AI arrived with a promise for quant: synthetic respondents enabling instant fieldwork, at a fraction of the cost. The survey stays exactly the same, the humans are replaced with AI-generated ones. Yet this misses the problem entirely. The problem with quant was never the humans. It was the design philosophy.

Survey research was built to measure customers: capture a snapshot, aggregate it, report it. The methods themselves were shaped around human limitations. Survey length capped by respondent fatigue. Task complexity limited by attention. Sample size constrained by cost and time. Conjoint, regression, segmentation, scenario simulation: good techniques, rigorous and proven. But the infrastructure they ran on was never equal to what they were capable of.

Replicating a cheaper version of an already challenged system doesn't close that gap. It amplifies it.

What simulation can unlock

At Verve, we started from a different question. Not: how do we field this methodology faster and cheaper? But: what does this decision need?

Verve Intelligent Personas and Simulations (VIPS) are award-winning dynamic customer simulations built from human truth. Anchored in trusted client data and augmented with curated cultural intelligence, they are constantly updated, fully auditable and validated against humans. Calibrated to reproduce known outcomes with more than 90% quantitative predictive validity.

And because simulations don't fatigue, don't satisfice at question 13, and can hold complexity across hundreds of tasks, the methods can finally run the way they were theoretically designed to.

Take segmentation. Traditional segment work stagnates. By the time findings land, the market has moved. With VIPS, distinct customer groups are grounded in the behavioural and cultural reality of a market and always updated. Not artefacts of a methodology. Actual intelligence.

Regression arrives with its interpretation already attached. The factors driving consideration, loyalty and purchase intent understood within the evolving contextual layer customers live in today. The numbers and the context, finally evolving together.

Conjoint run on an audience that already understands a category and a brand. Utility scores that reflect how specific customer archetypes make decisions — not how AI-generated generics respond to a stimulus they have no relationship with. That distinction matters enormously. A conjoint built on generic synthetic respondents is measuring how AI thinks customers behave. A conjoint built on VIPS is measuring how they do.

And scenario simulation for the question every CMO is asking: what happens if? A competitor enters. Positioning shifts. Regulation reframes the category overnight. Modelled on a validated audience, not demographic proxies and spreadsheet assumptions.

Simulation is only as good as the humans behind it

None of this works without human truth at its foundation.

The validity and contextual depth of VIPS is built on research gathered carefully, respectfully and with genuine fidelity to human behaviour. That doesn't change. What changes is how we get to that truth and what we can do with it.

Somewhere right now, a human being is sitting through a forty-five minute survey about toothpaste. They lost interest around question twelve — not because they stopped caring, but because we asked too much and made it too dull. That’s on us. We designed something nobody would want to sit through. And the data that comes out reflects it.

The most compelling promise of VIPS is not efficiency. It is that when simulations handle the repetitive, high-volume work, real people are freed for the conversations that genuinely need them — the ones requiring lived experience, emotional nuance and the kind of truth no model can manufacture.

The role that emerges

The MRS Delphi Group has always asked what the researcher of tomorrow needs to be. That question has never been more live than now.

When simulations handle the high-volume repeatable work, the human researcher doesn't diminish. But the craft shifts. Less executing a methodology. More building the environment in which insight is generated — and exercising the judgement to know when something is plausible but not true. The human research that remains gets to be what it was always capable of: deeper, more qualitative, more attuned to the nuance and edge cases where true insight lives. And the people being researched are finally asked only for what only they can give.

That’s not a smaller role for humans in research. It’s a better one.

Andrew Cooper is CEO at Verve. 

You can read the new MRS Delphi Group report ‘Who owns understanding?’ here.