The short answer
Get consent from every participant before the session, record with more than one microphone, transcribe with a service that separates speakers and processes data in the EU, and expect to spend more time checking the transcript than for a one-to-one interview. Pseudonymise every participant, including what they say about each other.
Privacy questions specific to focus groups
In an interview, only you hear what the participant says. In a focus group, everyone does. You can ask participants to keep the discussion confidential, but you cannot guarantee it, and your information sheet should say so honestly.
Each person in the room is a data subject. If one person withdraws, you may need to remove their contributions from the transcript, so plan how you will handle that before the session. Participants also mention third parties (colleagues, family members), which is personal data too.
- Collect individual written consent from every participant, not one consent for the group.
- Explain that recordings will be transcribed by an automated service in the EU and that the group cannot be fully confidential.
- Agree ground rules at the start: first names only or pseudonyms, no names of outside people.
- Decide in advance what happens if someone withdraws after the session.
Recording a focus group that transcribes well
Most transcription problems in focus groups come from the room, not the software.
- Use two recorders or a conference microphone in the middle of the table. A laptop microphone at one end will lose half the voices.
- Choose a quiet room with soft furnishings. Echo is the enemy of speaker separation.
- Ask people to speak one at a time and say their pseudonym or seat number the first time they speak.
- Have a co-moderator write a seating map and a speaking log with rough times. It makes speaker labelling much easier.
- For online focus groups, check that the meeting platform’s own recording and transcription settings are approved, or record locally.
Transcribing and checking
Modern AI transcription separates speakers automatically, but with many voices it will sometimes merge two similar voices or split one person in two. Plan to fix speaker labels by listening to the first minute of each speaker and using your seating map.
Overlapping talk is common in good focus groups. Mark it in the transcript ([overlapping]) rather than trying to capture every word, unless your method, like conversation analysis, needs it.
From transcript to analysis
Focus group analysis looks at what the group says and how it says it: agreement, disagreement and how views shift during the discussion. Keep speaker labels through coding so you can see whether a theme comes from the whole group or one strong voice.
In Kahubi, focus group recordings are transcribed by European providers with speaker labels in 99+ languages, with optional redaction of names. You rename speakers to pseudonyms, then code themes with AI suggestions you confirm. Each code stays linked to the speaker and timestamp, so you can check who said what before you write the findings.
Sources
Last updated 2026-10-09.