Entrepreneurship field guide

How to Conduct Customer Interviews Without Leading the Customer

Customer interviews are most useful when they reveal what people already do, where a workflow breaks, and what the problem costs. They become unreliable when a founder spends the conversation selling an idea or asking for compliments.

Start with a narrow learning goal

An interview should reduce one important uncertainty. You might need to learn who experiences a problem most often, how they handle it now, what triggers a search for help, who approves spending, or why a previous attempt failed. Trying to answer every business question in one conversation produces shallow notes.

Write the learning goal before recruiting. Define the participant by situation rather than a vague demographic. “Owners of firms with fewer than twenty employees” is broad. “Owners who prepared payroll manually during the last three months” identifies experience relevant to the question.

Recruit people with recent experience

Recent behavior is easier to recall and examine than a general opinion. Recruit from professional groups, customer-support records, referrals, communities, events, or direct outreach where appropriate. Explain that you are conducting research, how long the conversation takes, whether it is recorded, and how information will be used.

Avoid recruiting only friends who want to encourage you. If compensation is offered, state it clearly and do not connect payment to positive feedback. Protect confidential and personal information. Ask permission before recording, quoting, or identifying anyone.

Use a behavior-first question sequence

Begin with context: role, relevant workflow, and the last time the situation occurred. Then reconstruct the event. What triggered it? What happened first? Which tools or people became involved? Where did time, money, attention, or trust get lost? What did the participant try? What happened afterward?

Useful follow-ups include “What do you mean by that?”, “Can you show me how that works?”, “What happened the last time?”, and “How did you decide?” These questions invite detail without suggesting the answer.

Avoid the questions that manufacture demand

“Would you use a platform that saves time?” encourages agreement with an attractive promise. “Do you think this feature is useful?” asks the participant to evaluate your idea rather than describe life. “How much would you pay?” lacks the conditions of a real purchase.

Replace them with evidence questions: “What do you use now?”, “What does that cost in a normal month?”, “When did you last search for another option?”, “Who would need to approve a change?”, and “What prevented you from switching?”

Separate the interview from the pitch

If you need to show a concept, mark the transition. Finish the problem interview first, summarize what you heard, then explain that you would like a reaction to an early concept. Observe which part the person understands, questions, or ignores. Do not argue with a negative reaction.

A prototype reaction remains weaker than a real commitment. The next test may be a scheduled follow-up, introduction to a decision maker, data-sharing step, letter of intent, deposit, or paid pilot. Choose a commitment proportionate to what actually exists.

Analyze patterns without flattening differences

After each interview, record the situation, trigger, current alternative, cost, decision process, exact language, and surprises. Mark evidence separately from your interpretation. After several conversations, group repeated patterns and contradictory cases.

The U.S. Small Business Administration recommends using market research to examine demand, market size, location, saturation, and pricing, then combining it with competitive analysis. Interviews contribute depth, but they should be paired with market and behavioral evidence.

Turn learning into a decision

End the research cycle by deciding what changes. Narrow the segment, revise the problem, test a different offer, or stop the hypothesis. More interviews are not automatically progress. Continue when new participants still reveal decision-relevant information; move to a behavioral test when the pattern becomes stable.

Frequently asked questions