Customer conversations that reveal real workflows

What an interview can and cannot tell you

A discovery conversation helps you understand a person’s real workflow, language, constraints, and workarounds. It can reveal what to investigate next. A few interviews cannot estimate how common a problem is across a market, and polite enthusiasm is not a demand forecast.

Recruit people with recent experience

Choose a narrow group based on behavior or role, not on who is easiest to flatter your idea. Ask for a 20–30 minute conversation about a recent instance of the task. Explain that you are learning, not promising a product or asking them to endorse it. Avoid recruiting only friends, who may tell you what they think you want to hear.

A simple invitation:

I’m learning how independent consultants prepare client status updates. You mentioned doing this recently. Would you be open to a 20-minute conversation about your current process? I’m researching the workflow, not selling a finished product.

Ask permission before recording. If you take notes, store only what you need and remove names or sensitive details when identity is not essential.

Use a recent-event interview

Start with context and let the person tell the story. Useful prompts:

1. “Tell me about the last time you had to do this.”
2. “What triggered it? What happened first?”
3. “Can you show me the steps or artifacts you used?”
4. “Where did you get stuck, and what did you do next?”
5. “How often does this happen? When did it last happen before that?”
6. “What does the problem cost you or someone else?”
7. “What have you already tried? Why did you stop or keep using it?”
8. “Who else is involved in choosing or paying for a solution?”

Follow concrete details with neutral questions: “What do you mean by ‘slow’?” or “What happened after that?” Silence is useful; it gives the person space to remember rather than agree.

Avoid leading and hypothetical questions

Do not ask “Would you use an app that did this?” or “Would you pay $20?” before understanding the current behavior. These invite guesses. Do not pitch a solution, list features, or argue with a complaint. If you need feedback on a concept, label that as a separate usability or offer test, then watch the person use it rather than asking whether they like it.

If someone says the problem is important, ask for an example. If they say they already solved it, learn what works and what remains frustrating. A strong problem may already have a workaround; your product must be meaningfully better than that workaround, including switching effort.

Synthesize without turning anecdotes into statistics

Immediately after each interview, note direct observations separately from interpretation. Record repeated behavior, existing spend, frequency, consequences, workarounds, decision-makers, and any evidence against your hypothesis. Look for patterns across people, but keep the denominator visible: “3 of 6 interviewees” is honest; “half the market” is not.

Use this summary:

• Observed: what the person actually did or showed.
• Said: their words, in context.
• Interpreted: what you think it means.
• Unknown: what another method must answer.
• Next action: the smallest useful follow-up.

After five to eight focused conversations, decide whether the problem is clearer, not whether the sample has “validated” the market. If the pattern is weak, recruit a different segment or test a more precise problem. See idea to first experiment and the experiment brief.

For a reusable observation-to-finding workflow, see analyze a small sample of customer research.

If you already understand the workflow and need to learn how common a behavior is within a defined group, see design a useful customer survey. Do not use a survey as a shortcut for discovering the behavior in the first place.