A facilitated group discussion of 6–10 people. Useful for surfacing language and reactions — but vulnerable to groupthink and dominant voices, so it's the most misused method in research.
▸ Try the interactive toolA focus group is a structured group discussion in which a facilitator guides 6–10 participants through a set of topics, capturing their collective responses. Developed in the 1940s for market research, it became a standard tool — and also one of the most misapplied methods, routinely used for questions it's structurally bad at answering.
Focus groups are genuinely useful for a narrow set of jobs: surfacing the language people use, observing reactions and group dynamics, and generating a breadth of ideas quickly. But they're poor for assessing real behaviour or individual preference, because group settings introduce groupthink, dominant participants, and social desirability — people say what sounds good in front of others. Knowing what a focus group can and can't tell you is most of the skill.
Good for: surfacing vocabulary, observing reactions, broad idea generation, group dynamics.
Bad for: predicting behaviour, measuring true individual preference, sensitive topics — group effects distort all three.
The focus group's social setting is both its unique strength (you see group reaction) and its fatal weakness (the group distorts individual truth). Misuse comes from forgetting the second part.
| The shortcut | What it costs | What it gives you instead |
|---|---|---|
| Using them to predict behaviour | Group claims about future behaviour are especially unreliable. | Reserve focus groups for language and reaction, not behaviour prediction. |
| Groupthink | Participants converge on a dominant view; dissent stays quiet. | Skilled facilitation surfaces dissent; interviews capture individual truth. |
| Dominant voices | One confident participant skews the whole session. | Facilitation manages dominance; small groups limit it. |
| Social desirability | People say what sounds good in front of others. | Recognising this keeps you from trusting performed preferences. |
Choose a focus group when you want language, reactions, or idea breadth — not when you need to know what individuals actually do or truly prefer. The method choice is the first and biggest decision.
6–10 participants who fit the segment. Too much diversity within one group can suppress candour; match the group to the topic.
A skilled facilitator draws out quiet participants, manages dominant ones, and probes consensus that forms too easily. Passive facilitation lets groupthink win.
When the group agrees quickly, treat it with suspicion and dig for dissent. The convergence is often social, not genuine.
Treat focus-group output as hypotheses about language and reaction — to validate with interviews, behaviour, or experiments — not as findings to act on directly.
A team ran a focus group to decide which of two features to build, asking the group which they'd use. The group converged enthusiastically on one option, and the team built it. Adoption was poor — the feature people had endorsed in the room wasn't the one they actually used.
The focus group had been used for exactly the job it's worst at: predicting individual behaviour. In the group, a confident participant had championed the losing option and others had agreed rather than dissent. What the session was actually good for — the vivid language people used to describe the underlying problem — went unused. Interviews or a fake-door test would have predicted real behaviour far better.
The deliverable is captured language, reactions, and ideas — explicitly framed as signals to validate, not conclusions to act on.
| Use it for | Don't use it for |
|---|---|
| The words people use for the problem | Predicting what they'll actually do |
| Gut reactions to a concept | Measuring true individual preference |
| Generating a breadth of ideas | Deciding which feature to build |
| Observing group dynamics | Sensitive or private topics |
The focus group's social setting is simultaneously its only unique strength and its disqualifying weakness. It's the one method that shows you group reaction — and the group is exactly what distorts individual truth.
This makes method selection the entire game. A focus group used for language and reaction is a fine, fast tool; the same focus group used to predict behaviour or measure preference is worse than useless, because it produces confident-sounding answers that are systematically wrong. The groupthink, dominant voices, and social desirability that corrupt behaviour questions are baked into the format and can't be facilitated away entirely. A PM who reaches for a focus group should first ask whether the question is about words and reactions (use it) or about behaviour and preference (use interviews, analytics, or an experiment instead).
The single most common misuse. Group claims about future behaviour are unreliable — test behaviour directly.
Quick consensus is often social, not genuine. Facilitate for dissent.
One confident participant can skew everything. Manage the room actively.
Focus-group signals are hypotheses to validate, not findings to act on.
Where focus groups blur individual truth, one-on-one interviews (Tool 05) preserve it. Often the better default.
Focus-group reactions should be confirmed with real behaviour via fake doors or lean experiments (Tools 15, 18).
The vocabulary a focus group surfaces is genuinely useful for messaging and positioning (Module 2).
Groupthink and social desirability are specific biases covered in Tool 25.
List three questions a team might bring to a focus group. For each, decide: is this about language/reaction (suitable) or behaviour/preference (unsuitable)?
For any unsuitable one, name the method that would actually answer it — interview, analytics, or experiment.
If most of your real product questions turn out unsuitable for focus groups, you've discovered why the method is so often misused — it's reached for by default, not by fit.