Marcus Andrey Study
Ler em português Library

UX Research · Labs · 3 min read

Run three interviews

A guided lab: write a short interview guide, run three 30-minute interviews with consent and a recording, debrief after each one in an interview log, and change the guide between sessions. About four hours over a week, with a video call tool and a notes app.

You learn to interview the way you learn to cook: by doing it, tasting, and doing it again. Three interviews won't answer a big research question, and they aren't meant to. They're enough to feel what a good question sounds like, to notice when you talk too much, and to see your guide get better between Monday and Friday.

What you'll do:

  • Write a guide for a 30-minute interview
  • Recruit three people and get their consent
  • Run and record three interviews
  • Debrief after each one, in an interview log
  • Change the guide between sessions

You'll need: the plan from Write a research plan (or a small question of your own), three people who match it, a video call tool that records, a consent form, a notes app, and ideally a colleague to take notes. About four hours across a week.


1. Write a short guide

Take the top two research questions from your plan and write a guide around them: an introduction with consent, one or two warm-up questions, three or four core questions that each ask about a real past event, and a wrap-up. Follow The interview guide and Asking good questions. NN/g's steps end with one that's easy to skip: pilot your guide, to find missing questions, bad wording, a poor order, and whether it fits the time (Maria Rosala, NN/g, 2021).

For 30 minutes, plan about 20 minutes of core questions. You'll use the rest.


Three people who match the plan, not three teammates. Friends of friends are fine for practice, as long as they really do the thing you're asking about. Send the consent form before the session, say what you'll record, how long you'll keep it, and that they can stop at any time (see Research ethics and consent). Give each person a code (P1, P2, P3) and use only the code in your notes.


3. Run the first interview as a pilot

Run interview 1 exactly as planned, with the recording on. Steve Portigal's reason to record: notes end up as "filtered versions of what we heard" (Portigal, Great User Research for Non-Researchers, 2019). If a colleague takes notes, you can listen fully. During the session:

  • Ask for a specific time, not a habit: "Tell me about the last time…".
  • Let silences run. Count to five before you speak.
  • Follow the surprise, even off the guide. See Running the session and listening.

4. Debrief within the hour

SessionWho, whenConsent · recordingObservations, with timestampsSurpriseChange to the guide
P1 · guide v1, pilotFlatshare of 4 · Mon 2 Mar · 34 minSigned · video, 34:10Posts a photo of every bill in the flat's chat (06:20) · Keeps a spreadsheet "because nobody else will" (12:18) · Opened our app twice in the last month (19:40)She pays first and chases later, every timeQ3 "How do you track…" got a habit, not a story: now "Show me the last time you checked who owed what"
P2 · guide v2Couple · Wed 4 Mar · 31 minSigned · audio only, by requestNo running balance: "it evens out" (04:55) · Splits only the big bills, rent and energy (09:30) · Argued once about who paid a deposit (21:05)Exact splits feel unfriendly to themAdded probe after Q2: "What would feel unfair?"
P3 · guide v3Flatshare of 3 · Fri 6 Mar · 33 minSigned · video, 33:02Redid a split by hand to check it (14:40) · "I don't mind paying more, I mind not knowing why" (23:40) · Left a previous app after a flatmate moved out (27:15)Moving out is when the money gets messyNext round: add a question on the last time someone moved in or out
An interview log after three sessions. Each row is written within the hour, with timestamps for anything worth quoting, and the last column shows the guide getting better between Monday and Friday.Example study, invented for this course. Log: Marcus Andrey, CC BY 4.0.

Straight after each session, spend 15 minutes filling one row of the interview log with your note-taker. The GOV.UK Service Manual says to analyse "as soon as you can", while it's fresh (GOV.UK Service Manual, Analyse a research session). Write observations, what was said or done, not interpretations. Mark the timestamp of anything you might quote, so you can find it in the recording later (see Notes, recordings and transcripts).


5. Change the guide, then run interviews 2 and 3

Look at the last column of the log. If a question confused someone, reword it. If a question never got a story, replace it. If something surprising came up, add a probe for it. Keep a short changelog at the top of the guide, so you know which version each person heard.


6. Read across the three

When the third row is done, read the log across. Write three sentences that start with "All three…", "Two of three…" or "Only one…". Those aren't findings yet: three people can't tell you how common something is. They're the questions for the next round, and practice for Synthesise a study.


7. Check your work

Reflection questions: Where in the recording did you talk over a silence? Which question got the best story, and why?

The interview guide · Asking good questions · Running the session and listening · Short user interviews

Saved only on this device.

↑ ↓ to move · Enter to open · Esc to close