A focus group puts your personas in conversation with each other. Instead of each persona responding in isolation — as in a concept test or survey — they hear what others think and react to it. This unlocks the dynamics that matter most in real consumer decision-making: which opinions dominate, where consensus forms quickly, and where genuine disagreement lives.
Use a focus group when you want to:
- Understand how a concept holds up under social scrutiny, not just individual reflection
- Surface the themes that are contested versus those that everyone agrees on
- See which personas emerge as opinion leaders and which follow
- Go deeper on findings from an earlier concept test
Every Boses focus group runs across two structured rounds, moderated by an AI facilitator.
Moderator opening — The AI moderator introduces the topic and poses the first question to the group.
Round 1 — Each persona responds independently to the opening question. These are their initial, uninfluenced positions.
Moderator bridge — The AI moderator synthesises the Round 1 responses and poses a follow-up question that surfaces the most interesting tensions in what the group said.
Round 2 — Each persona reads the other participants’ Round 1 statements and responds again — agreeing, pushing back, qualifying, or shifting their position. This is where social dynamics emerge.
Aggregate report — After both rounds complete, Boses generates a full moderator report from the entire discussion.
Running a focus group
Create the simulation
Create the simulation with simulation_type: "focus_group" and your opening prompt_question. The simulation starts immediately.curl -X POST https://api.temujintechnologies.com/api/v1/projects/<PROJECT_ID>/simulations \
-H "Authorization: Bearer <TOKEN>" \
-H "Content-Type: application/json" \
-d '{
"persona_group_id": "<GROUP_ID>",
"simulation_type": "focus_group",
"prompt_question": "We are considering launching a subscription meal-kit service in Metro Manila priced at PHP 1,200 per week for two people. The kits include pre-portioned ingredients and recipe cards for 3 meals. What are your initial thoughts, and would this fit into your current routine?"
}'
id as SIM_ID. Poll for completion
Focus groups take longer than concept tests because they run two rounds plus the moderator transitions. Poll until status is "complete".curl https://api.temujintechnologies.com/api/v1/projects/<PROJECT_ID>/simulations/<SIM_ID> \
-H "Authorization: Bearer <TOKEN>"
progress field reflects which round is currently running. Status moves through running → generating_report → complete. Fetch the results
curl https://api.temujintechnologies.com/api/v1/projects/<PROJECT_ID>/simulations/<SIM_ID>/results \
-H "Authorization: Bearer <TOKEN>"
Reading the aggregate report
The focus group aggregate report is richer than a concept test’s because it captures the group dimension of the discussion.
Moderator summary — A narrative account of how the conversation evolved: what position the group started from, how it shifted between rounds, and what the dominant mood was by the end.
Consensus themes — The ideas that emerged across personas regardless of their starting sentiment. If seven out of eight personas mentioned “time savings” — whether positively or as a concern — it appears here as a consensus theme.
Disagreements — The topics where the group divided. These are often the most strategically useful findings: a split on price, for example, might correspond to an income segment boundary worth exploring further.
Sentiment distribution — The proportion of positive, neutral, and negative positions across all persona responses in both rounds.
Recommendations — Concrete next steps derived from the group’s collective output.
Example aggregate (abbreviated):
{
"aggregate": {
"moderator_summary": "The group was initially split on price — younger participants saw 1,200 PHP as aspirational but achievable, while participants with children immediately framed it as a household budget decision. Round 2 shifted the conversation toward convenience: after hearing others emphasise time savings, even the most price-sensitive participants reconsidered. By the end, 6 of 8 personas expressed conditional willingness to try it.",
"consensus_themes": [
"Time savings on weeknight meal preparation",
"Interest in recipe variety and learning to cook",
"Uncertainty about freshness and delivery reliability"
],
"disagreements": [
{
"topic": "Price acceptability",
"positions": "Younger single professionals accepted the price as a treat; parents of young children considered it a luxury that competed with groceries"
}
],
"sentiment_distribution": {
"positive": 55,
"neutral": 28,
"negative": 17
},
"recommendations": "Lead with convenience messaging for the 25–35 professional segment. For family-oriented buyers, reframe the value proposition around cost-per-meal versus restaurant delivery. Consider a starter 2-meal trial kit at 650 PHP to reduce the first-purchase barrier."
}
}
Tips
Run a focus group after a concept test on the same topic. Use your concept test themes as the foundation for your prompt_question, and let the focus group go deeper on the tensions your concept test surfaced. The two methods complement each other directly.
Your prompt_question sets the opening frame for the entire discussion, so it shapes both rounds. Write it as you would a real moderator’s opening question — specific enough to focus the conversation, open enough to let disagreement emerge.
If only one persona completes Round 1 successfully, Round 2 is skipped automatically and the aggregate report is generated from Round 1 data only. Ensure your persona group has been generated successfully before starting a focus group.
