A practical guide to understanding, using, and evaluating AI-simulated consumer audiences.
01 Defining synthetic audiences
A synthetic audience is a representation of a real-world consumer group, in the form of AI agents, designed to simulate how people within that group might behave in different situations.
These AI agents are given information about the consumers they represent, including their demographics, preferences, attitudes, purchasing behaviors, and other relevant characteristics.
They can then be placed in different scenarios, such as encountering a new product, seeing an advertisement, or considering a purchase, to simulate how those consumers might think, react, and make decisions.
This allows brands to explore consumer behavior, test ideas, and investigate research questions without recruiting human respondents for every study.
The people you want to understand, described by who they are and how they behave.
Each agent is given what is known about the consumers it represents.
The agents simulate how those consumers might think, react, and decide.
Real consumers AI agents
02 How to use them
Put AI-simulated consumers in real-world situations to explore how they might think, react, and behave.
Create a population of AI agents representing the consumers you want to understand, with their own characteristics, preferences, and behaviors.
Expose your audience to a product, advertisement, shopping experience, or real-world scenario, and simulate how they might respond.
Examine their simulated reactions, decisions, and behaviors to uncover patterns, compare outcomes, and identify insights worth validating.
03 Use cases
From the ads people see to the products they choose, explore how different consumers might respond in real-world situations.
A brand testing three advertising concepts before launching a campaign.
Consumers choosing between competing products at different price points.
A brand exploring how its positioning resonates with consumers in a new country.
A shopper navigating a digital experience, encountering friction, and deciding whether to purchase.
04 Trusting the results
Not all synthetic simulations are equally reliable. Their accuracy can vary depending on the type of study, the questions being asked, the audience being simulated, and the context in which decisions are made.
At Neon Blue, we compare synthetic results with real human research to understand where simulations are reliable, where they fall short, and how they can best be used.
Where do synthetic and human responses align?
Where do they diverge?
Which questions and scenarios are suitable for synthetic research, and which still require human validation?
“Which of these two messages would make you more likely to buy?”
Actual responses from the target consumer group.
AI-simulated responses for the same group.
Examine agreement, differences, and limitations across questions.
Human Synthetic
We're inviting selected brands to run a free study with human respondents, alongside an AI-simulated audience, to compare how the two respond to the same research questions.
Explore a free study