How to Run Customer Discovery Interviews That Actually Teach You Something

Most founders come out of customer interviews feeling great — and learn nothing useful. The problem isn't the concept. It's the execution. Here's how to run interviews that actually move your thinking forward.

Before You Start: Get Clear on What You're Trying to Learn

Don't schedule a single call until you can answer: what specific assumption am I trying to validate or invalidate?

Write it down. Something like: "I believe that HR managers at 50–200 person companies spend significant time manually chasing down approvals, and that this is a painful enough problem they'd pay to solve." That's testable. "I want to learn about HR problems" is not.

Once you have the assumption, design your interview questions backward from it. You're not building a survey — you're trying to understand the world as it actually is.

Recruiting Interviewees

Who to recruit. Your interviewees should match your target customer profile closely enough that their answers are predictive. If you're building for growth marketers at SaaS companies, don't interview CMOs at agencies. Specificity matters.

Where to find them.

• LinkedIn: Boolean search is powerful. Filter by title, company size, and industry.
• Reddit and Slack communities: Offer genuine value first, then ask.
• Your own network: Good for early momentum, but bias-check yourself — warm intros can make people want to help you.
• Customers of adjacent products: Conference attendees, meetup participants, newsletter subscribers in your space.

How to ask. Keep the ask low-friction. You're not pitching — you're genuinely curious. A simple message: "I'm researching how [job role] handles [problem area]. Would you have 25 minutes to share your experience? I'm not selling anything."

Aim for at least 15-20 interviews before drawing conclusions. Patterns don't emerge from five conversations.

Running the Interview: The Mom Test Approach

Rob Fitzpatrick's Mom Test has one core insight: people lie to your face to protect your feelings, so you need to ask questions they can't lie about.

Bad question: "Would you use a tool that automated your approval workflows?"
Good question: "Walk me through the last time you had to chase an approval. What happened?"

The good question gets you behavior. The bad question gets you optimism.

Rules for the interview itself:

• Talk less than 50% of the time. You're here to listen, not pitch.
• Ask about the past, not the future. "Have you ever..." is more reliable than "Would you ever..."
• Follow the energy. When someone's voice changes, dig in. "You mentioned that was frustrating — what made it particularly bad?"
• Don't explain your product. Not until the end, if at all. Once you pitch, the interview becomes a sales call.
• Ask for specifics. "How often does that happen?" "How long does that take?" "Who else is involved?" Numbers ground the conversation.

Questions That Consistently Work

• Tell me about the last time you dealt with [problem]. What actually happened?
• How are you handling that right now?
• What have you tried before? What happened?
• What's the worst part of your current approach?
• How much time does this take per week?
• Who else does this affect?

The One Question Most Founders Skip

At the end of every interview: "Who else should I talk to?" A warm referral from a qualified interviewee is worth more than any outreach campaign.

Synthesizing Insights

Raw notes from 20 interviews are just data. You need to turn them into beliefs.

Immediately after each call (within 30 minutes): write a short paragraph summarizing what you heard, what surprised you, and what you now believe differently. Don't rely on your notes alone — capture the emotional texture of the conversation.

After all interviews: look for patterns across three dimensions:

1. Behaviors — what people actually do, not what they say they do
2. Motivations — the underlying why behind the behavior
3. Friction — where current solutions break down

Build an affinity map: cluster quotes and observations into themes. If you're seeing the same theme in 70%+ of interviews, it's probably real. If only two people mentioned something, don't build a feature around it.

The output should be a set of crisp, evidence-backed statements: "8 out of 15 interviewees described manually compiling reports in spreadsheets as the most time-consuming part of their week." That's the kind of thing that should drive your roadmap.