A chat simulator for user research interview practice can help a researcher hear how a written guide may feel before speaking with a participant. It makes question order, follow-up branches, pauses, transitions, and closing language visible. The simulator is a rehearsal tool, not a participant, research method, transcript, or source of findings. A convincing simulated answer must never be reported as evidence about real users.
This guide uses a fictional study about a made-up community library search concept called Northstar Library. The participant profile, messages, task, organization, quotes, and outcomes are invented. The exercise is intentionally low risk: it shows how to practice an interview guide while protecting consent, privacy, accessibility, and the integrity of later research. You can build the rehearsal in the online Chat Simulator using invented profiles, timestamps, playback, and Mobile or Web export.
1. Define what the rehearsal can and cannot test
Start with one learning objective. In this example, the researcher wants to check whether a guide can explore how someone chooses search terms, notices filters, and responds when results look incomplete. The rehearsal can reveal awkward wording, abrupt transitions, missing probes, and an interview that is too long for its planned time.
It cannot establish that the concept is usable, that people prefer a feature, or that a real participant will answer in the same way. Those claims require an appropriate study with recruited participants, consent, real observation, documented analysis, and limitations.
- Rehearsal output: a revised interview guide.
- Not an output: user findings, quotations, or success metrics.
- Primary reviewer question: can a participant understand and answer each prompt?
- Stop condition: the guide fits the session and covers the research objective.
2. Create a fictional participant without stereotyping
Use only the context needed to test the guide. The fictional participant, Sam, is looking for beginner gardening material and sometimes searches from a phone. That is enough to create realistic follow-ups. Adding an invented employer, diagnosis, income, address, or detailed personal history would increase risk without improving the exercise.
Do not use a stereotype as a shortcut for behavior. A demographic label does not tell you whether someone understands a filter, trusts a result, or prefers one path. Build the rehearsal around observable actions and relevant context instead of assumptions about a group.
3. Turn the interview guide into a branching map
Separate core questions from optional probes. A core question supports the research objective and should be asked consistently when appropriate. A probe follows something the participant actually said or did. Mark transitions, permission checks, task boundaries, and places where the interviewer should wait rather than fill silence.
A useful branch records the trigger, neutral follow-up, reason for asking, and exit condition. This prevents the rehearsal from becoming a script that forces every participant through the same artificial conversation.
- Core prompt: open enough to invite the participant’s own account.
- Probe: connected to a word, choice, hesitation, or action already observed.
- Transition: explains why the topic is changing.
- Exit: returns to the guide or ends the topic respectfully.
4. Rehearse a complete fictional interview segment
The following example tests a search task without claiming that the fictional response predicts real behavior. The interviewer avoids praising one path, naming the expected answer, or treating task completion as proof of satisfaction. All names, wording, actions, and results below are fictional rehearsal material.
- Researcher: “For this practice task, imagine you want a beginner resource about growing herbs. What would you do first?”
- Sam: “I would search for herbs for beginners.”
- Researcher: “What are you looking for as you scan these fictional results?”
- Sam: “Something short, current, and available online.”
- Researcher: “You mentioned current. What on this screen would help you judge that?”
- Sam: “A publication date or an updated label.”
- Researcher: “Is there anything you expected to find but do not see here?”
- Sam: “I expected a way to filter for online access.”
- Researcher: “Thank you. Before we move on, is there anything else about this search step you want to add?”
5. Replace leading questions with neutral probes
Leading questions suggest the desired answer or introduce an idea the participant did not raise. “Was the filter easy to find?” assumes that it was findable and frames ease as the topic. A more neutral prompt is “What, if anything, did you notice about narrowing the results?”
Neutral does not mean vague. A good probe points to an observed moment without interpreting it for the participant. Compare “Why were you confused?” with “I noticed a pause before you selected that option. What were you considering?” The second version makes the observation visible and leaves meaning open.
- Avoid praise that rewards a particular path.
- Do not ask two questions in one message.
- Do not introduce a feature the participant has not encountered unless the study plan requires it.
- Use the participant’s own terms when clarification is needed.
6. Practice silence, confusion, refusal, and recovery
A rehearsal should include moments that do not produce a neat answer. The participant may pause, misunderstand the task, ask what a word means, decline to discuss something, or say they do not remember. The interviewer needs a respectful response that does not pressure, rescue, or turn the session into product instruction.
Plan short recovery lines: restate the task without adding hints, clarify unfamiliar language, offer to skip a question, and confirm that a participant can pause or stop according to the study protocol. If a task cannot continue because the prototype fails, record the failure and move to the approved backup instead of pretending it worked.
7. Protect consent, privacy, and accessibility
Consent is an ongoing process, not a single opening sentence. The actual study should explain purpose, recording, observers, data use, retention, withdrawal, and contact details in a form appropriate to the research plan. A chat simulator can help rehearse this explanation, but it cannot collect valid consent on behalf of the real study.
Use synthetic data in the rehearsal and remove exported images after review. Do not paste a real recruitment response, support conversation, prototype credential, access token, or participant quote. Plan accessible materials, flexible response modes, breaks, readable text, keyboard access, captions or interpretation when relevant, and a way for participants to request accommodations without having to justify them publicly.
8. Pilot, annotate, and revise the guide
Run the simulated sequence once for flow, then conduct a pilot with an appropriate person who understands that the goal is to test the study materials rather than generate findings. Mark questions that were unclear, repetitive, leading, unnecessary, or impossible to answer from the task. Record elapsed time by section and leave margin for consent, setup, pauses, and closing.
Revision should change the guide, not rewrite the fictional answers to make the interview look successful. Keep a short decision log explaining why a question was removed, reordered, or reframed. This creates a more trustworthy study record and makes future review easier.
- Check each question against the research objective.
- Remove probes that merely repeat the core prompt.
- Test transitions and closing language aloud.
- Version the guide and record who reviewed it.
Frequently asked questions
Can simulated answers be included in a research report? Only as clearly labeled examples of planned method or interface content, never as participant evidence. Can a chat simulator replace a pilot interview? No. It can reveal structural problems, while a pilot tests the materials with a real person under the planned process. Should every possible answer have a branch? No. Prepare high-value probes and principles; over-scripting can prevent the interviewer from listening. Can a real transcript be anonymized and reused? Removing a name may not remove identity or consent obligations. Follow the study’s authority, minimization, retention, and sharing rules.
User research interview rehearsal checklist
- The rehearsal objective is specific and does not claim findings.
- Profiles and answers are fictional, minimal, and free of stereotypes.
- Core questions, probes, transitions, pauses, and exits are distinguishable.
- Leading, double-barreled, and praise-based prompts are revised.
- Consent, refusal, privacy, accessibility, and prototype failure have safe responses.
- The guide is piloted, timed, versioned, and connected to the real research plan.
