Focus group analysis: a complete guide

Focus group analysis is the process of working through recorded group discussion data to identify themes, areas of agreement and disagreement, and findings that reflect how a group collectively makes sense of a topic. The core steps are: transcribe the recordings with speaker attribution, read for overall impression, code both the content and the group dynamics, compare findings across multiple groups, and write up with explicit attention to how the group interaction shaped what was said. The key difference from individual interview analysis is that the unit of analysis is the group, not the individual — which means the social and conversational dynamics are data, not background noise.

This guide covers the full process, including a worked example, common mistakes, and how AI tools can assist with the content analysis while you focus on the interpretive work that requires human judgement.

Why focus group analysis is different

The method shapes the data. Focus groups produce material that one-to-one interviews do not, and analysing it as if it were a set of individual interviews produces weaker findings.

The distinctive features of focus group data are:

Collective meaning-making: participants build on each other's contributions, correct each other's framing, introduce ideas others take up and extend. The analysis needs to capture these dynamics, not just the individual utterances.

Social influence: what participants say in a group is shaped by who else is in the room. Positions are moderated by social pressure, norms of the group, and the presence of authority figures. This is partly why focus groups are limited for sensitive topics — but when the phenomenon you are studying is itself social, it is analytically valuable.

Consensus and dissent: groups often produce a surface-level consensus that can be misleading. One or two vocal participants can make a majority view out of what is actually a minority position. Equally, dissenting voices are sometimes suppressed and may only emerge in how participants qualify their agreement. Both are worth tracking.

Turn-taking and sequence: who said what in response to whom matters. A comment made after a strong intervention from a dominant participant means something different from the same words said spontaneously at the start of the discussion.

For a deeper treatment of the unique challenges of focus group transcripts, see how to analyse focus group transcripts. This guide focuses on the broader analysis process, from setup to write-up.

When to use focus groups

Focus groups are well-suited to understanding shared meaning, exploring how people construct a view collectively, and capturing the range of perspectives in a community. They are particularly useful for:

Exploring how a concept or topic is understood and talked about in natural language (consumer research, brand perception)
Understanding social norms and what is considered acceptable or problematic within a group
Generating hypotheses for follow-up research
Policy and consultation work where the goal is to surface the range of stakeholder views

They are less well-suited to gathering individual experiences in depth, measuring the prevalence of views across a population, or exploring sensitive topics where social desirability is likely to suppress genuine responses.

For a comparison of when focus groups versus individual interviews serve different research goals, see focus groups vs individual interviews.

Transcribing focus group recordings

Focus group transcription is harder than transcribing one-to-one interviews for two reasons: multiple speakers often talk simultaneously, and speaker identification requires more careful attention.

What to include in the transcript

A useful focus group transcript includes:

Speaker labels: who said each utterance. Use consistent labels (Participant A, Participant B, or role-based labels if relevant). Without speaker attribution, the analysis loses its ability to track individual positions or identify dominant voices.
Crosstalk markers: where two or more people are speaking simultaneously, mark it. Do not transcribe it as if it were sequential dialogue.
Significant non-verbal cues: laughter, long pauses, expressions of surprise or discomfort, if your analytical questions make them relevant.

Auto-transcription tools handle single speakers well, but multi-speaker attribution is still imperfect. For focus groups, plan to review the transcript against the recording more carefully than you would for a one-to-one interview. Skimle's transcription feature supports multi-speaker identification, but verification is worth doing for group sessions specifically.

Step-by-step focus group analysis

Step 1: Read each transcript for overall impression

Before coding, read each group transcript from beginning to end. You are looking for the overall shape of the discussion: what the group focused on, where they agreed readily, where tension or hesitation surfaced, and whether any participants' contributions stood out as driving or disrupting the conversation.

Note anything that seems analytically significant in the margin, but do not code yet. This read-through builds the contextual knowledge you need to interpret codes correctly later.

Step 2: Code for content themes

Code each transcript for the themes relevant to your research question. The approach is similar to individual interview coding — mark passages, apply a label, and build a set of codes across all your transcripts. The complete guide to thematic analysis covers the coding process in detail.

The difference is that in focus groups, you are coding at the group level rather than the individual level for most purposes. A theme that appears in a discussion is a group finding; you do not need a separate code for each participant who contributed to it. Reserve individual-level coding for cases where individual positions are divergent or analytically significant.

Step 3: Code for group dynamics

Run a second pass specifically for dynamics:

Dominant contributions: passages where one participant's framing shaped the direction of subsequent discussion
Suppressed dissent: moments where a participant hedged, qualified, or withdrew a position after others responded
Consensus formation: moments where a group view crystallised — and whether it was genuine or conversational
Novel emergence: ideas that developed through the group interaction and would not have been articulated by any single participant alone

This layer of coding is what separates focus group analysis from interview analysis. It is often where the most analytically interesting material is.

Step 4: Compare across groups

If you have run multiple focus groups (which is standard practice), compare the theme structure across them. Questions to ask:

Which themes appeared consistently across all groups?
Which themes appeared only in one group? Why? Was it a different participant profile, a different moderator prompt, or a genuine finding about group-specific views?
Where did groups reach the same conclusions by different routes?
Where did groups reach different conclusions from the same starting point?

Cross-group comparison is the equivalent of cross-case analysis in interview research. It tells you how robust your findings are.

A worked example

Scenario: a consultancy has run three focus groups with eight participants each to understand how mid-market companies experience the procurement process for enterprise software. Groups were segmented by company size: 50–100 employees, 100–250 employees, and 250–500 employees.

After transcription and reading: the analyst notes that the smallest company group had a very different tone — more decision-maker certainty, less discussion of internal approval processes. The two larger company groups were more similar to each other, with more talk about sign-off chains and stakeholder alignment.

Content coding: recurring codes across all three groups include: budget scrutiny, integration concerns, trial period expectations, security and compliance questions, and internal stakeholder disagreement. These form the basis of the theme structure.

Dynamics coding: in the smallest company group, one founder-level participant drove nearly all the significant contributions. In the two larger groups, discussion was more distributed but consensus was reached more slowly. In Group 2, an IT director's comments on security visibly shifted the group's framing of risk.

Cross-group comparison: budget scrutiny appeared in all three groups but had different textures. In the smallest company, it was a binary question (can we afford this?). In the two larger companies, budget scrutiny was entangled with internal approval processes and the time cost of those processes.

Analytical finding: procurement friction is not primarily a cost problem — it is a coordination problem whose intensity scales with company size. This is a finding that would not have been visible without the cross-group comparison.

Write-up: the results section distinguishes between findings that held across all groups and findings that were group-specific, with explicit acknowledgement of how the group dynamics shaped the discussion in each session.

AI-assisted focus group analysis

AI tools handle the content layer of focus group analysis well: identifying what topics were discussed, how themes distribute across transcripts, and which passages illustrate each theme. Skimle reads each group transcript and builds a structured theme hierarchy across all sessions, with every insight traceable back to the verbatim passage it came from.

This is valuable for two practical reasons. First, it surfaces cross-group consistency and divergence automatically — you do not have to hold all the material in your head. Second, it applies consistent coding across all transcripts rather than allowing analytical drift across a long project.

The dynamics layer, including moderator effects, social pressure, consensus formation, and the emergence of novel group ideas, still requires human interpretation. The practical workflow is to use structured AI analysis for the content and apply your interpretive judgement to the dynamics. Once Skimle has given you a structured map of what was discussed, you can return to specific transcripts with focused questions: was this consensus genuine, or did it follow one dominant speaker's framing?

Common mistakes

Treating focus groups as discounted interviews: analysing focus group data by counting how many participants said something misses the social construction of meaning that is the method's core contribution.

Ignoring dominant voices: if three of your substantive codes in a group come from one participant, that group has not provided three independent pieces of evidence. Weight accordingly.

Reporting manufactured consensus as shared views: groups often reach agreement conversationally. Check whether the consensus held up or whether qualifications emerged if you look carefully at the transcript.

Running too few groups: one focus group is not a focus group study. The comparison across groups is where the method's analytical value lies. Three is a practical minimum for most research questions.

Not documenting moderator effects: different moderators prompt differently, which affects what emerges. Note this in your methods section so readers can assess it.

Writing up focus group findings

The write-up should reflect the group-level nature of the data. "Across all three groups, participants described procurement friction as a coordination problem" is the right framing. "Eight participants said X" is not — it imposes an individual-unit logic on data that was generated collectively.

Distinguish between findings that held across all groups (robust), findings that appeared in two groups (moderate robustness), and findings that appeared in only one group (contextual or emerging). This distinction is what makes focus group findings defensible.

For the methods section, document: how many groups, participant profile and recruitment approach, moderator and guide, whether sessions were recorded with consent, and how transcription was done. The analytical approach should state explicitly that you coded for both content themes and group dynamics.

For help structuring the write-up, see how to write a thematic analysis report.

Ready to bring structure to your focus group analysis? Try Skimle for free — upload your transcripts and see how AI-assisted analysis distributes themes across groups while preserving full traceability.

Related reading:

About the authors

Henri Schildt is a Professor of Strategy at Aalto University School of Business and co-founder of Skimle. He has published over a dozen peer-reviewed articles using qualitative methods, including work in Academy of Management Journal, Organisation Science, and Strategic Management Journal. His research focuses on organisational strategy, innovation, and qualitative methodology. Google Scholar profile

Olli Salo is a former Partner at McKinsey & Company where he spent 18 years helping clients understand the markets and themselves, develop winning strategies and improve their operating models. He has done over 1000 client interviews and published over 10 articles on McKinsey.com and beyond. LinkedIn profile