Creators often call a social media poll “customer research” and begin building from the winning option. Polls can start a conversation, but they rarely reveal the context, tradeoffs, or behavior behind a choice. Strong research combines several forms of evidence and treats participants with respect.
The purpose is not to manufacture certainty. It is to improve decisions while making limitations visible.
Write the decision you need to make
Begin with a concrete decision: choose between two audiences, understand why a workflow fails, decide which example belongs in the first release, or identify the current alternative your product must beat.
Then list the assumptions behind that decision. Separate what you know from what you believe. This prevents interviews from turning into broad conversations that are interesting but impossible to apply.
Recruit for the relevant situation
Define participants by behavior and context, not only demographics. “Freelance designers who onboarded at least two new clients recently” is more useful than “entrepreneurs aged 25–45.” The behavior connects directly to the problem.
Use several recruitment sources where possible. Explain who you are, why you are researching, how long it takes, whether there is compensation, what will happen with notes or recordings, and that participation does not require a purchase.
Do not add interviewees to marketing lists without appropriate consent. Keep research participation separate from sales permission.
Ask about a real recent event
Open with “Tell me about the last time…” and follow the sequence. What triggered the situation? What did the person try? What information did they seek? Who influenced the decision? What did it cost in time, money, risk, or frustration? What happened afterward?
Avoid leading questions such as “Would an automated template save you time?” Ask “How did you handle it?” before introducing your concept. Silence can be useful; do not rush to sell the participant a solution.
Distinguish statement from evidence
“I care about organization” is a statement. Showing the spreadsheet they created, describing a recent missed deadline, or naming a paid tool is behavioral evidence. Both are useful, but they should not be treated as equivalent.
Handle notes and recordings carefully
Collect only what the research decision requires. Tell participants whether you record and obtain the appropriate permission. Restrict access, set a retention period, and remove information that is not needed.
If you want to use a quote publicly, ask separately and explain the exact context. Approval to participate in research is not automatically approval to become a testimonial. Never invent or combine quotes to make a stronger story.
Add observational and market evidence
Interviews reveal perspective but not every action. When appropriate and consented, observe a person attempt the workflow or review anonymized support patterns. Combine this with public competitor positioning, product reviews, search intent, and your own sales or support data.
Each source has limitations. Public comments overrepresent people motivated to post. Existing customers may differ from people who never bought. Search volume does not explain urgency. Write these limitations alongside findings.
Analyze patterns without erasing differences
After each session, capture the trigger, goal, current method, friction, consequence, decision criteria, language, and unresolved questions. Group evidence by situation, not by the conclusion you hoped to reach.
Look for repeated behavior and important contradictions. A minority segment with a distinct high-value problem may deserve separate treatment rather than being averaged away.
Use direct quotes internally to preserve voice, but tie every quote to its context. Convert findings into a short opportunity statement and list what remains uncertain.
Share conclusions proportionately
Say “Several participants in this small study…” instead of “Customers want…” when evidence is limited. Document recruitment, dates, sample characteristics, method, and limitations so the team can judge how much confidence to place in the result.
Research should end with a decision or next experiment. If nothing changes, the research question may have been too broad.
Debrief immediately after each session
Reserve ten minutes to separate observations, interpretations, surprises, and follow-up questions while the conversation is fresh. Note where your own wording may have led the participant. Do not rewrite an ambiguous answer into a confident conclusion. A consistent debrief makes later synthesis more reliable and reveals when the interview guide needs correction before the remaining sessions.
Practical checklist
- Define the decision and underlying assumptions.
- Recruit by relevant behavior and situation.
- Explain purpose, time, recording, compensation, and data use.
- Keep research consent separate from marketing consent.
- Ask about recent real events before showing a concept.
- Record behavior, alternatives, constraints, and consequences.
- Collect and retain only necessary information.
- Obtain separate permission for public quotes or testimonials.
- Document contradictory evidence and study limitations.
- End with a decision, revision, or next test.
Use a simple evidence table
Create one row per observation with columns for participant situation, recent trigger, current method, friction, consequence, direct evidence, confidence, and open question. Keep interpretation in a separate column so the team can distinguish what happened from what it means. After several sessions, group rows by situation and look for both patterns and disconfirming cases. The table should never become a public customer database; minimize identifiers, restrict access, and preserve only what the research decision needs.
Make curiosity more valuable than confirmation
The strongest researcher is willing to hear that the problem is infrequent, the current workaround is good enough, or the buyer is someone else. That honesty leads to a product people can actually use and trust.