“Would you use this?” is easy to ask. It can also be difficult to interpret. A recent experience gives you more to explore: a trigger, a context, a sequence of actions, and an outcome.
Start with something that happened.
When someone talks about an actual event, you can follow the details. What were they trying to do? What else was going on? Which step took longer than expected?
“Tell me about the last time you needed to share a project with someone outside your team.”
This gives the participant a place to begin. It also makes it easier to ask a follow-up that connects to their experience, rather than a theory about what people generally do.
Follow the sequence.
Let the participant walk through the event in their own terms. Simple prompts can open up useful context:
- What prompted you to start?
- What did you do next?
- What were you expecting at that point?
- What happened when that did not work?
The aim is to understand the account. Avoid rushing to fill a pause or offering your own explanation before the person has finished.
Ask about the words that carry weight.
If someone calls a step “confusing,” “slow,” or “easy,” explore what that meant in the moment. The same word can describe very different experiences.
“You mentioned that step felt confusing. What were you trying to work out?”
A grounded follow-up can reveal a missing expectation, an unfamiliar term, or a decision the interface does not explain.
Keep the question open.
“Was the invite button difficult to find?” gives the participant an explanation to react to. “What were you looking for at that point?” leaves more room for their account.
Review a discussion guide for questions that assume a problem, suggest a preferred answer, or combine several ideas at once.
Apply the same care to AI interviews.
A conversational agent needs more than a list of questions. It needs a clear study goal, boundaries, and a way to follow an answer without inventing context or steering the participant.
A pilot can help examine whether the agent notices meaningful detail, handles interruption, and knows when an answer needs clarification. Review the actual exchanges before drawing conclusions about interview quality.
Try this before your next session: take one broad opinion question and rewrite it around a recent event. Then write two neutral follow-ups that could help you understand what happened.
This is a methodological note from Vellora Labs, not a report of a completed client study.