The Rise of Synthetic Audiences: Fast, Cheap… but Believable?

In this episode of The Intuitive Customer, Professor Ryan Hamilton is joined by Ben Shaw, seasoned brand strategist, to unpack the promises and pitfalls of synthetic audiences - AI-driven research used for faster, cheaper market research.

Synthetic audiences, powered by large language models (LLMs), can replicate customer segments and respond to creative concepts or product questions at scale cutting the time and cost of traditional surveys. But does it come with trade-offs?

Expect lively debate on the AI vs. LLM naming debate, the enduring value of ethnography and nuance, and practical tips for blending synthetic and traditional research to make smarter, more human-centred decisions. From democratising access to consumer insight to questioning the believability of robot-approved ad copy, this candid discussion highlights how to use these tools wisely.

🔑 Key Takeaways

1. Synthetic ≠ Real: Synthetic audiences use LLMs to predict human responses—they're powerful, but they're still not replacements for real people.

2. Direction not Proof: Treat synthetic audiences as a first-cut hypothesis tool i.e.great for inspiration, not yet reliable as stand-alone evidence.

3. Nuance & Minority Voices Matter: AI often defaults to the average response, risking loss of critical insights from edge cases or minority behaviours.

4. Democratisation of Insight: With the right data, teams from product to shop-floor staff can ask customer-centred questions directly.

5. Best Practice: Use synthetic audiences for speed and hypothesis testing, then validate with real humans—especially for high-stakes decisions.

📚 Resources & Mentions

• HubSpot CMO Kipp Bodnar "Better, faster, AND cheaper isn't a trade-off anymore"

• Harvard Business School study 'Using LLMs for Market Research'