Begin with a specific recent occasion
“Would you use a tool like this?” asks somebody to predict a future reaction. “Tell me about the last time this happened” invites an account of something they can describe. The second question leaves more room to learn how the situation works.
Choose participants whose responsibilities relate to the question you are investigating. Explain the purpose of the conversation and ask permission before taking notes or making a recording. Use an appropriate way to handle any personal or confidential information; a useful account of a process rarely needs a list of real customer names.
Follow the sequence without supplying the answer
Ask what started the task, what the person did next, which information they needed and how they knew it was finished. When a step is unclear, ask for more detail. The GOV.UK Service Manual recommends open, neutral questions for in-depth interviews.
A fictional example: a participant says that a record is “checked” before a handover. Ask what checked means in that situation. Who checks it? What are they looking at? What happens if something is missing? A familiar word can hide several different activities.
Keep a small set of prompts
- What was happening just before the task began?
- What did you do first, and what did you need to know?
- Could you walk me through what happened next?
- Where did you need help or extra information?
- How did the work move to another person?
- How did you know the task was finished?
Use the guide to support the conversation rather than turn it into a rapid questionnaire. Leave room for an unexpected part of the account.
Distinguish the account from your interpretation
Keep separate notes for what was said, what was observed and what you think it means. A participant reporting that they use three files does not, by itself, establish that they need a single application. The files may serve different purposes or people.
Write tentative explanations as questions to investigate. “The handover may depend on an informal message” is more faithful to an early observation than “customers need automated notifications.” Keeping the inference visible makes it easier to change when another conversation provides a different explanation.
Look for differences as well as patterns
Compare the situations, responsibilities and constraints behind the accounts. Similar words can describe different work. A person who owns a process may describe it differently from a person who carries out one step during a busy shift.
Do not treat a small interview set as a numerical estimate of an entire market. Use it to improve the questions, identify possible needs and decide what further observation or testing would help. Record contradictory accounts instead of smoothing them into one convenient story.
Let the next test follow the evidence
A conversation might suggest a clearer instruction, a change to a form or a prototype worth trying. Choose a next step that tests the uncertainty you actually found. When showing a proposed solution, give the participant a realistic task and watch where the context is still missing.
The findings can feed an audience hypothesis or a workflow map. A useful interview does not have to validate the original idea. It should improve your understanding enough to make a better next decision.
Sources & further reading
Published by Naltyx. Examples are illustrative; any Naltyx product mentioned is our own. How we write.