What research teams get right that AI personas can't
A real research team brings things to a study that no synthetic panel can replicate — not because the AI is bad, but because these are structurally human skills.
This page argues against our own product, on purpose.
A tool built to sell synthetic research has an obvious incentive to downplay what real research teams do well. We're doing the opposite here, because understating a real research team's value doesn't make thepanelist more useful — it just makes the eventual disappointment worse when someone learns the hard way.
Real-time follow-up judgment
A skilled researcher in a live interview notices when an answer doesn't quite make sense and asks the exact right follow-up question, in the moment, based on tone, hesitation, or a contradiction they caught live. A synthetic panel's follow-up is scripted by whoever's asking it, after the fact, with no live read on the persona's actual reaction.
Recruiting genuinely representative, hard-to-reach participants
Reaching a specific, narrow, or hard-to-access real population — a rare medical condition, a niche professional role, a specific regulatory context — is a real skill and infrastructure problem that a research team solves through networks and recruiting expertise a language model has no access to.
Institutional and domain context
A research team embedded in an industry carries context about what's already been tried, what past studies found, and what a surprising result would actually mean — context that shapes which questions are worth asking at all.
Genuine ethical judgment
Deciding whether a study design is fair to participants, whether a question is manipulative, or whether a finding should be reported cautiously requires human ethical judgment a synthetic tool doesn't carry.
None of these are things a synthetic panel does badly. They're things it doesn't do at all, by the nature of what it is.
Common questions about this comparison
Is thepanelist trying to make research teams obsolete?
No — this page exists specifically to name what real research teams do that a synthetic tool structurally can't, because that gap is real and worth stating plainly.
When should I use a research team instead of a panel?
When you need live follow-up judgment, access to a hard-to-reach real population, deep institutional context, or ethical judgment about a study's design — none of which a synthetic panel provides.
Can the two be used together?
Yes, and it's a common pattern: use a synthetic panel to sharpen which questions are worth a research team's limited time, rather than treating the two as competing options.
Use a panel to sharpen your research team's next study
A directional filter before the real, human-led research begins.
Start a panelMore Critical Perspectives
Synthetic panels are a gut-check, not a replacement
It would be easy to let the marketing copy drift toward 'skip user research entirely.' Here's exactly what a synthetic panel can and can't tell you, and why we keep saying so even when it costs us a cleaner pitch.
AI market research: how accurate is it really?
Accuracy isn't one number. It's a set of separate, checkable claims — and most of what gets called 'accuracy' in AI research marketing is actually a claim about something else entirely.
Why we said no to fake respondent counts
It would be easy to show '10,000 simulated respondents' on a panel result. We don't, because a panel of five distinct personas and a panel of ten thousand near-identical copies would look the same on that number.