What is a Focus Group?

A focus group is a moderated discussion, typically with five to ten participants, exploring perceptions, attitudes, and reactions to a topic: a product concept, a brand, a message, a category of behaviour. A skilled moderator works from a discussion guide, keeps the conversation on course, draws out quiet voices, and manages dominant ones. Sessions are recorded and analysed for themes, and multiple groups are usually run to distinguish a group's quirks from a genuine pattern.

The method's defining feature, and its double edge, is the group itself. Participants react to each other, not just to the moderator: they build on ideas, disagree, remind each other of forgotten experiences, and reveal the social dynamics around a topic. That interaction is data no one-to-one method produces. It is also a contamination mechanism, for reasons covered below.

A Brief History

The focus group descends from the "focused interview" developed by sociologist Robert K. Merton and colleagues at Columbia University in the 1940s, originally to study audience responses to wartime radio broadcasts and propaganda films. Market researchers adopted and renamed the technique after the war, and by the 1980s the mirrored viewing room with one-way glass had become a fixture of advertising and consumer-goods research, and of popular culture's image of what "research" looks like. The digital era moved many groups online, trading the physical room for video calls, which widened recruitment and lowered costs while making group dynamics somewhat harder to read.

What Focus Groups Are Good For

Used honestly, focus groups are an exploratory instrument. They excel at mapping the language people use around a category, surfacing the range of attitudes that exist (not their prevalence), generating hypotheses worth testing properly, exploring social norms around how people talk about money, health, status, and other topics where the group setting itself is informative, and reacting to early-stage stimulus like positioning, messaging, or concept directions where breadth of reaction matters more than depth. They are a starting pistol rather than a finishing line: findings should graduate to interviews, usability testing, or quantitative validation before anyone bets money on them.

Where They Mislead

The group that makes focus groups valuable also makes them dangerous. Dominance effects: one confident voice can tilt an entire session, with subsequent speakers anchoring to the first opinion voiced. Social desirability: people perform for an audience, professing healthier eating, higher-minded media habits, and more rational purchase decisions than their behaviour shows. Groupthink: consensus emerges because disagreement is socially expensive, not because minds changed. The articulation trap: groups reward opinions that are easy to say, and punish preferences people can't explain. This is why the Nielsen Norman Group has long cautioned against using focus groups to evaluate usability: watching people use something and hearing a group talk about it produce reliably different verdicts. Product history is littered with ideas that focus groups panned and users later loved, and vice versa. That is not an argument against the method; it is an argument against asking it questions it cannot answer.

How to Run One Well

1. Use it for the right question.
Exploration, language, reaction, range: yes. Usability, prevalence, purchase prediction: no.

2. Recruit for productive homogeneity.
Within a group, enough shared context that people speak freely (mixing bosses with their subordinates kills candour); across groups, the diversity your question needs. Tight screeners matter as much here as in any method.

3. Invest in moderation.
The moderator is the method. Redirecting a monologuer without shaming them, inviting the quiet without spotlighting them, probing a glib answer, sitting in silence: these are practised skills, and the gap between a good and bad moderator is the gap between data and noise.

4. Run multiple groups.
One group is an anecdote with a conference table. Patterns that recur across three or four independently recruited groups deserve attention; anything seen once does not.

5. Analyse the recording, not the memory.
The most quotable moment is rarely the most representative. Transcribe, code, and look for themes across groups with the same rigour you would apply to interview analysis, and report the disagreements, not just the consensus.

Focus Groups and Their Modern Alternatives

Much of what teams once used focus groups for now has sharper tools. For concept reactions at breadth, unmoderated studies where dozens of participants respond to stimulus individually (in platforms like Ballpark, often via video answers) deliver the reaction range without the contamination of a shared room. For depth on sensitive topics, one-to-one interviews. For behaviour, usability testing and analytics. What survives for the focus group is what only a group can do: show you how people negotiate a topic together.

The Benefits

Focus groups are fast (a week of sessions can canvass forty people) and the group setting surfaces contrasts, vocabulary, and social dynamics that individual methods miss. They are also unmatched as an organisational empathy tool: stakeholders behind the glass, or on the call, hear customers argue about their product in real time, which lands harder than any report.

The Limitations

Everything above: dominance, performance, groupthink, and the articulation trap, plus small unrepresentative samples and heavy dependence on moderator skill. Focus groups measure what people will say about a topic in front of strangers. Sometimes that is exactly the question, and then nothing else will do. Mistake it for what people privately believe, or what they will actually do, and the method will mislead you with total confidence.

The Takeaway

The focus group is neither the caricature its critics mock nor the oracle its schedule-pressed fans want. It is a specialist exploratory tool: superb at language, range, and social dynamics, unreliable at behaviour and prevalence. Ask it the right questions, moderate it well, run enough groups to see patterns, and then verify what you heard with methods that watch what people do.