Run a learner interview that changes your design

Choose a research question

Start with a decision you need to make: what support would help a new employee handle their first difficult customer conversation? Recruit people who actually perform that task, with different experience levels and access needs. Explain how you will use their input and ask permission before recording.

Ask for the last real example

Use prompts such as: “Tell me about the last time this happened”; “What did you do first?”; “What information did you need?”; “Where did you look?”; and “What made that difficult?” Ask to see relevant artifacts when appropriate and permitted. Avoid leading questions that imply a course is already the solution.

Separate observation, interpretation and decision

Observation: three participants could not find the escalation rule. Interpretation: the rule may be difficult to locate at the point of need. Design decision: test a searchable decision guide. Keep the chain visible so another designer can challenge your interpretation.

AI can help cluster de-identified notes when use is approved. Require links to evidence, check excerpts and look for disconfirming cases. A frequently mentioned theme is not automatically the most important problem.

Turn findings into a prototype

Select the highest-consequence problem supported by the research. Write a design hypothesis: “If the escalation rule is available beside the case form, new agents will choose the correct escalation path with fewer requests for help.” Test that hypothesis with a small realistic task. Record the result and what you still do not know.