FlashInsight

Glossary

Digital twin

Also called twin · individual twin

An attempt to replicate a specific, real individual rather than represent a segment. The strongest claim in the field — and therefore the most demanding to evidence.

A digital twin sets out to replicate a specific, identified person: their recorded answers, their documented preferences, the way they trade one thing off against another. It is a stronger claim than a persona, which represents a segment without claiming to be anybody in particular.

A looser second sense circulates too: an aggregate prediction offered as the “twin” of a population. Both usages exist and we are not going to legislate one of them away. But they do not call for the same evidence.

The hierarchy of proof

The more individual the claim, the heavier the burden. Saying a model recovers a group average is checkable on an aggregate. Saying it recovers this named person’s answers means showing that person’s answers, at the individual level, on questions that were not used to calibrate the model in the first place. That is almost never what gets shown.

What a buyer can check

At which level the headline accuracy was computed: individual, segment, or whole population? An “85% match” means nothing until you know 85% of what — and the answer changes the value of the promise entirely.

See also

  • Synthetic persona — The briefing given to the model: the traits, attitudes and characteristics that steer its answers. A specification, not a person.
  • Synthetic respondent — An instance of a model asked to answer questions as a human would, calibrated on data describing a real population. The output — the unit a sample is made of.
  • Agent-based research — A design in which respondents are not merely questioned but set in motion over time, so that one reaction can produce the next.

Further reading

Back to the glossary