MarketingFramework / modelModelIntermediate

Focus groups

How can focus groups improve people, teams, or organisational effectiveness?

IntermediateOperationalIndividual3 min read

Contents

A focus group is a form of qualitative research where a group of specially selected or randomly selected individuals come together to discuss a specific topic.

A focus group is a moderated qualitative discussion with participants selected because their experience or characteristics are relevant to a research question. The interaction among participants is part of the evidence: people compare, challenge and develop ideas in ways that an individual interview cannot reproduce.

When to use it

Use focus groups to explore language, beliefs, experiences, decision criteria and the range of views surrounding a topic. They are especially useful for developing hypotheses, refining a product or service, interpreting survey results and understanding how opinions are negotiated socially.

A typical session lasts between 90 minutes and 2 hours and covers five or six substantive questions. It can produce insight quickly, but it is not suitable for estimating how common a view is in a population or for subjects that participants cannot discuss safely in a group.

Origins

The modern focus group grew from the “focused interview” developed by sociologists Paul Lazarsfeld, Robert K. Merton and colleagues during mid-twentieth-century research into audience responses to radio and wartime communication. The method combined a group format with open, non-directive probing of participants’ interpretations. Marketing researchers adopted it widely, and it later returned to social science, evaluation, design and organisational research. Contemporary practice emphasises that the discussion is socially produced evidence, not a collection of independent survey answers.

What it is

A moderator introduces the purpose, creates safe discussion norms, asks open questions and probes differences without steering the group toward a preferred conclusion. An observer may manage notes and logistics. Useful formats include:

Two-way group: one group observes another discussion and then examines what it noticed, with consent from everyone involved.

Dual moderator: one moderator manages participation and rapport while another tracks the research topics.

Opposing moderator: moderators introduce contrasting interpretations to test assumptions, while making clear that the positions are prompts rather than hidden agendas.

Respondent moderator: a participant temporarily proposes or leads a question, which can reveal the group’s own priorities.

Client observation or participation: a commissioning team observes transparently or joins under a disclosed role. Covert client participation is deceptive and should not be presented as a route to more candid data.

Mini group: 4 or 5 participants allow more individual depth than the more common 8 to 12, and may suit technical or sensitive subjects.

Why it matters

Focus groups reveal how people describe an issue, where views converge or conflict and which social cues shape the conversation. Product, service and change teams can hear needs and objections before committing to a design.

The output is interpretive. Silence does not mean agreement, confident speakers are not necessarily representative and non-verbal behaviour is ambiguous. Strong analysis preserves disagreement and context instead of collapsing the group into one supposed opinion.

How to use it

Define the decision and what the group can realistically illuminate. Recruit purposively across relevant experiences, avoiding relationships or power differences that could make honest participation unsafe. Obtain informed consent, explain recording and confidentiality limits, and provide an accessible way to participate or withdraw.

Create a discussion guide that moves from easy, broad questions toward the core topic. Use neutral prompts, activities or stimuli only when they serve the question. Pilot ambiguous wording and plan how the moderator will handle dominance, distress, disclosure and disagreement.

During the session, invite quieter voices without forcing disclosure, probe concrete examples and test whether apparent consensus is genuine. Afterward, compare groups, cases and contradictions. Link each theme to supporting evidence and state what the sample cannot establish.

Possible data sources

Data may include moderator and observer notes, flip charts, participant-created artefacts and consented audio or video. Transcripts can support Text Analytics, Voice Analytics or Sentiment Analysis, but automated outputs must be checked against tone, interaction and context.

A short questionnaire may capture individual background or a private response before group influence occurs; see Quantitative Surveys and Qualitative Surveys. Store recordings and identities separately, restrict access and honour the stated retention period.

How difficult or costly is it to collect?

One group is relatively straightforward to convene, but credible research requires skilled moderation, careful recruitment, accessible participation, transcription, analysis and privacy controls. Multiple locations, languages or audience segments increase both cost and interpretive complexity.

Practical example

A business considering organisational change could conduct separate groups across functions, grades and affected roles. Separating groups where power could inhibit candour may reveal different concerns about workload, fairness and implementation.

The team would compare themes rather than attribute comments to individuals, test findings through other evidence and explain which decisions changed. A focus group should not substitute for formal consultation, representation or a confidential reporting channel.

Top practical tip

Design the group around the research decision. Recruit for relevant variation, use a neutral moderator and end by inviting missing or minority views. Analyse interaction as well as topics: who agreed, challenged, withdrew or changed position, and under what prompt?

Top pitfall

Do not mistake a lively discussion for representative truth. Dominant participants, conformity, social desirability and the moderator’s wording can shape the result. Never insert an undisclosed client as a participant. Preserve dissent, disclose the sample and validate important conclusions with other methods.