Data Gurus Podcast

The episode contains several genuinely useful research findings - the negligible benefit of deep profiling over demographics, the 0.2 correlation benchmark, and the systematic pro-technology bias - but these are spread thin across 29 minutes of conversational padding and soft transitions.

“we find that adding the 500 questions that you answer doesn't really help versus just your demographics”

“we found a correlation of only 0.2 between the human answers and the twin answers, which is about the same as the correlation between height and iq”

The episode draws directly from primary research rather than recycled frameworks, and the hyper-rationality finding and within-person A/B test insight are genuinely non-obvious; the neuromarketing hype-cycle comparison is a familiar trope but applied with relevant specificity.

“Digital Twins tend to be a bit more hyper rational than humans because they're basically, it's an LLM that has information about you, but the ALM knows almost everything and it has unlimited reasoning capabilities”

“I can assign A and B to the same Persona because there would be different API calls, there would be no memory, no contamination. Like I can do true a B test within a person, which is fantastic”

Toubia is a genuine practitioner-researcher with 25 years of relevant work, a multi-institution study of real scale, and publicly released data - not a thought-leader but someone who has actually built and tested the thing being discussed.

“I got into marketing from operations research and in, uh, 1999 I started working on how do we optimize online surveys”

“we interviewed over 2,000 people, uh, on the Prolific...We asked them, basically we put together all the possible scales we could find from psychology, business, economics, uh, IQ questions, scales, Big five, lots of different economic games”

The episode is anchored in concrete numbers - sample sizes, question counts, correlation coefficients, co-author counts, and named platforms - which is unusually grounded for a market research podcast, though some claims about domain-level performance differences remain tentative and unquantified.

“Over 2,000 people, each answering over 500 questions over four different waves”

“we found a correlation of only 0.2 between the human answers and the twin answers, which is about the same as the correlation between height and iq”

The host has genuine domain familiarity and lands one sharp clarifying challenge, but most questions are validating or open-ended softballs, and the guest's tentative claims about domain-specific performance are never pressed for harder evidence or practical thresholds.

“It does help or it does not help?”

“I'm curious from your perspective, I know you're still researching everything, but do you feel like a lot of the excitement around it is really based on m. The commercial aspect?”