User research interview notes: capturing what people actually said

Quote the words people use for their problems, note what they did as well as what they said, and treat the transcript — not your memory or your summary — as the source you go back to.

This guide covers how to take notes during a research interview, a template for writing them up, how to turn several interviews into findings, and how to handle consent and recordings.

Quotes over paraphrase

A paraphrase is already an interpretation. "User finds exporting confusing" could mean they could not find the button, did not understand the formats, or were confused by what happened after. "I clicked Export, got a zip, and had no idea which file was the one I wanted" tells you which.

Paraphrase also drifts. Notes taken by someone who believes the onboarding is the problem will find onboarding problems in every interview. The participant's own words are the best protection against hearing what you expected.

In practice:

• Quote descriptions of the problem. The words people use are often the words they would search for, and the words your product copy should use.
• Quote reactions. "That's more than I'd pay" and "that's about right" are different findings even if both people say yes.
• Note behaviour separately from opinion. What someone did in a usability test ("scrolled past the Save button twice") is stronger evidence than what they said about it afterwards.
• Mark your own interpretations. If you write something that is your inference, label it — "[my read: they do not trust the auto-save]" — so nobody later mistakes it for something the participant said.

During the interview

Live notes in a research interview are for the interviewer's follow-ups, not the final record. If someone else is taking notes, the interviewer can write almost nothing. If you are alone, keep it to:

• Timestamps of moments to revisit. "12:40 — the workaround with the spreadsheet." With a recording, this is how you find it again.
• Follow-up questions. Things to come back to before the end.
• Non-verbal observations a transcript will not capture: a long pause, a sigh at a particular screen, where they looked.

Everything else can come from the transcript, if you have one and the participant agreed to it.

The write-up template

Write one of these per interview, the same day.

# Interview [number]: [participant code] — [date]

Interviewer: [name]
Note-taker: [name, or "from recording"]
Participant: [code, not name] — [segment, role, relevant context]
Recording: [yes, consent form signed / no]
Research question: [what this round of interviews is trying to learn]

## Context

- How they do the task today:
- Tools they use:
- How often:

## Key quotes

> "[exact words]" — [timestamp]
> "[exact words]" — [timestamp]

## Observed behaviour

- [what they did] — [timestamp]

## Problems and workarounds

- Problem: [in their words]
  Workaround: [what they do instead]
  How much it matters to them: [what they said, not your estimate]

## Surprises

- [anything that contradicts what the team assumed]

## My interpretations

- [clearly marked as yours]

## Follow-ups

- [questions for next interviews, or for this participant]

Use a participant code rather than a name in the notes. It keeps the research shareable inside the team without spreading personal details, and makes deleting identifying data later much simpler.

A short fictional example

Harbour and Northwind are invented; so is this participant.

# Interview 4: P4 — 29 September 2026

Interviewer: Ana Ruiz
Note-taker: from recording
Participant: P4 — operations lead at a Northwind-sized logistics firm
Research question: how teams hand over shipments between shifts

## Key quotes

> "The handover is a WhatsApp message and a prayer." — 06:12
> "If Harbour could tell me what changed since I left, I'd open it first thing." — 18:40

## Observed behaviour

- Opened three tools to answer "what's late today?" — 09:05

## Problems and workarounds

- Problem: "I don't know what the night shift already chased."
  Workaround: a shared spreadsheet with a "chased?" column, often out of date.
  How much it matters: "It's the first ten minutes of every morning."

## My interpretations

- [my read] The spreadsheet is a symptom; the need is a record of who did what overnight.

From interviews to findings

Synthesis is where paraphrase does the most damage, because it is where notes from different people get merged.

1. Pull quotes onto cards — one observation or quote per card, with the participant code and timestamp.
2. Group by what people are trying to do, not by feature. "Finding out what changed overnight" is a group; "notifications" is a solution.
3. Count participants, not cards. One talkative participant can fill a group on their own.
4. Keep the link to the source. Every finding should point to the quotes it rests on, and every quote to its interview and timestamp. When someone asks "who said that?", you can answer in a minute.
5. Write findings as statements with evidence: "Four of six participants rebuild overnight context by hand each morning (P1, P2, P4, P6)."

If an AI tool drafts a summary, treat it as a first pass and check it against the transcripts. Summaries flatten disagreement and occasionally attribute a view to someone who did not hold it.

For how research notes feed decisions and roadmaps, see meeting notes for product managers. For a general template to adapt, see the meeting notes template.

Consent and recordings

Research participants give you something personal: how they work, what frustrates them, sometimes details about their employer. Treat it that way.

• Use a consent form or a clear verbal script covering what is recorded, who will see it, how long it is kept, and how they can withdraw. Get it before you start recording. How to ask for consent to record has wording for the start of a call.
• Check your organisation's research policy and the rules where the participant is. Participant data is personal data under data protection law in many places.
• Set retention before the round starts. Many teams delete raw recordings once transcripts are checked and quotes pulled, and keep the anonymised notes.
• Limit who hears the audio. Share quotes and findings widely; share recordings only with the people who need them.

Working from the transcript with Notey

Notey records a remote interview on your Mac with nothing joining the call, and keeps your side and the participant's on separate tracks, so the transcript labels your questions You and their answers Them without anyone's voice having to be learned. Transcription happens on your Mac and the recording stays on your disk.

For synthesis, that gives you the two things this guide keeps asking for: exact words and a way back to the source. Click a line's timestamp to hear the moment again before quoting it. Search every interview by what was said with ⌘K, and file a research round under a project so its interviews stay together — more on that in building a searchable archive of your meetings. Anything Notey's AI writes is labelled as AI-generated, so a drafted summary is never mistaken for a quote.

Checklist

1. Consent form or script agreed before recording.
2. Participant codes, not names, in notes.
3. Quotes with timestamps; behaviour noted separately.
4. Your interpretations marked as yours.
5. Write-up the same day.
6. Findings linked back to quotes and interviews.
7. Retention applied once the round is done.

Frequently asked questions

Should I paraphrase or quote in user interview notes?

Quote wherever the wording matters — descriptions of a problem, workarounds, reactions to a price or a design. Paraphrase loses the detail synthesis depends on and tends to drift towards what the team already believes.

Who should take notes in a user interview?

Ideally someone other than the interviewer, so the interviewer can keep eye contact and follow up. If you are alone, record with the participant's consent and write notes from the transcript afterwards.

How soon after the interview should I write it up?

The same day. Write a short debrief within the hour while the context is fresh, and do the fuller write-up from the transcript before the next interview.

Do I need a consent form to record a research interview?

Most research teams use one, covering what is recorded, who sees it, how long it is kept and how to withdraw. Your organisation or the law where the participant is may require it; check before you start.