How do I conduct customer interviews?

The fast version

The Mom Test (Rob Fitzpatrick, 2013) is the canonical discipline for customer interviews that produce real signal:

1. Never mention your idea. It contaminates the conversation.
2. Ask about their life, not your idea. Their past behavior > your hypothetical future.
3. Ask for specifics in the past. “What did you do last time” beats “would you” every time.
4. Talk less, listen more. 80% them, 20% you. Count to 5 in silence after their answer; what they volunteer next is gold.
5. Look for commitments, not compliments. Time, money, reputation are the only signal.

Run 10-15 interviews per buyer segment. Below 10, you have anecdotes. Patterns stabilize around interview 10.

The questions that work

Past behavior questions:

• “When you last had [problem], what did you do?”
• “How much time did it take? How much did it cost?”
• “Who else was involved?”
• “What did you try that didn’t work?”
• “How often does this happen to you?”

Buying-journey questions:

• “Walk me through what you did from first realizing you needed [solution] to actually buying”
• “Who else weighed in on the decision?”
• “What almost stopped you?”

Commitment questions (only after the past-behavior ones):

• “Could I introduce you to someone else who has this problem?”
• “Would you be willing to pay $X right now to use this when it’s ready?”
• “Can we schedule a follow-up where I show you a rough mock-up?”

The questions that don’t

These feel useful and produce noise:

• ❌ “Would you pay for this?” (Predicts nothing.)
• ❌ “Would you use a tool that…?” (Hypothetical.)
• ❌ “What’s your dream solution?” (Imagination, not behavior.)
• ❌ “Do you think this is a good idea?” (Compliment-fishing.)
• ❌ “Don’t you hate when X happens?” (Leading.)
• ❌ “Would your team find this useful?” (Speculation about others.)

What you’re listening for

In any customer interview, three signals matter more than the rest:

1. Verbatim pain phrasing. The exact words the buyer uses to describe the problem. These become your marketing copy. If three unrelated buyers describe the pain in the same words, you have your headline.

2. Spending evidence. “I built a spreadsheet for this” or “I paid $X for [competitor] but stopped because Y” tells you the pain crossed the action threshold. Pain that hasn’t been spent on isn’t real pain to that buyer.

3. Workarounds. What’s the duct-taped version of the solution they’re using today? That’s your competitor. If the workaround is “they suffer in silence,” there might be no market. If the workaround is “they pay $50/month for [legacy tool],” there might be a great one.

Common mistakes

1. Pitching in the first 5 minutes. Once you describe your idea, the interview is contaminated. Save the pitch for the last 5 minutes, if you need it at all.

2. Asking leading questions. “Don’t you find X frustrating?” produces “yes.” Replace with “tell me about the last time you did X.”

3. Talking too much. Founders talk 80% of the time and conclude the customer loves their idea. Flip the ratio.

4. Interviewing friends. They lie out of love. Interview strangers, ex-colleagues, or people one LinkedIn-degree away.

5. Counting one good interview as proof. Patterns emerge around interview 10. Below that you have noise dressed as signal.

6. Not writing down verbatim phrasing. Their words are your copy. Yours are not.

How long does an interview take

30-45 minutes per interview. Add 10-15 minutes for review/transcription. Plan for ~50% no-show rate when scheduling, so to land 10 confirmed interviews, schedule ~14 slots.

For 10 interviews: realistic wall-clock time is 7-14 days, mostly bound by scheduling friction (3-7 days of back-and-forth per buyer to land a slot).

How ShipFit operationalizes this

ShipFit’s stage 3 (What Hurts?) generates a tuned set of Mom Test interview questions based on your buyer + hypothesis. After interviews, ShipFit helps synthesize transcripts into:

• The 3-4 most-cited verbatim pain phrasings (your headlines)
• The 2-3 buyer sub-segments with different behavior patterns (your segmentation)
• Contradictions between what buyers said and what they described doing (your bias check)

ShipFit doesn’t run the interviews for you (no AI can; the buyer needs a human across the table). It makes sure you ask the right questions and don’t fool yourself with the answers.

Further reading

• The Mom Test framework, full breakdown with worked examples.
• Buyer Persona Canvas, what to do with the persona insights interviews surface.
• Idea validation, the four-method ladder, where customer interviews fit in the wider validation flow.
• Rob Fitzpatrick, The Mom Test (2013), the source.