qa-aman/customer-discovery

Overview

Based on "The Mom Test" by Rob Fitzpatrick. The core principle: most customer conversations are worthless because founders ask questions that invite polite lies. The Mom Test fixes this by shifting from opinion-seeking ("would you use this?") to behavior-seeking ("how do you currently handle this?"). Every question must pass the test: would your mom give you an honest answer even if she wanted to protect your feelings?

Workflow

Step 1: Define the assumption you are testing

Before writing a single question, state the riskiest belief your business depends on.

Format:

• Assumption: [specific belief about customer behavior, pain, or willingness to pay]
• Why it matters: [what breaks if this assumption is wrong]
• What evidence would confirm or deny it: [observable behavior, not stated preference]

Do not run the interview until the assumption is written down. The interview exists to pressure-test this specific thing.

Step 2: Write your question set using Mom Test rules

Rule 1: Ask about their life, not your idea. Rule 2: Ask about specifics in the past, not generalities or futures. Rule 3: Talk less. Listen more.

Core questions to use:

• "Walk me through the last time you dealt with [problem area]."
• "How are you handling it today?"
• "What have you tried before? What happened?"
• "How much time or money does this cost you right now?"
• "Who else is involved when this happens?"
• "What would need to be true for you to change how you do this?"

Questions to cut entirely:

• "Would you use [your product]?"
• "Do you think this is a good idea?"
• "How much would you pay for this?"
• "Can you imagine a world where..."

Step 3: Run the interview (20-30 minutes)

Structure:

• 2 min: warm-up ("I'm not selling anything - I'm trying to understand how you work")
• 20 min: open-ended questions, following their answers rather than your script
• 5 min: close ("Is there anything I didn't ask that I should have? Who else should I talk to?")

Take verbatim notes. Do not paraphrase in real time - you lose signal when you summarize. After each answer, ask one of: "Tell me more about that." / "What happened next?" / "Why?" Do not advance to the next question until the current thread is exhausted.

Step 4: Synthesize within 30 minutes of the interview

Capture immediately after:

• 3 direct quotes that surprised you (verbatim)
• The specific workflow or behavior they described
• Emotional tone: frustrated, resigned, indifferent, or actively seeking a fix
• What they are already doing or paying to solve this
• The one thing that changed your understanding of the assumption

Keep each interview raw. Do not aggregate across interviews yet.

Step 5: Identify patterns across 5+ interviews

After enough sessions, look for:

• Repeated phrases - exact customer language is your copywriting
• Consistent workflows or workarounds that reveal the real problem
• Who suffers most - this is your early adopter profile
• What people are already paying for - reveals real willingness to pay
• What nobody mentions - may signal the problem is not acute

Update your original assumption: confirmed, denied, or did a more important problem surface?

Step 6: Write the insight memo

One page. No decks.

Sections:

1. Original assumption
2. What we learned (3-5 concrete findings with supporting quotes)
3. Early adopter profile (who suffers most and what they look like)
4. What we were wrong about
5. Next assumption to test

Anti-Patterns

1. Compliment collection Bad: "Everyone I talked to loved the idea." Good: "Three of five people described spending 4+ hours a week on this manually. Two had already paid for a tool that failed them."

2. Future-tense answers treated as data Bad: Recording "they said they would definitely pay for this." Good: Asking "have you paid for anything like this before? What happened?" and recording what they actually did.

3. Pitching during discovery Bad: Describing [your product] mid-interview because you sense interest. Good: Staying in question mode for the entire session. Save the pitch for a separate conversation.

4. Averaging across interviews too early Bad: Running 2 interviews, declaring the assumption validated, and building. Good: Running at least 5-7 interviews before drawing patterns. Outlier responses often contain the real signal.

Quality Checklist

• [ ] Assumption is written down before the interview starts
• [ ] Every question asks about past behavior, not future hypotheticals
• [ ] Verbatim quotes captured during or immediately after the session
• [ ] Notes include what the person currently does, not just what they wish existed
• [ ] Synthesis memo updates the original assumption with a clear confirmed/denied/pivoted verdict
• [ ] Early adopter profile becomes more specific after each round
• [ ] At least one thing believed before the interview was changed by it