Making climate risk readable for an insurer
- AXA Climate
- Finance
- Strategy
- 2024 → 2025
Sound climate models, and underwriters who could not act on any of it.
01.The challenge
Show uncertainty without hiding it
The data was there, the science was sound, and the screen stayed unreadable. We worked the question the other way round: what decision does the underwriter have to make, and what do they need to make it?
We started from the decisions an underwriter makes, then worked back to the data strictly needed for each — and looked for a representation of uncertainty the climatologists validate and the underwriters can read.
02.The product approach
Start from the decision to be made, work back to the data.
03.The method
Our approach: draw uncertainty rather than round it off.
001.
Uncertainty that has to be shown
A climate model gives ranges. Hiding them would be false, showing them raw is unreadable.
002.
Two trades that were not talking
Climatologists and underwriters shared neither the vocabulary nor the expectations.
003.
A decision to be made fast
An underwriter assesses a file in minutes, not in half a day of reading.
Filmed underwriting sessions on real files at every iteration, rather than lab tests. An underwriter now handles three times as many files a week, and sends half as many questions back to the science team.
04.Our impact
× 3
files assessed per underwriter per week.
-52%
fewer questions escalated to the science team.
9/10
stated satisfaction after six months of use.
“Six weeks of framing saved us a year of building in the wrong direction.”
