Why focus groups are harder to transcribe
A focus group is many people in one room, often reacting to each other. People interrupt, laugh, agree in chorus and start sentences at the same time. That is good research and hard audio. Any transcription method, human or automatic, has more trouble with a group discussion than with a one-to-one interview.
Knowing this helps you plan. Spend your effort on the recording setup, then expect to review the passages where several people spoke at once.
Recording setup for a group
In a facility, the session is often recorded on video for clients watching behind the mirror, which gives you an MP4. Moderators running their own groups usually use a portable recorder (WAV or MP3) or a phone. Online groups on Zoom or Teams leave an MP4, and Zoom usually adds an audio-only M4A. Upload whichever file you have; audio is extracted from video automatically.
- Use a boundary or conference microphone in the middle of the table, or two recorders at each end for a long table.
- Avoid rooms with hard walls and a humming air conditioner. Echo makes overlapping voices harder to separate.
- Ask participants to speak one at a time at the start, and gently repeat it if the group gets lively.
- Have each person say their first name or participant number at the start. It makes matching voices to people much easier later.
- For online groups, ask everyone to use headphones and mute when not speaking.
Labelling participants
Speaker identification is what turns a block of text into a usable focus group transcript. It is available on paid plans (Plus, Unlimited) or with a Minute Pack, and not on the Free plan or the no-sign-up trial. You can tell it how many speakers there are, anywhere from 1 to 20. Count the moderator, any assistant moderator and every participant who spoke.
Then rename the generic labels: Speaker 1 becomes "Moderator", the others become participant codes or first names. Each rename updates the whole transcript. In a group, labels are less reliable than in an interview. Short interjections, overlapping speech and voices that sound alike can be assigned to the wrong person. Use the synced player to spot-check the sections that matter for your findings, and click any line to hear it.
Exports for reports and analysis
Choose the export based on who reads it next. Speaker names and timestamps can be turned on or off for each export.
- DOCX: for a client report appendix, or to import into qualitative analysis software that accepts Word files.
- CSV: each segment in its own row with time and speaker, handy for sorting comments by participant in a spreadsheet.
- PDF or a read-only share link: for stakeholders who only need to read the discussion.
- AI prompt export: copies the transcript with a prompt you can paste into ChatGPT or Claude to draft a topline summary. RocketScribe does not write summaries itself.
Running a series of groups
Projects often have several groups across cities or segments. Create a folder per project, upload each session into it and bulk export the folder at the end. Paid plans allow longer files and batch uploads, which suits sessions that run well past an hour. Groups held in other languages work too, with more than 90 languages supported.
Use the same participant codes and speaker names across every session in a project, for example P1 to P8 in each group plus "Moderator". Consistent labels make it much easier to compare groups later and to search the exported files. If a client is observing, a read-only share link lets them read a session soon after it ends. Keep the original recordings until the analysis is signed off, in case a passage needs to be heard again.