Three signals that tell you Claude adoption is actually working

Provisioning seats is easy. Knowing whether those seats turned into leverage is the hard part. After watching a lot of rollouts, we keep coming back to three signals.

1. Weekly active rate by team — not org-wide

An org-wide average hides everything that matters. A 55% average can be two teams at 95% and four teams at 15%. Segment by team and the real story appears: where the habit is forming, and where a champion (or a blocker) needs your attention.

2. Depth, not just breadth

One session a week is a tourist. The teams getting value show depth — multiple sessions, longer arcs of work, repeated use across surfaces like Claude Code and Desktop. Watch the shift from "tried it" to "reach for it first."

3. Week-over-week velocity

A single week is noise. The trend is signal:

• Accelerating — usage climbing week over week. Protect it.
• Flat — plateaued. Usually a workflow gap, not a people problem.
• Decelerating — the early adopters moved on and the rest never started. Intervene now.

The goal isn't a big launch-week spike. It's a line that keeps climbing after the novelty wears off.

Each of these is observable from your first week of telemetry — which is exactly the point. You don't need a quarter of hindsight to course-correct. You need the signals in front of you while there's still time to act.