How many customer interviews do you need before you build?

How many customer interviews are enough?

For one clearly defined audience, plan for 8 to 15 customer interviews. Stop when two or three interviews in a row add nothing new about the problem, the current workaround, and what it costs people. If your interviews keep surprising you after 15, the problem is usually not the number. It is that you are talking to several different audiences at once.

That range is a working rule, not a law. It comes from qualitative research on when interviews stop producing new themes, and from the practical limit of what a solo founder can schedule and write up in two or three weeks.

What the research says about interview numbers

Researchers call the point where new interviews stop adding new information "saturation." In a well-known study, Guest, Bunce, and Johnson coded 60 interviews and found that most of the themes appeared within the first twelve, with the basic elements already visible after six (Field Methods, 2006).

Two details matter for founders. The participants in that study were a fairly similar group, and the researchers asked everyone the same questions. Both conditions made saturation arrive sooner. If your audience is mixed, or you change your questions every call, you will need more interviews to see the same patterns.

Count interviews per audience, not in total

"I did 20 interviews" can mean very different things. Twenty conversations with freelance designers tell you a lot about freelance designers. Five designers, five agency owners, five developers, and five friends tell you a little about four groups and nothing reliable about any of them.

Before you book the first call, write down who qualifies in one sentence, based on something they did recently. For example: "Freelance designers who sent at least three client proposals last month." Then count interviews against that sentence. If you want to understand a second audience, start a second count.

How to know when to stop interviewing

Keep a simple tally while you go. After each interview, write one line for each of these:

1. The trigger: what happened that made the problem matter to them.
2. The workaround: what they do about it today, including "nothing."
3. The cost: time, money, or risk the problem causes, in their words.
4. Anything new: a point you had not heard in an earlier interview.

When the "anything new" line stays empty for two or three interviews in a row, you have probably heard the main patterns for this audience. Stop and move to a test. If it never stays empty, look at who you have been talking to. Usually the answers split into two groups that should be separate audiences.

When fewer interviews are enough

You can stop earlier in two situations.

• Every answer points the same way. If six people in a row describe the same trigger, the same spreadsheet workaround, and the same lost hours, more interviews mostly confirm what you know. A small test will teach you more.
• Nobody has the problem. If the first five qualified people say they do not experience the problem, or it costs them nothing, you do not need ten more to be sure. Change the audience or the problem before you continue.

When you need more interviews

Plan for more than 15 when the audience is broad, the purchase involves several people, such as a team buying software, or the problem shows up in very different ways. In those cases, consider narrowing first. A narrower audience usually needs fewer interviews and gives you a clearer first offer.

What interviews cannot tell you

Interviews are good at explaining what people do today and why. They are poor at predicting what people will do with something new. People are polite, and "I would use that" costs nothing to say.

So treat interview results as assumptions to test, not as demand. A good next step is a small test that asks for a behavior: a reply to a specific offer, a signup on a page that says what happens next, or a pre-order you can actually deliver. Use the growth experiment template to write it down before you start.

A worked example: 11 interviews, then a test

Here is an illustrative example, not results from a Marketing Loop customer. A founder wants to build a tool that reminds freelance designers to follow up on proposals.

• Interviews 1 to 4: Every designer describes proposals going quiet. Three track them in a spreadsheet, one in their email inbox.
• Interviews 5 to 8: Two new points appear: designers feel awkward following up, and they lose most proposals in the first week.
• Interviews 9 to 11: No new points. The same trigger, workaround, and cost repeat.

The founder stops at 11 and writes one assumption: "Freelance designers who send three or more proposals a month will reply to an offer for a follow-up template pack." The next step is a message to 30 designers, not interview number 12.

Marketing Loop is in development. It is being built to keep each test, its result, and the decision it led to in one place. If you want to know when it is ready, join the build updates.

FAQ

Is five customer interviews enough?

Five is enough to learn whether a problem exists for one audience and to spot obvious patterns. It is rarely enough to be confident that the patterns hold. Use five interviews to decide whether to continue, and around ten to decide what to test first.

Should I pay people for customer interviews?

You can, but it changes who agrees to talk. People who respond to a payment may not be the people who care about the problem. For early interviews, a short, specific request to people who recently acted on the problem usually works better. The customer discovery interview guide has an example.

How long should a customer interview be?

Twenty to thirty minutes is enough for one recent experience, the workaround, and the cost. Longer calls tend to drift into opinions about your idea, which are less useful than facts about their past.

Can I use a survey instead of interviews?

A survey can count how common a problem is once you know which questions to ask. It is a poor way to discover the problem in the first place, because people can only answer the options you thought of. Interview first, then survey if you need numbers.

Start counting

Write the one-sentence audience description, book the first five calls, and keep the four-line tally after each one. When the "anything new" line stays empty, stop interviewing and write your first test.