Seventeen studies. One population. 0.90–0.98 agreement.
A statutory insurer’s research, rebuilt on Elaiia: 624 twin interviews against 609 real ones, with the same questionnaire and the same sample structure — and one population of twins carried through 17 studies.
agreement with real data
0.90–0.98
Across the full questionnaire — 0.90 over all 96 matched questions, rising to 0.98 by campaign route.
studies, one population
17
The same 208 twins answered study after study instead of being rebuilt for every brief.
decisions, same winner
10/12
Where the real interviews picked a winner, the twins picked the same one in 10 of 12 cases.
Live enterprise validation
Not a one-off benchmark. A research programme, run twice.
Most validations test a single question. This one ran a statutory insurer’s live studies in parallel — real interviews on one side, Elaiia twins on the other — and kept the same twin population for every study that followed.
The setup
Two sides, one design. Real respondents and Elaiia twins answered the same questionnaire, recruited to the same sample structure, so every answer could be compared line by line.
Side | Interviews | What it measures |
|---|---|---|
Elaiia twins | 624 | Simulated answers |
Real respondents | 609 | Observed answers |
The twin population was built once — 208 twins — and then reused across 17 studies, the way an insights team would actually work with it.
What matched
The whole questionnaire, not a handpicked part. Agreement ran from 0.90 across all 96 matched questions to 0.98 by campaign route.
The decisions, not just the averages. The twins pointed to the same winner as the real interviews in 10 of 12 decisions.
Over time, not once. One population of 208 twins carried 17 studies without losing its fit.
The full method
The complete method and results are being published in a joint Suva + Delta Labs whitepaper, planned for November 2026. Until then, the Knowledge Forum recap of Res Unternährer’s talk is the public reference.
At a glance
Full method in the joint Suva + Delta Labs whitepaper, November 2026.
FROM ELAIIA ACADEMY
Stay close to what we’re learning.
New case studies, validation studies, articles and Forum insights from Elaiia Academy. Monthly, no more.
