Documentation

How a focus group can be launched

There are two fundamentally different ways to run a session. Which one you use depends on whether you are exploring an idea on your own, or you want the platform to target weaknesses discovered in a GEO analysis and feed results back into that GEO report.

1. Manual (single) run from your dashboard

You configure the assignment yourself: the scenario, how many discussion rounds to run, how many personas participate, and persona attributes such as age range and gender. The system follows your brief.

This mode is ideal when you want to sharpen a product or narrative, test how an idea lands in conversation, and iterate quickly—both for “market” style reactions and for how AI-styled personas push back on claims.

2. Automatic run linked to a GEO analysis

When a focus group is triggered from GEO, the system uses your GEO results to infer where the company is comparatively weak. The moderator brief and other session parameters are generated from those weaknesses—not from a blank form you fill manually.

Participants do not know which brand is the subject of the study. They only see a set of competitors; your brand is mixed into that set without being singled out by name as “the client.” Personas are tasked both to evaluate the offering associated with each labeled brand and to vote—which brand they would choose in specific situations.

3. Voting data back in the GEO report

For GEO-linked runs, aggregated voting and related outputs from the focus group are written back into the GEO report. That lets you connect “what generative answers look like” with “which brand simulated buyers pick when they compare alternatives”—in one analytical thread.

What the Overview tab in results shows

Once a session completes, Overview summarizes the synthetic run: how many personas, how many rounds, how many objections and conflicts were detected, and headline scores (Confidence, Idea) when present. It is the executive briefing for “did the simulation hang together and what is the headline verdict?”

How it is built

The focus-group engine runs a moderated multi-persona dialogue against your brief (manual) or against a GEO-derived brief (linked). Counts are tallied from structured outputs and parsed transcript segments. Confidence and Idea scores compress many turns into headline percentages for tracking over time.

How to read the cards

• High objections with low conflicts can mean “lots of skepticism but people stayed engaged”—good for sales enablement, less comfortable for brand.
• Low rounds or thin personas may cap reliability—check the Participants section for depth.

How to interpret for action

Use Overview to decide whether to trust the rest of the tabs: if Confidence is low, read Discussion and Transcript cautiously and consider re-running with a clearer brief or stronger product context.