The computer keeps a list of who used the paid helper and when. You can look at the list to see who really uses it, and give spare ones to people who would.
Microsoft's Copilot usage report shows who has a licence, who used it and when they last did. It's useful for spotting unused licences, but "active" can mean one try in a month, so you need to look at how often and how recently people use it before moving a licence.
The Microsoft Copilot usage report gives enabled users, active users (any user-initiated Copilot feature in the period), prompts and per-app last activity dates over 7, 28, 90 or 180 days, with up to 72 hours' latency. Use recency plus active days for reallocation, and treat "active" as a floor.
What the report is
The Microsoft 365 admin center has a set of Copilot reports under Reports, then Usage. The one I use most for licence decisions is the Microsoft Copilot usage report, which covers people who hold a Microsoft Copilot licence. People without a licence who use Copilot Chat appear in a separate Copilot Chat usage report.
Microsoft's documentation describes the main figures like this. Enabled users are the people with Copilot licences in the period. Active users are enabled users who tried a user-initiated Copilot feature in one or more Microsoft 365 apps in the period. The active users rate divides one by the other. You can choose the last 7, 28, 90 or 180 days.
The user table shows each licensed person's prompts and their last activity date in each app, such as Teams, Word or Excel. Microsoft notes that the last activity date stays fixed whatever period you choose.
The trap in "active"
Active means at least one user-initiated action in the period. Someone who asked Copilot one question in 28 days counts the same as someone who uses it every hour. So an active users rate of 80% can hide a long tail of people who tried it once.
That's why I'd never take a licence decision on the headline rate. I'd look at two columns per person: how recently they used it, and how many days they used it.
Turning it into a decision
These thresholds are my working assumptions, there to be argued with.
| Pattern over 28 days | What I'd do |
|---|---|
| No activity at all | Ask whether they still need it; offer help or move the seat |
| Active on 1 to 3 days | Talk to them before deciding; they may have one high-value use |
| Active on 4 or more days | Keep the licence |
| No licence, heavy Copilot Chat use | Consider them for the next spare seat |
The last row uses the Copilot Chat usage report, which shows prompts submitted and active days for unlicensed users. It's often the best evidence of who would use a licence if they had one.
Timing and privacy
Microsoft says Copilot activity for a day becomes available in the report within 72 hours of the end of that day, in UTC. I'd run licence reviews on data at least three days old for that reason.
By default the admin center hides user names in usage reports. An admin can change that in the organisation settings. Before showing named usage to managers, I'd agree with HR and the works council, if you have one, what the data will and won't be used for. Using it to move licences is one thing. Using it to judge people's performance is something else, and I wouldn't let a usage report do that job.
Where I got stuck
The report tells you who used Copilot. It can't tell you what that use was worth. Someone active on two days a month might use it for the one task where it saves them a day. I haven't found a report that captures that, so for borderline cases I still ask the person. The break-even calculator helps frame the question.
Sources
- Microsoft Learn, Microsoft Copilot usage report, accessed 11 October 2026.
- Microsoft Learn, Microsoft Copilot Chat usage report, accessed 11 October 2026.
- Microsoft Learn, Microsoft 365 admin center usage reports overview (concealed names setting), accessed 11 October 2026.