Run a synthetic focus group in two minutes
A synthetic focus group is the classic qualitative format with the recruitment removed: six participants of a defined profile, a moderator, a stimulus, and a conversation where they read and react to each other. The difference is that the participants are AI archetypes grounded in a specific market, so the session starts the moment you type the question instead of three weeks later.
How a session runs
The whole flow fits in a paragraph, and that is half the argument.
- Pick the market — Spain, Mexico or Chile — and up to six audiences from the base panel, or build your own by age, city, income, education, occupation and household.
- Optionally paste your brand's URL: the engine reads it once and treats it as what the brand says about itself, not as fact.
- Type the question. The six answer in streaming, one at a time, reading what the others said.
- Follow up on any answer. The room remembers the whole session, so round five is still the same conversation as round one.
- Export the full transcript and an executive report with drivers, barriers and prioritised recommendations, in PDF, PowerPoint or CSV.
Where it fits against a recruited group
A recruited group in Madrid, Mexico City or Santiago costs thousands per session and takes two to four weeks, almost all of it recruitment. It gives you something a simulation never will: real people, sensory reaction, and a result you can defend to a committee.
The synthetic session gives you the thing the recruited group cannot — twenty iterations for the price of none. Use it to arrive at the real group with three hypotheses instead of ten, and to keep iterating afterwards without recruiting again.
The moderation still matters
Removing the recruitment does not remove the craft. The questions that ruin a real focus group ruin a synthetic one identically: asking whether someone would buy, asking what they would pay, asking whether they like 'our' product, and asking the group whether they agree.
Start without the brand, get behaviour before opinion, show the stimulus and then stay quiet, and only ask directed questions once the spontaneous reaction is on the record.
Honest limits
No sensory reaction, no representativeness, no forecasting, and no standing as formal evidence. Unanimous agreement across the panel is usually a sign the question was not sharp enough, not a finding.
Where the profiles disagree is where to look, and where to point the human research that follows.
Frequently asked questions
Six is the default and matches the classic qualitative group. Fewer and one voice dominates the transcript; more and the session becomes long without becoming richer. Studio Pro allows up to twelve and two parallel rooms for A/B comparison between segments.
Keep reading
Last reviewed: 2026-08-21. Participants are AI-generated synthetic archetypes, not real people.