How Many Focus Groups Do You Need? 2026 Evidence

Qualitati Research Team · 2026-08-14 · 10 min read

Short answer. Empirical evidence points to fewer focus groups than most rules of thumb suggest. Guest, Namey and McKenna (2017) found more than 80% of themes were discoverable in two to three focus groups, and 90% within three to six. Three groups surfaced every one of the most prevalent themes. Plan three to six groups per distinct segment, not per study.

How many focus groups do you need? The evidence, not the folklore

Ask five researchers how many focus groups you need and you will get five confident answers: "three to five," "until saturation," "one per segment plus a spare." Most of these are inherited convention rather than measured findings. The question — how many focus groups do you need — has actually been tested empirically, and the answers are more useful than the folklore.

This guide summarizes what the published evidence says, explains why "per study" is the wrong unit of planning, and gives you a decision matrix you can defend to a stakeholder who wants to cut your sample in half.

Key takeaways

More than 80% of themes appeared within two to three focus groups; 90% within three to six (Guest, Namey & McKenna, 2017).
Focus groups reach thematic saturation in fewer data collection events than individual interviews — a median of 3 events for 80% saturation versus 8 for interviews (Namey et al., 2016).
The planning unit is the segment, not the study. Three groups covering three different populations is one group per population, which is not saturation of anything.
Saturation is contested as a stopping rule for reflexive thematic analysis (Braun & Clarke, 2021). Use it as a planning heuristic, not a claim of completeness.
High-prevalence themes emerge early; low-prevalence and minority-perspective themes do not. If your research question is about the margins, sample-size math based on common themes will underserve you.

What the saturation studies actually measured

The most-cited empirical answer comes from Greg Guest, Emily Namey and Kevin McKenna, published in Field Methods in 2017. They analyzed a corpus of 40 focus groups on health-seeking behaviors among African American men in Durham, North Carolina, then measured how many themes had appeared after each successive group.

Their headline results: over 80% of all themes in the full data set were discoverable within two to three groups, and 90% within three to six. Three focus groups were sufficient to identify all of the most prevalent themes. The authors were explicit that these numbers sit below many conventional recommendations.

A companion study by Namey, Guest, McKenna and Chen in the American Journal of Evaluation (2016) compared the cost-efficiency of focus groups against in-depth interviews on the same corpus. The median number of data collection events needed to reach 80% saturation was 3 for focus groups versus 8 for interviews; for 90%, it was 5 versus 16. Per event, focus groups covered ground faster — a useful counterweight to the assumption that interviews are always the more rigorous default.

The important caveat: prevalence

Saturation curves are steep at the start and flat at the end for a reason. Themes that most participants hold come up almost immediately. Themes held by a minority — the accessibility edge case, the objection only power users raise, the cultural framing that one subgroup brings — sit in the long tail. A study that stops at three groups reliably captures the mainstream and unreliably captures the margins.

This is the practical translation: match your group count to whether your research question is about the center or the edges of your population.

Why "per study" is the wrong unit

The saturation figures above were computed within a single, reasonably homogeneous population. They do not license running three groups across three different markets and declaring saturation.

Plan in segments. A segment is any grouping where you have reason to expect the discussion to differ materially — role, market, tenure, product tier, language, or care setting. Then apply the group count within each segment.

Qualitati Focus Group Sample-Size Decision Matrix

Your situation	Groups per segment	Why
Exploratory scoping; you need the dominant themes only	2–3	Captures 80%+ of themes and all high-prevalence ones
Standard discovery feeding a design or strategy decision	3–4	Approaches 90% theme coverage within segment
Research question targets minority views, edge cases, or dissent	5–6+	Long-tail themes need more groups to surface
Comparative claim across segments ("A differs from B")	3+ per segment, minimum 2 segments	One group per segment cannot separate group effect from segment effect
Sensitive topic where disclosure is suppressed in groups	Reconsider the method	Individual interviews usually outperform groups here
High-stakes, publication-bound, or regulated research	Plan 4–6, monitor and report a saturation curve	Reviewers expect a stated, evidenced rationale

A simple planner: total groups = segments × groups-per-segment. Three markets at three groups each is nine groups, not three. If nine is unaffordable, the honest move is to cut segments, not to cut groups per segment — the second choice quietly destroys your ability to say anything about any segment.

Saturation reporting checklist

State the segment definition and how many groups ran within each.
Report when new themes stopped appearing, by group number — not just the assertion "saturation was reached."
Distinguish code saturation (no new codes) from meaning saturation (no new understanding of existing codes).
Name the themes that appeared only once. These are candidates for follow-up, not noise to discard.
If you stopped for budget or calendar reasons, say so. A stated constraint is more credible than a saturation claim that was never measured.
Note whether any group was dominated by one or two speakers, which inflates apparent saturation.

The methodological objection you should know about

Virginia Braun and Victoria Clarke, in Qualitative Research in Sport, Exercise and Health (2021), argue that saturation is conceptually incoherent as a sample-size rationale for reflexive thematic analysis. Their case: themes in that tradition are generated through interpretation, not discovered lying in the data waiting to be counted. If themes are constructed, there is no fixed total to be a percentage of, and "we found 90% of themes" is a category error.

This does not make the Guest et al. numbers useless. It does change how you should cite them. Treat the saturation evidence as a planning heuristic — a defensible basis for choosing an initial group count and budget — rather than as a post-hoc claim that your analysis is complete. If your study is explicitly reflexive-TA, follow Braun and Clarke's guidance and justify your sample by information power and research design instead.

Where Qualitati fits

Qualitati is an AI user research platform for product, UX, and customer insights teams. Two parts of it bear directly on group count.

First, AI-moderated focus groups change the cost curve. Traditional groups are expensive per session, which is what makes "cut it to three" a standing temptation. When the moderator cost of an additional group is lower, running the full segments × groups-per-segment plan becomes realistic, and you can afford the extra groups that minority themes require. Qualitati's AI moderator probes, brings in quiet voices, checks consensus, and counters groupthink — the last two matter because a group dominated by one voice produces a falsely flat saturation curve.

Second, ThemeLens makes saturation observable rather than asserted. It runs a map-reduce thematic analysis across up to 100 transcripts, maps codes to your research questions, and synthesizes themes with participant-anchored quotes. That lets you look at when each theme first appeared across your groups and report an actual curve, instead of a claim.

Qualitati supports research in 10 languages and publishes per-credit usage rates, so multi-segment designs can be costed before you commit. Start free with 30 credits, no credit card required.

Limitations and when not to use this approach

Do not use focus groups at all when the topic is stigmatized, legally sensitive, or when participants have a professional reason to manage their image in front of peers. Group disclosure norms suppress exactly the material you need. Individual interviews are the right instrument.

Do not treat the two-to-three group figure as universal. It comes from one corpus, one population, one topic domain. Homogeneous groups on a narrow topic saturate faster; heterogeneous groups on a broad topic saturate slower.

Do not use group count as your only quality lever. Six weak groups with a leading discussion guide produce less than three well-moderated ones. Sample size cannot rescue instrument design.

Finally, if you use AI moderation or AI-assisted analysis, keep a human in the loop on theme construction. Automated theme extraction is a strong first pass and a poor final authority, particularly for the low-prevalence themes that motivated running extra groups in the first place.

FAQ

How many focus groups do you need for a typical UX study?

Three to four groups per distinct segment is a defensible default for discovery work. That range approaches 90% theme coverage within a segment based on Guest, Namey and McKenna's 2017 findings, while staying affordable.

Is three focus groups really enough?

Three groups reliably surface the most prevalent themes within one homogeneous segment. It is not enough if you need minority perspectives, if you are comparing segments, or if your population is heterogeneous.

How many people should be in each focus group?

Conventional practice is six to eight participants, balancing airtime per person against range of perspectives. This is convention rather than an empirically derived optimum, and smaller groups are common for specialist or professional populations.

Do focus groups or interviews reach saturation faster?

Per data collection event, focus groups. Namey and colleagues (2016) reported a median of 3 focus groups versus 8 interviews to reach 80% saturation on the same corpus. Per participant, the comparison is less favorable to groups.

Should I still claim saturation in my write-up?

Report evidence rather than a claim. Show when new themes stopped appearing, define what you mean by saturation, and note your constraints. If you are doing reflexive thematic analysis, consider justifying sample size by information power instead.

Can AI-moderated focus groups change the number I need?

Not the underlying theme-emergence curve, which is a property of the population and topic. What changes is affordability — lower per-session cost makes running the full number of groups your design requires more feasible.

Bottom line

How many focus groups do you need? Plan three to six per segment, multiply by your segment count, and weight toward the higher end when your research question concerns minority views rather than mainstream ones. Treat published saturation figures as a budgeting heuristic, measure and report your own theme-emergence curve, and never let one group per segment masquerade as a comparative finding.

Ready to run more groups without the per-session cost that forces the compromise? Start free with 30 credits, or view transparent pricing. Related reading: Focus Groups vs Interviews: What the Evidence Says, Dominant Participants in Focus Groups, and AI-Moderated Focus Groups: A 2026 Guide.

Sources

Guest, G., Namey, E., & McKenna, K. (2017). How Many Focus Groups Are Enough? Building an Evidence Base for Nonprobability Sample Sizes. Field Methods, 29(1). Link
Namey, E., Guest, G., McKenna, K., & Chen, M. (2016). Evaluating Bang for the Buck. American Journal of Evaluation. Link
Braun, V., & Clarke, V. (2021). To saturate or not to saturate? Qualitative Research in Sport, Exercise and Health, 13(2). Link
Gandy, K. (2024). How many interviews or focus groups are enough? Evaluation Journal of Australasia. Link

Last updated: August 14, 2026. This is an independent editorial summary of publicly available research; methodology claims about specific studies should be verified against the original papers before citation.