Last updated: September 27, 2026
\n\nA focus group is a qualitative research method in which a moderator guides a small group of pre-screened participants, usually 5 to 8 people, through a semi-structured discussion for 60 to 120 minutes. Its defining feature is interaction: participants react to, build on, and challenge each other, so the data shows how opinions form, not only what individuals think.
\n\nA focus group is a moderated group discussion used to generate qualitative data. Kitzinger's classic BMJ paper describes it as a form of group interview that uses communication between participants to produce data (Kitzinger, 1995). That is the difference from simply interviewing several people at once. In a focus group, participants hear each other, and the researcher studies how views are voiced, negotiated, and sometimes revised in a social setting.
\nThe method comes out of mid-20th-century sociology and communication research. Today it is standard in health research, education, public policy, market research, and UX research. Academic teams use it to explore a topic before designing a survey, to test the wording of an intervention, or to understand shared norms in a community. Research agencies use it for concept and message testing.
\n\nThis guide is for PhD students writing a methods chapter, academics planning a funded study, and qualitative or market researchers deciding between focus groups and interviews. If you already know the basics and need a full planning walk-through, see our focus group methodology guide.
\n\nShort answer: 5 to 8 participants for most research focus groups. Groups of 10 to 12 are common in commercial concept testing, but they give each person little airtime.
\nThe range is a trade-off. With fewer than about five people, the discussion can stall and depends heavily on one or two voices. With more than about eight, quieter participants are crowded out and the moderator spends the session managing turn-taking instead of probing. Kitzinger (1995) gives four to eight as the ideal, and Krueger and Casey's Focus Groups: A Practical Guide for Applied Research recommends five to eight. Mini-groups of three or four suit expert participants or topics where each person has a lot to say.
\n\nPlan several groups, not one: a single group can be skewed by the people who happen to attend. The best empirical benchmark is Guest, Namey and McKenna (2017, Field Methods). In their analysis of 40 focus groups, more than 80% of all themes were identified within two to three groups, 90% within three to six, and three groups were enough to capture all of the most prevalent themes. Treat these numbers as a starting point for a fairly homogeneous population. Each distinct segment you want to compare usually needs its own set of groups. For a fuller treatment, read how many focus groups you need and our guide to data saturation.
\n\nShort answer: use a focus group when the social exchange is part of what you want to study. Use individual interviews when you need each person's full story or the topic is private.
\n| Criterion | Focus group | One-on-one interview |
|---|---|---|
| Unit of data | Group interaction and shared talk | Individual account |
| Best for | Reactions, shared language, norms, agreement and conflict | Depth, personal histories, decision processes |
| Sensitive topics | Risky: peers can suppress candor, though some groups encourage disclosure | Usually safer; privacy supports disclosure |
| Airtime per participant | Roughly 10 to 15 minutes in a 90-minute group of 6 to 8 | The full session |
| Main bias risk | Conformity, dominant voices, social desirability | Interviewer effects, social desirability |
| Analysis focus | Content plus interaction | Content within and across cases |
The evidence on sensitive disclosure is mixed. Some studies find more disclosure in groups, others in interviews. We review both sides in focus groups vs interviews: what the evidence says. For the interview side of the choice, see what a semi-structured interview is.
\n\nAn original Qualitati framework. Score your study on each question. If two or more answers point to \"interview\", default to one-on-one interviews or combine the two methods.
\n| Question | Points to focus group if… | Points to interview if… |
|---|---|---|
| Is the social exchange part of the phenomenon? | Yes: norms, shared meanings, collective reactions | No: individual experience or decisions |
| Would peers inhibit honesty? | No, or participants share an identity that encourages talk | Yes: stigma, workplace hierarchy, health, money |
| Do you need each person's full narrative? | No, breadth of views is enough | Yes: life histories, journeys, critical incidents |
| Can you schedule 5 to 8 matched people at once? | Yes | No: dispersed, senior, or hard-to-reach participants |
| Will you analyze interaction? | Yes, and you have a plan to code it | No, you only need content themes |
Short answer: start with a standard approach such as thematic analysis, then add an interaction layer. Code who said what, how others responded, and whether apparent consensus was real.
\nKitzinger notes that analyzing focus groups is largely the same as analyzing other qualitative data, with extra attention to group dynamics. She stresses deviant case analysis: minority opinions and examples that do not fit the emerging pattern (Kitzinger, 1995). Onwuegbuzie, Dickinson, Leech and Zoran (2009, International Journal of Qualitative Methods) take this further with micro-interlocutor analysis. It records which participants respond to each question, in what order, and with what verbal and non-verbal signals, so that a \"consensus\" voiced by two people is not reported as the group's view. More recently, Dedios-Sanguineti and colleagues (2025, IJQM) proposed a seven-indicator model for assessing meaningful interaction and applied it to 12 asynchronous focus groups run over WhatsApp in Colombia. Interaction analysis therefore applies to online and asynchronous formats as well as in-person rooms.
\n\nAn original Qualitati template. Code each substantive turn on these five dimensions alongside your thematic codes:
\n| Dimension | What to code | Why it matters |
|---|---|---|
| Participation | Who speaks, how often, in what order | Shows whether a theme belongs to the group or to one voice |
| Response type | Agree, extend, qualify, challenge, change topic | Separates real consensus from passive assent |
| Shift | Participant revises a stated view | Shows persuasion or conformity at work |
| Minority view | A dissenting point, and whether the group takes it up or drops it | Supports deviant case analysis |
| Shared language | Terms and metaphors others adopt | Shows collective meaning and the vocabulary to use in later instruments |
When you report findings, state the level of each claim: \"raised in all four groups\", \"raised by one participant and contested\", or \"voiced by two participants, others silent\". That precision is what makes focus group findings credible to reviewers. For writing up themes, see our thematic analysis guide. For how LLMs perform on this kind of data, see can AI do thematic analysis on focus groups?
\n\nAI now appears in focus groups in three roles. In a June 10, 2026 arXiv paper, Wang and Dow organize AI support along two axes: the AI's role (tool, co-host, or host) and the modality (text, voice, or embodied). They also flag interactional trade-offs that still need evaluation. We summarize the paper in the role x modality playbook.
\nQualitati is a European, budget-friendly qualitative research platform for universities and research companies, with data privacy and GDPR as central priorities. For focus group studies it offers:
\nQualitati supports your methodological choices; it does not make them for you. Group composition, the questioning route, and the interpretation of dynamics remain the researcher's responsibility.
\n\nA focus group is a guided group discussion, usually 5 to 8 people led by a moderator, used to explore opinions, reactions, and shared language about a topic. Its value comes from participants responding to one another.
\n\nMost methodologists recommend small groups: Kitzinger (1995) gives four to eight, and Krueger and Casey recommend five to eight. Commercial groups sometimes run to 10 to 12, which reduces each person's airtime.
\n\nIn Guest, Namey and McKenna's (2017) analysis of 40 groups, more than 80% of themes appeared within two to three groups and 90% within three to six. Plan at least three groups per distinct segment and check for saturation.
\n\nAn interview produces one person's account. A focus group produces interaction: agreement, disagreement, and shared language. Use groups for reactions and norms, and interviews for depth and private topics.
\n\nUse a standard method such as thematic analysis, then add an interaction layer: who spoke, how others responded, where views shifted, and which minority views were dropped. Report each theme's level of support across participants and groups.
\n\nSynthetic participants cannot. They simulate responses and are useful only for exploration. AI-moderated groups with real participants are a newer format whose effects on group dynamics are still being studied, so keep a researcher in the loop.
\n\nA focus group is a moderated discussion among 5 to 8 participants whose value lies in their interaction. Use it to study reactions, shared language, and where a group agrees or splits. Run at least three groups per segment, code the interaction as well as the content, and choose one-on-one interviews for sensitive or deeply individual topics. Designed and analyzed this way, a focus group is still one of the most useful tools in qualitative research.
\nStart free with 30 credits, no credit card required. Create an account to run an AI-moderated or synthetic focus group and analyze transcripts with ThemeLens, or view transparent pricing first.
\n","related":[]}