The stake
Every week, five models answer thousands of questions where a Maison could be named. Being absent from those answers is invisible on every dashboard you own. And presence is uneven by construction: in the June 2026 Index, the same Maison scored up to 15.0 points apart between models. Reading one model tells you almost nothing about the four others.
What you see
Weekly share of AI answers citing the Maison, per model. Illustrative · sample data.
What you do with it
What a falling share becomes:
- A named gap in the Weekly Briefing.
- A PR brief aimed at the outlets that feed the weak model.