Synthetic focus groups are fast and cheap enough to screen many concepts and pre-test a discussion guide, but their participants are simulations and their output is directional. In-person focus groups are slower and costlier, and give real reactions, body language and physical product handling. Most teams use synthetic groups to narrow the options and in-person groups to confirm what matters.
Synthetic focus groups and in-person focus groups answer different needs. A synthetic focus group asks AI-simulated participants, conditioned on a profile of your audience, to react to concepts, messages or questions; it is fast, cheap per round and good for narrowing options. An in-person focus group brings six to ten recruited people together with a moderator; it is slower and more expensive, and it gives you real reactions, body language and hands-on product use. Synthetic output is directional, so the usual pattern is to screen with synthetic groups and confirm the shortlist with real people.
A synthetic focus group starts with an audience definition, such as "first-time parents in Germany who buy organic baby food". Simulated participants are generated within that definition and, in better tools, grounded in evidence: published statistics, research about the segment, and your own survey results or interview transcripts. You then show them the stimulus (a concept, claim, packaging image, ad or landing page) and ask the discussion questions. Some tools also run structured tasks, such as rankings or MaxDiff, so that options are forced to compete.
The output is a set of answers and themes you can compare across segments. Because every participant comes from the same underlying model, the answers can be wrong in the same direction at once, and they often show less disagreement than a real group would. For the method in more depth, see AI focus groups.
An in-person group is recruited through a screener, usually six to ten participants per session, and led by a moderator with a semi-structured discussion guide over one to two hours. Several sessions are typical. The sessions are recorded, transcribed and analysed.
The value lies in real people reacting in their own words, handling the product, and building on or disagreeing with each other. The weaknesses are equally well known: small samples, dominant voices, social desirability and moderator effects, and the time and cost of recruiting the right people.
In-person research runs on a linear timeline: brief, screener, recruitment, scheduling, sessions, transcription and analysis. Synthetic groups compress the early rounds into a working day, which lets a team test, revise and test again before committing to fieldwork. Speed is only useful if the output is good enough for the decision, which is why the synthetic round should end in a shortlist, not a final answer.
Evidence quality
Published studies of language-model respondents show both promise and clear limits. Models conditioned on real survey backstories reproduced many subgroup response patterns (Argyle et al., Political Analysis, 2023), while other work found averages that matched but too little variation, sensitivity to prompt wording and drift over time (Bisbee et al., Political Analysis, 2024). In-person groups have their own biases, but they are biases of real people, and a small group is never a statistical sample either.
Cost
In-person costs come from recruitment fees, incentives, venues or platforms, moderators and analysis time, so teams run few groups. Synthetic rounds cost little per question, so the constraint shifts from budget to the time it takes to read and judge the results.
Privacy
In-person groups collect personal data, recordings and consent forms. Synthetic groups involve no real participants, but the files, customer data and account data you use to build and run them can still be personal data. Review the provider's privacy policy, data processing agreement and subprocessors with your privacy team; no tool is compliant for every use by default.
In Minds, a synthetic focus group is a Study asked of an Audience of Minds. Minds are grounded in sources and in files you add, and a Study can ask open and closed questions, run MaxDiff and other methods, and show stimuli such as concepts, images, websites and copy. Audience Validation checks an Audience against real published surveys, or survey files you upload, and shows a score out of 100 with a 95% range for each survey, so you can judge how far to trust that Audience before you rely on its answers. Minds' published survey replications are in We Tested Synthetic Audiences Against Reality.
Minds does not replace in-person research for physical products, authentic quotes or decisions that need evidence from real people.
Use synthetic focus groups to explore and narrow: many options, a draft guide, an audience that is hard to reach. Use in-person focus groups to confirm and to hear real people: the shortlist, the physical product, the decision that has to stand up to scrutiny. Run them in that order and the in-person groups become smaller, better aimed and more useful.
How do synthetic focus groups compare to in-person focus groups in accuracy?
They measure different things. An in-person group records what real participants say and do, with the known biases of small groups and moderation. A synthetic group produces estimates of how an audience might react; published research finds these can match some survey patterns and miss others, with too little variation among simulated participants. Treat synthetic output as directional and check it against real data before a costly decision.
What is the cost and speed difference between the two methods?
In-person groups need recruitment, incentives, a venue or video platform, a moderator and analysis, and typically take weeks from brief to report. A synthetic group runs in minutes to hours and costs little per question, which is why it suits early rounds with many options.
When should a research team choose synthetic focus groups over in-person sessions?
Choose synthetic groups to screen many concepts or claims, pilot a discussion guide, or get a first read on a hard-to-recruit audience. Choose in-person groups when you need real reactions to a physical product, authentic quotes, observed group dynamics, or evidence that a stakeholder or regulator will scrutinise.
How can I check whether a synthetic focus group is reliable?
Compare its answers with real data for the same audience. In Minds, Audience Validation asks an Audience the questions of real published surveys, or of survey files you upload, and scores how close the answers are, with a 95% range and a list of the questions left out and why.
