them in time or money, what they tried and any exact phrases they used.
7. Look for patterns after every five conversations. Count how many people had the problem recently, how many already pay for a workaround and how many described it the same way.
8. Decide the next step. If the pattern is strong, test demand with a small offer. If it is weak, change the group or the problem and run another round.

Keep each round small. Ten calls with a clear group teach you more than thirty calls with a vague one.

What to ask and what to avoid

The most common mistake is asking people whether they would use your product. Most people say yes to be polite, and they cannot predict their own future behavior well. Ask about what already happened instead. The table shows the difference.

Two more questions work almost every time. Ask "what have you already tried?", because people who tried several fixes feel the problem strongly. Ask "who else should I talk to?", because it gives you your next interview.

Avoid describing your idea early. If you must show something, wait until the last five minutes, and treat the reaction as one data point, not a verdict.

A worked example

Say you are thinking about a tool that helps freelance photographers send quotes faster. This is an illustration, not a real case. Your assumption is that photographers lose bookings because they reply to inquiries too slowly.

You book 12 calls with working photographers. In each one you ask them to walk through the last inquiry they received. Say that by the sixth call you notice something. Most photographers reply the same day without trouble. What they describe as painful is chasing clients who agreed to book but never signed or paid the deposit.

That finding changes your idea. The problem you assumed (slow replies) is not the one people feel. A different problem (follow-up after the quote) comes up again and again, and several people mention workarounds such as reminder spreadsheets. You would not have found this by asking "would you use a quoting tool?".

The next step is not to build. It is to run a second round of five or six calls focused on follow-up, then offer a simple paid trial to a few of the people who described the problem most strongly.

How to take notes and read the answers

Notes that only say "seemed interested" are useless later. Use the same template for every call so you can compare them.

• Who: role, situation and how long they have dealt with the problem.
• Last time it happened: the specific story, in order.
• What it cost: time, money, missed work or stress, in their words.
• What they tried: tools, people or habits, and why each fell short.
• Exact phrases: words you could later use on a landing page or in an ad.
• Surprises: anything that contradicted your assumptions.

After 10 or more calls, put the summaries side by side. Mark each person yes or no on three questions: did they have the problem in the last month, did they spend money or real time on a workaround and did they describe it in a similar way to others? If most people say yes to all three, you have a real signal. If answers are scattered, you probably have the wrong group or a problem that is too general.

Common mistakes and a quick checklist

Most failed interview rounds share the same few errors. Check your plan against this list before your first call.

• I wrote my assumptions down before the first call.
• My target group is narrow enough to describe in one sentence.
• My questions ask about the past, not about opinions of my idea.
• I have a notes template and I fill it in within an hour.
• I know what result would make me change direction.

If you want a second pair of eyes on your idea before you start, the free roast my idea tool gives quick critical feedback with no signup. It does not replace talking to real people, but it can show you which assumptions to test first.

Where interviews fit in the bigger path

Interviews are one step in a longer process. Before them you choose an idea, perhaps with the free idea generator. After them you test demand with an actual offer, and the guide on how to validate a startup idea covers that stage. Once you have a few believers, how to find your first 10 customers explains how to turn them into paying users.

Fonda is an AI co-founder that follows this order. Its 14-step journey puts Interviews inside the Validate phase, before the Verdict, and your co-founder Marty remembers your project and can answer questions at each step. You can read how the method is built on peer-reviewed entrepreneurship research on the research page. It is free to start, and you can see how paid use works on the pricing page. You can also follow everything in this article with a notebook and a calendar.