Prepare interview questions that uncover how work really happens
Draft neutral questions, follow a concrete example and keep interview evidence separate from your interpretation.
Take the next useful step.
When you want to improve a process, a useful conversation starts with what people actually do. AI can help you prepare questions and notice assumptions in them. It cannot supply the participants’ answers. Use a fictional process to practise asking about a recent example, then follow your workplace’s consent and research procedures for real conversations.
Choose a decision the conversation will inform
Our fictional team wants to improve requests for shared equipment. The proposed decision is whether the request form needs clearer instructions. That is narrower than asking whether employees like the whole workplace. Write down what you need to learn before drafting questions.
Separate your hypothesis from the evidence. “People miss the return-date field because its label is unclear” is a possibility. The conversation may reveal another cause, such as uncertainty about the project schedule. Keep enough room for the participant to surprise you.
Replace leading questions with recent examples
“Would a clearer form save you time?” invites agreement with your proposal. “Tell me about the last time you requested equipment” gives the participant a concrete event to describe. Follow with “What did you do next?” and “Where, if anywhere, did you have to stop?”
Ask one thing at a time. A question about speed, clarity and satisfaction may produce an answer you cannot interpret. AI can flag those bundled questions and suggest simpler alternatives while preserving the decision you are investigating.
A request you can adapt
Draft six neutral interview questions about a recent equipment request. Start with a concrete example, then ask about steps, information needed and workarounds. Flag leading assumptions in my proposed questions. Do not invent participant responses, summarize nonexistent interviews or claim that the guide validates our hypothesis.
Follow the surprising detail
In an authored example, the participant says, “I waited until someone confirmed the project dates.” A useful follow-up is “What did you need to know before you could submit?” It explores the dependency. Asking “So the label was confusing?” would pull the conversation back toward your preferred explanation.
Keep a note with three fields: what was said, what you think it means, and what remains uncertain. A participant’s description is evidence of their account, not automatically proof of every event or a measure of how common the issue is.
Inspect the guide before a real conversation
Read the questions aloud and remove jargon that the participant may interpret differently. Ask whether any question requires unnecessary personal or confidential information. Agree on how notes or recordings will be used before collecting them through the appropriate process.
Use AI only on material you are permitted to process. An anonymous summary can still reveal identity through a distinctive event. A practice role-play can test question clarity, but simulated answers must stay labeled as invented material.
Turn evidence into a smaller next step
For a new exercise, imagine two fictional accounts: one person did not know the return date; another could not find the form. Propose a separate next check for each. Do not combine them into a claim that a redesigned form has solved the process.
Keep the interview guide, permitted notes and a decision memo together. The memo should say what the conversations support, what they do not establish and what you will test next. This makes the research useful without turning a handful of accounts into a universal conclusion.
Take this into your next task
Ask about a recent real example and keep the answer separate from the explanation you hoped to hear.
Questions people ask
Can AI simulate a participant first?
Yes, as a clearly fictional rehearsal of your questions. Simulated answers are not customer or employee research.
How many people should I interview?
That depends on the decision and research design. Do not treat a convenient small sample as representative of everyone.
Should I ask people which solution they want?
You can discuss options, but first understand the task, constraints and current behavior that a solution needs to support.
Sources and editorial notes
Reviewed 2026-09-07. Product capabilities depend on the app, plan, region and workspace settings. Workflows and prompts are authored teaching material, not recorded model results or measured time-saving claims.
Keep going
- AI at Work: Practical Skills, Examples and a Learning Roadmap
- Turn meeting notes into actions people can verify
- AI Mission Control: From Idea to Verified Result
Related Power of AI pages
Keep reading with Start here, everyday uses, the tool guide, the writing workshop, and sources and standards.