Focus groups: how they work and which decisions they support

1. What a focus group is and what it is not

A focus group is a moderated group discussion of about two hours with 5-8 participants selected on precise criteria, led according to a discussion guide of open questions. The aim is not to reach a consensus, nor to count how many prefer option A, but to understand how people think, what they feel, which words they use and what makes them choose or avoid something. The interaction between participants is the instrument: one person's remark triggers memories and reactions in the others, and the moderator watches an opinion form and change in real time.

A focus group is not a small survey. Eight people, however well chosen, do not statistically represent any population, and the results are not expressed in percentages. It is not a brainstorming session, nor a product presentation with questions at the end, nor a meeting in which the client explains to participants why its product is good. It is a qualitative research method with rules of its own, and breaking them produces exactly what happens most often: a pleasant conversation from which every observer retains what they wanted to hear.

2. When the focus group is the right method

The focus group answers the question "why?" and the question "how do people talk about this?". The typical situations:

• Testing concepts, messages and packaging before launch: what people understand from them, what attracts them, what confuses them, what is missing.
• Explaining a quantitative result: the survey shows satisfaction fell by ten points, and the focus group discovers what changed in the customer experience.
• Preparing a questionnaire: focus groups identify the themes, the answer options and the wording people actually use, so that the survey measures the right things in the right language.
• Understanding a category of products or services the team is not familiar with: how it is bought, who decides, what matters.
• Social programmes and information campaigns: what members of a community think about a topic, what barriers they face, which messages and messengers would work.

The focus group is not the right method when a decision is wanted by "vote", when consensus is sought, when the topic is intimate or confrontational, when participants would not speak honestly in front of others, when knowledge is being tested, or when the result must be a figure. For sensitive topics and for experts, the right method is the in-depth interview. For figures, a survey.

3. How many participants and how many groups

Group size. The standard is 5-8 participants. Below five, the discussion has too few perspectives and stalls easily; above ten, some participants stay silent and the others start talking among themselves. For complex or emotional topics, or for participants with expertise, mini-groups of 4-6 people are preferred, in which everyone has time to develop their ideas. The sign that a group is too large: participants whisper to their neighbours.

Number of groups. There is no statistical formula; the criterion is saturation: groups are held until the discussions stop bringing new themes. The practical rule is 3-4 groups for each category of participants, after which the researcher judges whether one more is needed. A study with two categories, for example current and former customers, means 6-8 groups. In most applied studies, saturation appears between 3 and 12 groups in total.

Homogeneity. Each group brings together people who are similar on the criterion relevant to the study: age, status, experience with the product, place of residence. People speak openly with those they perceive as similar and censor themselves in front of those who are different. Categories that must be compared, for example urban and rural, young and old, users and non-users, have separate groups and are never mixed in the same room.

4. Recruitment: who ends up in the room

Recruitment is the stage that decides the most and shows the least. Participants are selected through a screening questionnaire that checks the study's criteria: that they belong to the target category, that they have the relevant experience, that they do not work in marketing, advertising or research, that they have not taken part in a focus group in the last six months. Usually 9-10 people are recruited for each group of 6-8, to cover absences. Participants receive a reward for their time, and the reward must not be the main reason for taking part, otherwise the room fills with "professional participants" who say what they think the moderator wants to hear.

Recruiting among customers, employees or acquaintances of the project team is the most frequent error. Loyal customers say nice things, employees say what is safe, and acquaintances say what pleases you. None of them says what the market says.

5. The discussion guide and the moderator

The discussion guide usually contains 10-12 open questions, ordered from general to specific and from easy to hard. The classic structure has five categories of questions: opening questions, through which each participant says something simple about themselves and gets used to speaking; introductory questions, which bring the topic into the discussion; transition questions, which lead towards the heart of the subject; the key questions, two to five of them, which receive the most time; and the ending questions, which ask participants to summarise, to say what was most important and what was missing from the discussion.

Good questions are short, carry a single idea, are phrased in the participants' language, and ask for descriptions of experiences rather than abstract opinions: "tell me about the last time you…" produces more than "what do you think about…". The question "why?" asked directly produces rationalisations; experienced moderators replace it with "what made you…" or "what would have to change for you to…".

The moderator leads without influencing: asks, listens, requests examples, brings the discussion back when it drifts, gives the floor to the quiet and tempers the dominant, without ever approving or disapproving of an answer. It is a skill learned over years, and the difference between a professional moderator and a project colleague who is "good with people" shows in the transcript: in the first case the participants speak, in the second the moderator does.

6. The session: room, recording, observers

The discussion takes place in a neutral room, with participants seated in a circle or around a table, and is recorded in audio and, preferably, video, with the participants' consent. Observers from the client side follow from another room or via a live feed, not from inside the room, and do not intervene. The materials to be tested, concepts, packaging, advertisements, are shown after participants have talked about the product category in general, and their order is rotated between groups so that the first material shown is not always favoured. In the Republic of Moldova, groups are run in the language participants speak at home, Romanian or Russian, with native moderators, not through translation.

7. Analysis: patterns, not impressions

The most frequent error after a focus group is for the client team to leave the observation room with ready-made conclusions based on the most memorable remarks. A memorable remark is not a result. Proper analysis starts from the full transcripts of all groups and looks for patterns: which themes appear in all groups, in which categories, how frequently, with what emotional intensity and in what degree of detail. An opinion expressed by one person, however articulate, carries different weight from one that comes up spontaneously in three groups out of four.

The report of a focus group study contains: a synthesis of themes by participant category, the differences between categories, illustrative quotes chosen to represent the pattern rather than the exception, the language participants used for the concepts studied, the reactions to each material tested, and the resulting hypotheses. Statements such as "60% of participants preferred" have no place in it; "in all the groups of young people, option B was perceived as more modern but less credible" is the correct form.

8. What happens next: from hypotheses to decision

The focus group produces understanding and hypotheses, not measurements. When the decision is important and costly, the hypotheses are verified through a survey: the concept preferred in the groups is tested quantitatively on a sample, the messages identified are measured for awareness and relevance, the barriers discovered become questionnaire items. The sequence qualitative, then quantitative, is the safest road from "we don't know why" to "we know what to do and how sure we are". The guide Which type of research solves your problem describes this sequence.

There are also decisions for which the focus group is enough: adjusting a text, choosing between two wordings of the same message, identifying the obvious problems of a prototype. When the reaction is unanimous and strong in all groups, a survey would confirm, not add.

9. What to ask the agency for a focus group study

• The number of groups per participant category and its justification.
• The screening questionnaire and the exclusion criteria.
• The recruitment source and the reward offered.
• The discussion guide, for approval before the first group.
• Who moderates, with what experience, in which language.
• Audio and video recording, full transcription.
• The analysis method and the structure of the report.
• The possibility of observing the groups without intervening.

Done properly, the focus group is the fastest way to hear the market speak in its own words. Done badly, with single groups, friendly participants and conclusions drawn from behind the mirror, it is the most expensive way to confirm what you already believed. The difference lies in recruitment, the guide, the moderator and the analysis, and all four can be checked before signing. The other qualitative and quantitative methods are described on the methodology FAQ page.

See also: Focus groups and IDI Qualitative research Monitoring and testing In-depth interviews Which type of research solves your problem Methodology FAQ

The Perspicuus research team · 17 August 2026