The true cost of unused Microsoft 365 Copilot licences, and how to calculate yours

The number most finance teams haven't seen

Most organisations that have deployed Microsoft 365 Copilot know, roughly, that adoption is lower than expected. The Microsoft 365 admin dashboard tells you active users. The number is usually disappointing.

What most finance teams have not seen is what that number means in pounds (or in unrealised productive capacity. The licence cost is visible on the budget. The cost of paying for a tool most people have stopped using is less visible. The cost of the productivity gain you are not realising) the reason you purchased the licences in the first place, is almost never calculated.

This article gives you a framework for all three.

The direct cost: what you are paying for capacity you are not using

Start with the basics. Your Copilot licence cost per user per month, multiplied by total licences, gives your monthly spend. Check your current contract for the exact per-seat figure, pricing has moved since launch and varies by agreement.

Now look at your active user rate. Microsoft defines active use as at least one Copilot interaction in the past 28 days. That is a low bar. A user who opened Copilot once, got a mediocre result, and has not touched it since can still register as active. If your genuine, habitual-use rate is around 20%: which is consistent with what most organisations report after the initial deployment period, the effective economics look like this:

How to pull your current adoption data

Before the calculation means anything, you need your actual adoption rate. In the Microsoft 365 admin centre, navigate to Reports → Usage → Microsoft 365 Copilot. The report shows active users over 7, 30, and 90-day windows, broken down by product area, Copilot in Word, Excel, Teams, Outlook, and so on. The 30-day active user figure is the most useful baseline.

Two things to note when you pull this data. First, Microsoft's activity threshold is low (any interaction counts, including clicking the Copilot button and dismissing it. Treat the number as an upper bound on genuine habitual use. Second, the breakdown by product area is usually more revealing than the headline figure. Most organisations find that a small number of people use Copilot in Teams meetings consistently, while almost nobody is using it in Word or Excel) the areas with the most potential for habit formation and measurable output improvement.

The opportunity cost: the calculation that is usually bigger

The direct cost of unused licences is significant. The opportunity cost is typically larger.

Copilot's productivity case rests on time recovery (hours saved from drafting, summarising, searching, and formatting. Published figures vary, and self-reported time savings should be treated with scepticism. A conservative assumption) 30 minutes saved per active user per week, is a reasonable starting point for a back-of-envelope calculation.

Adjust these figures for your headcount, your adoption rate, and your employment cost assumptions. The gap is almost always larger than the figure that first prompted concern about low adoption numbers.

The compounding problem most organisations haven't priced in

The calculations above treat adoption as a fixed number. In most unstructured deployments, it is not fixed, it is declining.

Usage spikes when licences are first deployed. People try Copilot. Some integrate it into their workflow. Most do not. Over time the gap between licence holders and active users widens, and the people who have drifted away develop a settled view: Copilot is not particularly useful for their work.

That view is usually wrong, but it is rational given what they experienced. An undirected first attempt at an unfamiliar tool (no specific task, no brief, no feedback on whether the output was good) produces an average result. One average result is enough to put most people off trying again. Our article on why Microsoft 365 Copilot adoption fails covers the root causes of this pattern in detail.

An organisation twelve months into a Copilot deployment with 20% adoption has not just lost twelve months of potential productivity from 80% of its users. It has created a cohort of people who have decided, based on limited experience, that the tool is not for them. Reversing that perception requires more effort than forming a good habit would have in the first place.

The breakeven question

At what adoption rate does the investment make financial sense? Using the same assumptions:

For most organisations running an unstructured deployment, this breakeven is at or above their current adoption rate. The licence investment is, by this measure, not yet paying for itself.

This is not an argument against the technology. It is an argument about what it takes to make the investment work, and the answer is not a better tool. It is a structural adoption programme.

What changes the number

The organisations that have moved past the breakeven threshold share three structural features: a weekly challenge that gives participants something specific to do, a visible scoring mechanism that makes participation a social act rather than a private one, and a progression path that builds from basic prompting toward capabilities that genuinely change how people work. The evidence from structured adoption programmes (where all three elements are in place from week one) suggests active participation rates of 80–90% by the end of a nine-week period.

At 80% active participation, the numbers in the 100-licence example look like this:

The programme is not an additional cost on top of the licence investment. It is the mechanism that makes the original investment return anything. To build the full financial case, use our framework for measuring Copilot ROI.

Before the next budget review

If you have not run these numbers for your organisation, it is worth doing before Copilot comes up in the next budget conversation. The calculation frames the adoption problem in financial terms rather than dashboard terms, which is a different and more productive conversation to have with finance or with the board.

The variables that matter most are your actual per-seat cost, your genuine (not Microsoft-reported) active user rate, and your average all-in employment cost. The framework above works with any set of inputs.

And if you want to understand the three root causes of poor adoption before you run the numbers, the earlier article (Why Microsoft 365 Copilot adoption fails) covers the structural picture in more detail.