The Frontier Score is 0–100 from three counted pillars at fixed weights — Adoption 50, Activation 25, Security 25 — and lands in one of four bands: Emerging 0–29, Developing 30–54, Advancing 55–84, Frontier 85–100. Counted from your tenant’s own signals; a pillar your team answers instead is labeled self-reported.
What fed this page
- Aristo Frontier Scale — The score, its band and its movement — three pillars at fixed weights. On this run the security pillar is self-reported (a leader answered for it, Microsoft did not measure it), so the score is labeled self-reported wherever it appears and its movement is an estimate. As of 24 September 2026.
- Microsoft Graph — adoption signals — Seats with recent, genuine Copilot activity. Aggregate only. As of 24 September 2026.
- Microsoft license records — Paid, assigned and unassigned seats, and what they cost. As of 24 September 2026.
- Labeled illustrative Shadow-AI scenario — Illustrative aggregate exposure, labeled as an estimate and never a name. As of 24 September 2026.
No figure on this page is a person, and no number here is unlabeled — every one is counted, measured or an estimate, and the word is on the figure.
How the privacy floor holds, and what each of those three words means
Team-level readings are blended at five people or more before any leader surface sees them, the unsanctioned-AI signal is a count and never a name, and there is no per-person view, screen or export of anyone in this system for any leader to pull. Counted means read from your Microsoft records; measured means taken from your own billing export for the period shown; an estimate is a labeled envelope over real inputs, built to under-count rather than to flatter.
The Report carries the full methodology, question by question → · Where your data goes · Every permission, explained