European AI adoption, measured properly. A methodology note.

Published European AI adoption statistics vary by a factor of five depending on the question asked. That variance is not measurement error. It is four different definitions wearing the same word.

The four definitions in circulation

Any use, which counts an organisation where a single employee used a chat assistant once. Procured, which counts a signed contract regardless of use. Deployed, which counts a system in production. And routine operational use, which counts a system that a defined group relies on to do their work.

The first two produce the headline numbers. The last is the only one that predicts anything.

What we count

A system qualifies for the Future Chronicle index when it is used at least weekly by a defined role group, is integrated into a system of record rather than accessed separately, and has a named owner inside the organisation.

Pilots, evaluations and individually adopted consumer tools are excluded, as is any deployment that cannot produce a usage figure.

Why the integration criterion matters

Across every sector examined, the single strongest predictor of whether an AI deployment is still in use twelve months later is whether it lives inside the tool the user already works in.

Separately accessed tools show a consistent decay curve. Integrated ones do not.

Sector and country coverage

The index covers eight sectors across the Nordics, the Netherlands, Germany, France and the United Kingdom, with figures gathered through direct organisational interviews rather than through respondent self-classification in a panel survey.

Panel surveys reliably overstate adoption, because the person who answers the survey is disproportionately the person who cares about the topic.

What the first results show

Two findings, stated ahead of full publication. Routine operational adoption is substantially lower than headline figures suggest, in the range you would expect of an early technology rather than a diffused one.

And the variance between organisations within a sector is much larger than the variance between sectors, which argues that adoption is an organisational capability rather than an industry characteristic.