Measure usage, not licences

Buying licences is the easy part of an AI rollout. It is also the number most dashboards stop at.

A licence count tells you one thing: that money was spent. It says nothing about whether the tool changed how anyone works.

Four numbers that tell you more

1. Weekly active users. The share of licensed people who used the tool this week. Monthly figures hide the people who tried it once.
2. Task coverage. Which of the tasks you bought it for people actually do with it. A tool used only to rewrite emails is not the tool you planned.
3. Quality. How much of the output passes the team’s own checks. If nobody has defined good output, start there.
4. Drop-off. Who tried it and stopped, and when.

Measure without spying

Usage data is about people, so treat it that way. Report it in aggregate, by team, never as a ranking of individuals. Tell people what is measured and why before the first report goes out. A rollout that feels like surveillance loses the trust it needs.

Drop-off is the most useful number

The people who stopped are your best source. Ask them. The answer is usually one of three things: they were never shown how it fits their work, they don’t trust the output, or the use case they needed isn’t there.

Each of those has a fix. None of them shows up in a licence count.

What does your AI dashboard measure today? Tell me on LinkedIn.