Who should get a seat, whether it pays off, and how to measure it. A practical guide to licensing Copilot without paying for idle seats.
Give Microsoft 365 Copilot licences to the users whose work is writing, analysis and meetings, and prove it with a small pilot before you roll it out. It pays off per user when the weekly time saved, valued at the loaded hourly rate, exceeds the monthly licence cost. The bar is low on paper and hard to evidence in practice. Measure usage and outcomes against a baseline, and reclaim seats nobody uses first.
Select by work pattern, not by seniority or department.
| User profile | Recommendation | Why |
|---|---|---|
| Heavy writers and analysts: proposals, reports, long email threads, spreadsheets | Paid licence | The work is text and data that Copilot can draft, summarise and structure. Time saved is large and easy to observe. |
| Meeting-heavy roles: project managers, team leads, account managers | Paid licence | Meeting summaries, action items and catch-up on missed threads are the use cases that show up most reliably in pilots. |
| Occasional users: a few documents a week, mostly reading | Copilot Chat first | The free, web-grounded chat covers general questions and drafting. Move to a paid seat only if usage shows a need for work data. |
| Frontline and shift staff with little desk time | No licence | Too little time in documents and mail to recover the seat cost. Revisit only for specific, evidenced workflows. |
| Executives and senior staff by title alone | Decide by work pattern | Seniority is a poor predictor of use. A seat held for status and not opened is a common source of idle Copilot spend. |
Run a pilot first: a small group across several roles, a fixed period, a baseline taken beforehand. Copilot is assigned per user, usually as an add-on on top of an eligible Microsoft 365 or Office 365 base licence, unless it is included in your suite (for example Microsoft 365 E7), so check what each candidate already holds. An admin can reassign a seat from one user to another, but the number of purchased seats usually cannot be reduced mid-term (under the New Commerce Experience, only within seven days of a purchase or renewal, otherwise at renewal), so avoid buying for the whole company on day one. The free Copilot Chat, which is grounded in the web and available to work accounts, is the right default for everyone else.
A user breaks even when time saved per week × loaded hourly rate × weeks per month ≥ monthly licence cost. Loaded rate means salary plus on-costs, not salary alone.
Worked example with illustrative assumptions only: an assumed licence cost of USD 30 per user per month (use your own price), a loaded rate of USD 60 per hour and 4.33 weeks per month. The break-even is 30 ÷ (60 × 4.33), about 0.115 hours, or roughly seven minutes per week. If a user truly saves two hours a week, the value is around USD 520 a month against USD 30 of licence.
The formula is the easy part. It assumes the saved time is used productively and that the seat is opened at all. A seat with zero usage costs the full licence and returns nothing, whatever the average says. That is why the real question is how many of your seats are active, not what one active user could save.
Take the baseline before rollout: time per recurring task, meeting load, turnaround on typical deliverables, and a short user survey. Without it, any later number is an opinion.
Then track two kinds of metric. Usage shows whether a seat works: active days, apps used, prompts per week. Outcome shows whether it is worth the price: hours saved, faster turnaround, survey results. Microsoft offers usage and adoption reporting for Copilot in the Microsoft 365 admin center, in the Microsoft Copilot Dashboard and in Copilot Analytics within Viva Insights; which reports you see depends on your licences and settings, so check what your tenant exposes.
The first ROI lever needs no survey: reclaim Copilot seats that show no usage over a defined period and give them to people on a waiting list. It is the same logic as any other add-on, covered in the Microsoft 365 cost optimization guide and the licence reclaim guide.
The Zation Platform profiles activity per user across the Microsoft 365 family and Copilot, and shows which Copilot seats are idle and which are actively used, so reclaiming seats rests on evidence instead of a guess. It does not replace the baseline or the outcome measurement above.
Per-user usage evidence before you buy more seats.