Can you trust a simulated customer? Findings from 20,000 replications

Can you trust a synthetic persona that answers in seconds? In this two-person chat, Viewpoints CEO Leo Yeykelis and Stanford professor Byron Reeves will run a few short synthetic persona studies live and ask the audience to vote on if they would trust this result for a real decision. We will debate the verdict from two angles, speed to insight versus scientific rigor, and share the checks that turn a convincing simulation into defendable research results.

Key takeaways:

- How to set up and run synthetic studies you can trust, including best practices for sampling, context, measures and the red flags that tell you when not to use the output.
- Where synthetic research fits right now – a practical map of the near-term use cases that perform well and the situations where human research is still the safer choice.
- How to conduct “superhuman” research with synthetic respondents – iterating on studies in minutes to see what makes findings no longer generalized, run head-to-head comparisons across 10+ assets per respondent to produce a true stack rank and model hard-to-reach audiences on demand.