What happened

A strategy expert explains that ROI for generative AI can be measured, but current practice is often declarative; a durable framework is needed as models and costs evolve, with three main reasons for measurement gaps.

Why it matters

Understanding how to measure ROI helps firms pilot AI responsibly, avoid overclaiming results, and align metrics across departments.

Levyed by Harmonie Reybaud, the discussion highlights two company types: large groups already using AI at many levels but unsure what the returns are, and PME-ETIs exploring where AI could help, without a clear path. A durable, cross-department measurement framework is proposed to keep indicators meaningful as models and token costs evolve.

The article emphasizes that ROI is often reported rather than proven, urging careful, multi-KPI analysis,such as adoption, costs avoided, reputation effects, and customer satisfaction,to avoid misinterpreting the data and to support scalable implementation.

What this does not tell us

Quotes and conclusions reflect interview-based reporting and practitioner opinion; results are not generalized to all organizations or populations.

FOR PEOPLE

No direction stated

This piece explains real-world measurement without hype, including where numbers come from and what they don’t prove.

FOR AI AND ITS OPERATORS

No direction stated

Not specified in the text.

These are two separate readings of what the sources describe. Reported claims and risks do not by themselves establish a real-world effect.

Original sources · 1
  1. How to Implement a Dynamic ROI Measurement for Generative AI ↗Stratégies · 2026-09-11

Reporting discovered in France. Discovery market does not mean the event happened there.