A Founder’s Guide to Useful Customer Interviews

Recruit for experience, not agreement

Define the behavior or situation you need to understand, then invite people who have encountered it recently. Job title alone may be a weak filter: two people with the same title can have different responsibilities, tools, and buying influence. Explain the purpose honestly, how long the conversation will take, and whether you are researching a problem or demonstrating a concept. Do not imply that participation is an endorsement.

A hypothetical team researching appointment scheduling for small clinics might speak with reception staff who recently handled rescheduling, rather than only asking clinic owners whether a new tool sounds useful. The team can include different clinic sizes later, but should first learn one workflow in enough detail to understand its steps and exceptions.

• Describe the experience you want to learn about.
• Invite without promising a product, result, or benefit.
• Ask permission before recording or quoting.
• Keep the conversation length and purpose clear.

Ask questions that point to a real event

Begin with context: “How does this process work today?” Then ask for the most recent example. “What happened the last time an appointment changed?” is more informative than “Is scheduling difficult?” Follow the story: what triggered the change, what the person did, which tools they used, where they waited, and how they knew the task was complete.

Use neutral follow-ups such as “Can you tell me more?” or “What happened next?” Pause after asking. Resist the urge to fill silence with a suggested answer. If someone says a process is frustrating, ask what specifically made it so and how often it occurs. Their answer may reveal that the issue is rare, concentrated in one step, or solved by a workaround your initial framing missed.

Avoid leading and hypothetical questions

Questions can bias the evidence. “Would a dashboard save you time?” describes a benefit before the person has explained the problem. “How do you currently keep track of the appointments that need attention?” leaves room for a surprising answer. Likewise, “Would you pay for this?” asks someone to forecast future behavior without the constraints of a real choice.

If you show a concept, do it after learning about the current workflow. Ask the participant to describe what they think the concept does, what they would do first, and what is unclear. This tests comprehension and usability signals; it does not establish willingness to buy. Make clear that you are evaluating the concept, not the participant.

Turn notes into a learning record

Use a short template for every conversation: context, recent event, steps, tools, workaround, consequence, exact phrases, surprises, and open questions. Distinguish quotations from your interpretation. After several conversations, compare patterns and exceptions. A repeated story can justify a narrower experiment; it still does not prove that every person in the market shares the same need.

Hypothetical synthesis: three receptionists describe rescheduling, but only one reports difficulty finding available staff coverage. Do not summarize this as “clinics struggle with scheduling.” State the smaller finding and recruit more people with the relevant responsibility. Good synthesis makes uncertainty visible so the team knows whether to broaden, narrow, or change the research question.