What Is a Focus Group? Asking People Together and the Problems That Causes
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
A focus group puts six to ten people in a room with a moderator to discuss a product, a policy or an idea, on the reasoning that discussion surfaces reactions a questionnaire would miss. It does, and it also introduces problems that individual interviews do not have, which is why the method is genuinely useful for some questions and misleading for others.
What it is good at
The method suits exploration rather than measurement, and its distinctive value is specific:
- •Generating the vocabulary people actually use about a subject, which is essential before writing a survey, since a questionnaire written in the researcher's language measures the researcher's assumptions
- •Surfacing considerations nobody anticipated, because participants raise things the moderator did not think to ask
- •Revealing how opinions are formed and justified when challenged, which an individual interview cannot show
- •Observing reactions to something physical, including packaging, prototypes and advertisements, where the first unguarded response matters
- •Exploring disagreement, since the moderator can probe why two participants see the same thing differently
- •Speed and relative cheapness compared with an equivalent number of individual interviews
What goes wrong in a group
The group setting introduces well-documented distortions. Dominant participants take a disproportionate share of the talking and shift the discussion, while quieter ones contribute less than their views warrant, which is a straightforward failure of representation within the room. Conformity effects mean people adjust their stated views towards an emerging consensus, an effect demonstrated in classic experiments where participants gave obviously wrong answers about line lengths when everyone before them had. Social desirability bias is amplified by an audience, so participants report what sounds acceptable rather than what they do, which is why focus groups consistently understate consumption of things people feel judged for. Moderator effects are substantial, since phrasing, tone, follow-up and even nodding shape responses. And participants asked to predict their own future behaviour do so badly, which is the deepest problem, because the question a client wants answered is usually whether people will buy something and the honest answer is that they do not know.
The famous failures
Several commercial disasters are attributed to research of this kind and the attributions are worth examining rather than repeating. The reformulation of Coca-Cola in 1985 followed extensive taste testing in which the new formula was preferred, and the testing measured a sip rather than a purchase and never asked how people would feel about the original being withdrawn, which was the variable that mattered. Several celebrated products were reportedly rejected by early research, and the pattern behind such cases is consistent: research is reliable about reactions to things people can evaluate from experience and unreliable about genuinely novel propositions, because participants judge unfamiliar things against what they already know. That limitation is frequently summarised with a quotation about faster horses attributed to Henry Ford, which he probably never said, and the underlying point stands regardless of its provenance. The practical conclusion drawn in the industry is that focus groups should inform decisions rather than make them, and that using them to validate rather than to explore is where they mislead.
Doing it better and the alternatives
Methodological improvements address specific failures. Recruiting for homogeneity within a group and heterogeneity between groups reduces the inhibition of speaking across social difference. Running several groups rather than one guards against a single unusual dynamic being mistaken for a finding. Structured techniques including individual written responses before discussion capture views before conformity operates. Skilled moderation actively manages dominant participants and probes silence. Analysis should treat the discussion as qualitative data to be coded rather than as a set of quotations to be selected, since selective quotation is how focus groups are most commonly abused. The alternatives suit different questions: individual depth interviews avoid group effects, ethnographic observation records what people do rather than what they say, behavioural experiments and trial sales measure actual choices, and large-scale surveys measure prevalence, which focus groups never can, since a group of eight people is not a sample of anything.
The takeaway
Focus groups are exploratory, valuable for surfacing the vocabulary and considerations people actually use and for observing reactions to something tangible. The group setting introduces dominance effects, conformity, social desirability bias and moderator influence, and participants predict their own future behaviour badly. The 1985 Coca-Cola reformulation tested a sip rather than a purchase and never asked about withdrawing the original. A group of eight people measures nothing, so the method informs decisions rather than making them.