Three Minimum Viable Interviews Beat a Hundred Surveys

Why Surveys Are Not the First Choice

I've seen too many small teams launch surveys early, collect a hundred responses, and spend two weeks analyzing an "average need." The result: features users said they wanted go unused; features they said were indifferent turn out to be pain points.

Surveys aren't useless. But small teams lack the resources to design them well. Sample bias, leading questions, and Likert scale neutrality—these pitfalls require professional researchers to avoid. Moreover, surveys tell you what, not why.

So I recommend small teams prioritize interviews. Not ten or twenty, but three.

The Iterative Logic of Three Interviews

"Minimum viable interview" means each interview tests one hypothesis, not collects data. After three, you should be able to make one of these calls:

• The direction is right; do more validation.
• Major deviation; adjust.
• You discovered a completely different core problem; redefine the product.

Step 1: Define Your Most Uncertain Hypothesis

Set a specific, falsifiable hypothesis before each interview. For example:

"Small team PMs evaluating AI features worry most about accuracy, not cost."

If you write "users need a better product," the interview is pointless.

Step 2: Find Three Users, But Not "Typical" Ones

Small teams often pick friendly users—friends, family, early adopters. They rarely give negative feedback.

Better: find one extreme user (someone with strong frustration), one edge user (occasional use, you don't understand why), and one churned user (used but left). These three expose hypothesis flaws best.

Step 3: Design a Semi-Structured Guide with Only Five Questions

Too open → drift; too structured → lose flexibility. I use a five-question framework:

1. Background: How do you currently do X?
2. Pain point: What frustrates you most in this process?
3. Attempt: What have you tried? Why didn't it work?
4. Hypothesis test: If a solution could do Y, would it solve your problem?
5. Open: What haven't I asked that you think matters?

Avoid asking "Would you pay?" directly—it's meaningless. Ask about specific scenarios and behaviors.

Step 4: Do a 15-Minute Debrief Immediately After Each Interview

Right after each interview, spend 15 minutes writing three things:

• What evidence supported my hypothesis?
• What evidence contradicted it?
• Should I adjust my hypothesis for the next round?

Don't wait until all three are done. If the first interview disproves your hypothesis, adjust and test the new one in the second.

What After Three?

• Hypothesis validated: Do larger validation (e.g., 10 more users, an A/B test).
• Hypothesis rejected but new direction found: Redefine hypothesis and start another three-interview cycle.

Three interviews are not the end. They're a one- to two-week sprint to fail fast before committing major resources.

A Hypothetical Example

Suppose I build an AI writing tool. Core hypothesis: "users need longer articles, not better headlines." I interview three users:

• User A (content creator): his biggest pain is "keeping focus in long articles," not length.
• User B (marketer): he needs "batch generate different headline styles" because long articles can be outsourced.
• User C (student): complains about "reference formatting."

After three interviews, my hypothesis is disproven. But a common thread emerges: users need structural assistance, not just length. I pivot to a "smart outline generator" instead of a "long article generator."

Without these three interviews, I might have spent two months building a long-form feature nobody uses.

Common Pitfalls

1. Trying to cover everything in one interview: Focus on one hypothesis.
2. Asking "what if" instead of "what did": Users can't predict future behavior; ask about past actions.
3. Treating the interview as a sales pitch: Listen, don't pitch.
4. Only recording confirmations: Seek disconfirming evidence—it's gold.

When to Use Surveys

Surveys work after you've validated core hypotheses through interviews and need quantitative confirmation. For example, after interviews reveal "speed is the top concern," a survey asking "Would you pay $10 more per month for faster loading?" is meaningful.

Final Thought

For small teams, time is more precious than money. Three interviews cost little (usually a week) but can save months of wasted effort. Next time you're about to launch a survey, ask yourself: can I talk to three people first?

PaxLee