Know what you’re learning from.
Audience Analysis creates fictional AI personas from a market description. Their responses can suggest hypotheses, objections, and useful follow-up questions. They are not interviews with real people, a representative survey, or evidence of market demand.
How a panel is created
Your target-market description, panel size, and research purpose are used to generate a set of persona outlines. AI models expand those outlines into profiles with backgrounds, goals, problems, and decision habits. The generation pipeline uses multiple models to vary perspectives. Variation across models does not establish accuracy or statistical independence.
How questions are answered
A question is presented to the selected panel or an individual persona. When you add product context, that information is included to help the models interpret the question. The system collects generated responses and synthesizes themes, objections, sentiment, and—when relevant—pricing suggestions. A summary can omit nuance, so read the individual responses too.
How to interpret the numbers
- Simulated interest: a model-generated score summarizing the panel’s apparent interest. It is not a probability of purchase.
- Response sentiment: the distribution of generated response labels. It describes this synthetic panel, not the target market.
- Suggested price range and median: a synthesis of hypothetical willingness to pay. Validate pricing with real buying behavior.
What can go wrong
Models can repeat stereotypes, agree with a leading question, invent details, or produce confident but inconsistent explanations. Detailed personas can feel realistic even when their assumptions are wrong. Larger panels can explore more generated perspectives, but do not create a statistically representative customer sample.
A responsible research workflow
- Define a specific market and a decision you need to make.
- Review generated profiles for fit and contradictions.
- Ask neutral questions and inspect dissenting responses.
- Record hypotheses and what could disprove them.
- Validate important conclusions through real customer interviews, prototype tests, or observed behavior.
Public and private research
Public panels and their public profiles and questions can be read by anyone and appear in search engines. Advanced and Ultimate plans provide the option to create private panels. Choose visibility before generating a panel and avoid placing confidential information in public research.
About this service
Audience Analysis is operated by MB Skydis. These guides explain the product’s workflow; they do not claim that synthetic research is independently validated. Send feedback or contact support@audienceanalysis.ai to report a misleading result.