Written by: Amanda Herriman, Director of Strategy, WAX Healthcare Marketing
Why do 80% of healthcare consumers say one thing in surveys and do another when choosing care? Traditional audience research misses what matters most: actual behavior.
Focus groups capture what people say. Surveys record stated preferences. Both methods cost thousands of dollars and weeks of time. Neither reveals what people actually do when making healthcare decisions. Synthetic audiences use AI to simulate real decision-making behaviors, uncovering insights traditional research cannot capture. Healthcare organizations are already using this technology to identify hidden barriers in the patient journey.
Understanding Synthetic Audiences
Synthetic audiences are AI-generated models that simulate how real people think, decide, and act within specific healthcare contexts. Unlike traditional personas (static documents describing demographics and preferences), synthetic audiences are dynamic, interactive representations that respond to scenarios and predict behaviors.
A traditional persona might describe "Maria, 62, Medicare-eligible, lives alone, concerned about prescription costs." A synthetic audience model of Maria can interact with different plan options, react to messaging approaches, and demonstrate how her concerns balance against other priorities like provider networks or coverage details. This simulation capability comes from training AI models on vast datasets of actual consumer behavior and decision patterns.
Why Traditional Research Falls Short
Focus groups and surveys provide value but face inherent limitations. They suffer from social desirability bias, capture stated rather than revealed preferences, and represent small samples in artificial environments. Most critically, they ask hypothetical questions rather than observe actual decision-making behaviors. What people say they'll do and what they actually do often diverge significantly, especially in healthcare where decisions carry emotional weight and social stigma.
Synthetic audiences extend traditional research rather than replacing it. By simulating hundreds of decision scenarios, they reveal patterns impossible to test with real participants and uncover "silent filters" in decision-making that people may not consciously recognize or articulate.
Case Study: Uncovering Hidden Barriers
A North Carolina behavioral health organization partnered with healthcare marketing agency WAX to conduct a brand assessment. Beyond standard qualitative research, the agency introduced synthetic audiences using AI models built from five detailed audience profiles and integrated with claims data specific to NC behavioral health patients.
Each synthetic persona was prompted to navigate the patient journey online: search for care, identify three potential providers, and explain which they'd choose and why. The Medicaid persona revealed a critical insight: this persona couldn't easily find information about Medicaid coverage acceptance while navigating the client's website. The information existed but wasn't prominent. The synthetic model selected a competing provider that displayed Medicaid acceptance clearly on its homepage.
This discovery revealed a fundamental barrier beyond website navigation. For Medicaid recipients already facing access challenges, difficulty confirming coverage could eliminate a provider from consideration entirely – resulting in lost patients and revenue for the organization.
The Power of Behavioral Simulation
The behavioral health case demonstrates synthetic audiences' core value: revealing what people don't say or don't realize themselves. In focus groups, Medicaid recipients might not mention difficulty finding coverage information because they don't consciously recognize it as a decision factor. But synthetic modeling revealed it acted as a silent filter, eliminating options before other factors came into play.
Getting Started Responsibly
Healthcare leaders considering synthetic audiences should start with specific, high-value questions rather than broad exploration. Combine synthetic modeling with traditional research because each method strengthens the other. Maintain vigilance against bias: AI models trained on historical data can perpetuate existing biases, requiring careful attention to diverse data sources and regular validation with actual community members.
Done right, synthetic audiences bring us closer to understanding patient needs in all their complicated, human complexity.
**Key Takeaway:** Synthetic audiences amplify—not replace—human insights, turning complex data into empathy-driven understanding.