Persona Research Interviews: Questions That Produce Usable Evidence
Persona interviews fail when they collect opinions about products instead of evidence about lives. Recruit people who recently did the job your product supports, ask about the last time they did it in detail, and record what they did, paid, and struggled with. Personas built on behaviour survive scrutiny. Personas built on preferences do not.
Recruit for recency and reality
The best participants did the relevant activity in the last few weeks, unaided, in the real context. Screen for that behaviour explicitly. People who might someday do the job give hypothetical answers, and hypothetical answers build fictional personas no matter how carefully you analyse them.
For more on this, see this related guide and our earlier explainer on the same theme.
Pay participants properly and recruit beyond your existing customers. Customers explain your product back to you. Non customers doing the same job with other tools, or with no tool, reveal the market your persona actually needs to describe.
Ask about the last time, in order
Open with the most recent occasion and walk it chronologically: what triggered it, what they did first, where it got hard, what they used, what it cost, how it ended. Specific episodes produce concrete detail. General questions produce the participant's self image, which is the least reliable data in research.
Follow surprises, not your script. When a participant mentions a workaround, a person they asked, or a tool used off label, that detour is usually the insight the persona needs. Scripts keep interviews comparable. Detours make them valuable. Budget time for both.
Separate evidence from interpretation live
Take notes in two columns: what the participant said and did, and what you think it means. Mixing them in the moment contaminates the record, and weeks later nobody can tell a quote from a conclusion. Verbatim phrases are gold for persona writing. Keep them exact.
Readers also found this companion piece from our archive useful.
Tag every claim in the eventual persona with its evidence: how many participants, from which segment, in what context. A persona whose claims carry their receipts can be argued with productively. One without them can only be believed or ignored.
How many interviews are enough
For a focused persona, patterns usually stabilise between eight and fifteen good interviews per distinct group. The signal is repetition: when new interviews stop changing your understanding of goals, triggers, and obstacles, you have enough for this round.
Personas expire. Work patterns, tools, and budgets shift, and an interview set older than about eighteen months should be treated as a draft to revalidate, not a fact base. Schedule light refresh interviews before any major product bet that leans on persona claims.
Frequently asked questions
Should interviews be remote or in person?
Remote interviews recruit faster and wider and work well for most persona research. In person or contextual sessions earn their cost when the physical environment, tools on a desk, a workshop setup, shapes the behaviour you need to understand.
What questions should I never ask?
Avoid asking participants to design the product, predict their future behaviour, or rate hypothetical features. All three produce confident, useless answers. Ask about the past, the specific, and the costly, where memory is anchored in real events.
How do I stop personas becoming stereotypes?
Anchor every trait to observed behaviour with evidence tags, describe ranges instead of single types, and include what the persona avoids and fails at. Review drafts with someone who does the job and let them strike anything that reads like a caricature.
Can survey data replace interviews for personas?
Surveys measure how common something is. Interviews explain why it happens and what it means. Strong personas use interviews for structure and surveys for sizing. Skipping interviews leaves you counting behaviours you do not understand.