How can you find the right audience for a product before you spend on fieldwork?
A worked example from a real simulation run on an invented product, with the full sample report to explore.
Audience discovery puts one proposition in front of simulated personas from several segments and asks each the same questions. The report shows which segments find it relevant, which find it unclear and what would block adoption. Treat the result as a shortlist of segments to test with real people, not as a measure of demand.
Before a team commits to interviews or a survey, it needs a short list of who to ask. Audience discovery builds that list. It puts one proposition in front of simulated personas from several segments, asks every persona the same questions and reports who finds the proposition relevant, who finds it unclear and what would stop each group from adopting it.
When is audience discovery the right method?
Use it early, when the buyer is still uncertain and the positioning is broad. It suits a new product, an existing product entering a new market, or a proposition that several kinds of team might want. If you already know the segment and want to improve the wording, a concept test is the better fit. If you need to know how common a view is among real people, you need a survey.
What does a sample audience discovery report look like?
The example below is a real run on an invented product. “Brightloom” is described as the operating system for research, insights and strategy firms. Four simulated personas, drawn from research teams of different sizes and roles, each answered eight questions twice. Every score and quotation comes from that run; the company is fictional.
Brightloom: which research teams are most likely to adopt it?
Identify which segments of research, insights and strategy teams are most likely to adopt the product, how well the proposition fits each segment, and what would drive or block adoption.
Revise the concept and retest the weakest dimensions before committing to rollout.
- Personas4
- Runs each2
- Simulations8
- Study typeAudience discovery
Concept clarity by segment
- Director · SMB (50–500)27
- Lead / Principal · NGO32.5
- Manager · SMB (50–500)33
- Manager · Solo / freelance36.5
The proposition is too broad to judge. An 'operating system' could add another layer of complexity without replacing existing tools, and I would be concerned about cost, migration effort, source confidentiality and whether it genuinely supports qualitative humanities research.
It would be most useful when a small research firm is running several concurrent projects and needs to connect source material, interview findings, analysis, decisions and final strategic recommendations in one traceable workspace.
A simulated example for a fictional company, run on synthetic personas from a stated catalogue. It estimates how a defined audience might respond to the material shown; it does not measure real people, and company names are invented.
The headline is a mixed signal of 42.5 out of 100, and the reason is visible in the bars. Preference strength is the highest at 54.8, but concept clarity is only 32.2. The personas said an “operating system” was too broad to judge and could not tell which existing tools it would replace. All eight runs ended at “unsure” on adoption, and in six of the eight the persona described itself as a recommender rather than the final decision maker. That points to a proposition to sharpen, not an audience to drop.
How do you read the report?
Every persona-research report follows the same order. These six steps walk through the example above.
Start with the decision brief
The ring is the overall signal: 42.5 out of 100, labelled “Mixed signal”. It combines the dimension scores with equal weight, so read it as a direction, not a verdict. Beside it sits the line the evidence supports acting on.
In the report: Decision briefFind the dimension pulling the score
Preference strength is the strongest dimension at 54.8; Concept clarity is the weakest at 32.2. An average can hide one weak dimension, and that is usually where the next round of work belongs.
In the report: Performance across dimensionsLook for gaps between segments
Concept clarity by segment runs from 27.0 to 36.5 across the panel. A gap between segments shows who a finding applies to; a flat line means the panel largely agreed.
In the report: Overall and by personaRead findings with their prevalence
Each finding states how many of the 8 simulations it appeared in, next to a quoted simulated answer. A point raised in 8 of 8 runs carries more weight than one raised once.
In the report: Reported findings, prevalence and simulated voiceCheck who was simulated and why
The panel is 4 personas drawn from a catalogue with stated filters, each answering 2 times. The report lists every profile and why it was selected, so you can judge whether the right people were in the room.
In the report: Who was simulated—and whyFinish with the limits and the next step
Every report ends with what the evidence cannot show and the real-world checks to run next. The simulated answers are hypotheses to test with real people, not findings about them.
In the report: Limitations and appropriate use · Where to go next
Can you explore the whole report yourself?
Yes. The frame below holds the complete sample report, with its tabs for findings, panel, next steps, method and the run-by-run ledger. It is the format a real study produces, with fictional names.
What would you do after a result like this?
The report ends with concrete next steps. For this example they were to:
- show concrete workflows for qualitative, archival and interview-based research
- clarify which existing tools the product replaces, complements or integrates with
- demonstrate source provenance, traceability, search and export
- provide evidence that the product saves time without reducing analysis to generic summaries
The sensible follow-up is to rewrite the proposition, run the same study again to see whether clarity moves, and then test the segments that responded best in real conversations.
What can audience discovery not tell you?
It cannot size a market or measure demand. Simulated personas answer from the profile they are given, and the research is clear about the limits. Nielsen Norman Group tested AI-generated “users” against real studies it had already run and found the answers over-optimistic and shallow. It advises treating the output as hypotheses to check with real people (Nielsen Norman Group).
How much grounding a simulated person has matters. In a study of 1,052 people, agents built from a two-hour interview with each person reproduced held-out survey answers at about 83% of how consistent people were with their own answers two weeks later, while agents given only demographics reached 74% (Park et al., latest version). That suggests the less a persona is grounded in a real person’s own words, the lower the ceiling is likely to be. Use audience discovery to choose where to look, not to decide.
A short sequence
- Write the proposition in two sentences, as you would show it to a buyer.
- List the segments you suspect might care and the ones you doubt.
- Fix the questions before you see any results.
- Read clarity and barriers first, then fit.
- Rewrite, run again, then test the strongest segments with real people.
Frequently asked questions
How many personas does audience discovery need?
Enough to cover each segment you care about more than once. The example on this page used four personas that answered eight questions twice each. More runs show whether a finding is stable; more personas show how far it spreads across segments.
Is audience discovery the same as a survey?
No. A survey samples real people and can estimate how common a view is. Audience discovery asks simulated personas and returns directional signals about fit, clarity and barriers, which is useful for deciding whom to survey.
What should I do with a mixed signal?
Read it as a proposition that is not ready, not as a verdict on the market. Fix the weakest dimension (in the example, clarity), run the same study again to see whether it moves, then take the segments that responded best to real conversations.
Sources
Sources for the facts on this page, last checked October 17, 2026.
Run research end to end. Keep the knowledge working.
One AI-native operating system for market research and insight professionals. Study design, evidence generation, agents, institutional knowledge, delivery and action, all in one place.