In short: A large share of paid Microsoft 365 Copilot licences sits idle. Independent surveys put weekly active usage at only around 20 to 30 per cent of the seats purchased (Value Add VC, 2026). The bottleneck is not the technology, it is adoption. Three evidence-backed levers turn this around: lock in a handful of use cases, enable people continuously instead of training them once, build internal champions.
The problem: paid for, but unused
Imagine you have licensed software for an entire department and after three months only a quarter of the people work with it regularly. That is exactly what happens with Microsoft 365 Copilot in many organisations, except it is more expensive than it looks at first glance.
The figure that circulates most often, claiming that only around 35 per cent of assigned Copilot licences are actively used and that roughly two thirds therefore lie idle, comes from vendor and consultancy blogs without a traceable primary source and does not hold up methodologically (Peafowl IT / AppWT, 2025). So I treat it only as a rough illustration. The more solid statement is the cautious one: depending on the survey, only around 20 to 30 per cent of purchased seats are used actively on a weekly basis (Value Add VC, 2026; Worklytics, 2025). Well-run rollouts land noticeably higher, but they rarely reach the majority of licences.
An important note for context: even Microsoft's own measurement is generous. In the Copilot usage report, a single use within 28 days counts as active (Microsoft Learn, 2025). A Gartner analyst considers that threshold too low and recommends assuming at least one use per week, ideally per day, if you want real value creation (techpartner.news, 2025). Anyone who wants to measure adoption seriously therefore does not rely on the standard 28-day rate. The genuinely used share is likely to be even lower than the values quoted above.
For the management team and the CFO, the maths is simple. It is not the licence that creates value, it is the usage. Here are the Swiss prices for Microsoft 365 Copilot, each excluding VAT and on top of the existing Microsoft 365 base licence (Microsoft Switzerland, 2026):
- CHF 14.57 per user per month on an annual subscription (promotional price until 30 June 2026)
- CHF 17.00 per user per month on an annual subscription (standard price, after the promotion)
- CHF 20.40 per user per month with monthly billing
With a thousand seats and active usage of 25 per cent, month after month you are paying for 750 dormant licences. Unused seats therefore need to be identified regularly and either activated or cancelled.
Why this happens
It is not the product. The gap between expectation and real-world usage is first and foremost a change and enablement issue, not a technical fault. The contrast in the data is striking. In the 2024 Gartner survey (director analyst Max Goss, 132 IT leaders), 90 per cent said employees would fight for access, while at the same time 57 per cent reported that usage drops off quickly shortly after the rollout and 72 per cent said people struggle to integrate Copilot into their daily work (Computerworld, 2024).
The most common root cause: licences are bought before concrete use cases are defined. Gartner attributes the obstacles overwhelmingly to the organisational side, missing use cases, weak enablement, unclear data governance, rather than to the technology (Gartner, 2025). In the 2024 Gartner survey, only 3 per cent of IT leaders said Copilot was already delivering significant value (Computerworld, 2024). And the jump to scale rarely succeeds. Of the organisations that had completed a pilot, only around 5 per cent moved to a broader rollout in 2026 (Gartner, 2025, 187 leaders).
On top of that, the change effort is systematically underestimated. 73 per cent reported that the rollout required more change management than expected and 87 per cent said end users need ongoing support and training (Computerworld, 2024). If you hand out licences and point people to the onboarding video, the actual work is still ahead of you.
Lever 1: lock in a handful of use cases
Organisations with high Copilot usage do not roll out broadly, they deliberately concentrate on a few everyday tasks. Microsoft itself explicitly recommends in its Adoption Playbook that you focus the licences on a small number of areas first and define concrete use cases instead of spreading them thinly across the organisation (Microsoft Adoption Playbook, 2025).
Practice confirms the direction. Where employees handle a few recurring tasks with Copilot as a fixed routine, stable weekly usage emerges. Microsoft's own user data shows the same focus. Users report the greatest benefit with email processing, first drafts and meeting follow-up (Microsoft WorkLab, 2023). Industry analyses describe the pattern behind high usage as settling on two to three frequently recurring use cases (Value Add VC, 2026). That exact threshold comes from an aggregated market assessment, whereas the principle of few use cases is well supported.
In concrete terms: instead of Copilot for everyone, go for three clearly named routines per team. Proven candidates:
- meeting summaries
- email triage and drafts
- Excel analysis
- first drafts of text
These get practised until they become habit. A document-heavy example shows the potential is real. In Vodafone's legal department, contract review saves around 4 hours per person per week (Microsoft Customer Stories, 2024). Figures like that apply to one specific field of work and cannot be transferred one to one to an entire workforce, but they do show where Copilot creates value particularly quickly.
Lever 2: enable continuously instead of training once
A one-off training session fizzles out. Microsoft describes practical, recurring training as the single most effective investment in adoption and recommends making ongoing training the standard and building a culture of continuous learning, which is precisely not the one-off course (Microsoft Adoption Playbook, 2025). The reasoning: only regular use makes Copilot a natural part of the workflow.
That matches the friction points from the surveys. If 87 per cent of organisations report that end users need ongoing support (Computerworld, 2024), then a single kick-off webinar is structurally too little. What works is structured, continuous enablement, short and regular nudges, tied to concrete tasks, with the chance to ask questions. What that kind of ongoing adoption support looks like is both plannable and measurable.
Some analyses even observe mandatory enablement where usage is particularly high (Value Add VC, 2026). What holds up is that training should be structured and continuous. Whether it strictly has to be mandatory is only supported by secondary sources. The responsible way to put it is therefore: firmly embedded and continuous, rather than voluntary and one-off. This is exactly where the difference lies between a dormant licence and a working one.
Lever 3: build internal champions
Adoption spreads from person to person, not by management announcement. Microsoft names peer-to-peer support among colleagues as one of the strongest levers and explicitly recommends Copilot champions plus a user community in the Playbook (Microsoft Adoption Playbook, 2025). In its own rollout, Microsoft used a hub-and-spoke model with nearly 10,000 champions, train-the-trainer sessions and regular demo appointments (Microsoft Inside Track, 2024).
The core observation from the champs community is remarkably simple. Often a single convincing use case is enough to turn sceptics into users (Microsoft Inside Track, 2024). Champions who live Copilot in their own working context have more impact than any announcement from the top, because they know the language and the tasks of their colleagues.
Another finding fits with that. Cohorts with a clearly assigned task and active support from their line manager keep usage considerably higher than broad blanket rollouts, with a magnitude of two to three times being quoted (consultancy syntheses, 2026). Read that multiplier with caution, as it does not come from a controlled study, but the direction is clear. A concrete trigger per team plus support from the line manager increases sustained usage. Microsoft's own rollout at least confirms the cohort approach by business area and region (Microsoft Inside Track, 2025).
A secondary driver: the EU AI Act makes enablement mandatory
There is a second, often overlooked reason to enable employees in their use of Copilot, and it is a regulatory one. Since 2 February 2025, Article 4 of the EU AI Act requires providers and deployers of AI systems to ensure, to their best extent, a sufficient level of AI literacy among their staff, measured against prior knowledge, experience, education and context of use (EU AI Act, Art. 4, 2025; European Commission, 2025). This obligation applies to every provider and deployer, regardless of the risk class of the AI system, so it applies even at minimal risk.
Relevant for Swiss companies: the EU AI Act has extraterritorial reach. If you place an AI system on the EU market, or its output is used in the EU, you may be covered even without an establishment in the EU (Lenz & Staehelin, 2025; CDBF, 2025). Anyone deploying Microsoft Copilot in their operations generally counts as a deployer of an AI system, and where there is an EU connection, employee enablement falls under the Article 4 obligation (aiacto, 2025; MinnaLearn, 2025). Whether it applies in a specific case depends on the actual EU connection and should be clarified legally.
A purely minimal measure, a single onboarding video or a mere pointer to the user manual, explicitly does not suffice. The measures have to be tailored to role, prior knowledge and context, and ideally documented (European Commission, 2025; Travers Smith, 2025). The obligation is already in force. Official enforcement by national market surveillance authorities only starts from August 2026, and no separate fine is attached to Article 4 (European Commission, 2025; Mayer Brown, 2025). Even so, you should not postpone enablement, since it is the lever that puts the licences to work anyway. The nice part: exactly those measures that make Copilot economically worthwhile also satisfy the regulatory duty of care.
What to do now
Three steps, in this order:
- Make the inventory visible. How many seats are assigned, how many are used weekly? Do not take the flattering 28-day figures, apply a weekly threshold instead. Flag unused seats, for activation or cancellation.
- Use cases before licences. Define two to three frequent routines per team and practise them, instead of rolling out broadly.
- Make enablement permanent. Continuous enablement plus internal champions, that is the lever with the best evidence base and it satisfies the AI literacy obligation along the way.
An honest note on the business case: the Forrester study on Microsoft 365 Copilot, commissioned by Microsoft, quotes around 9 hours of time saved per person per month (Forrester TEI Copilot, 2025). That is a self-assessment from 16 interviews and 367 survey respondents and marks the upper end. A telemetry-based study by Microsoft Research (randomised controlled, 56 companies, more than 6,000 employees) measures more soberly: for regular users, around 30 minutes less per week spent reading emails (across all employees, more like 12 minutes), faster Word documents and no measurable saving for meetings (Microsoft Research, 2025). So calculate conservatively with roughly 0.5 to 2 hours per week. Even the lower end pays off. At a fully loaded cost rate of CHF 60 to 90 per hour, the licence amortises at around 15 to 20 minutes of real time saved per week, provided that time is then used productively.
This is exactly where I come in. I do not sell licences, I extract the value from the licences you are already paying for. That means recurring nudges and support instead of a one-off training session, a few clearly chosen use cases, internal champions who carry the thing, measured by weekly usage rather than by seats handed out.
If Copilot licences are sitting idle in your organisation, a sober look at the current state is worth it. In a free initial conversation we look at your usage situation and name the one or two levers with the biggest effect. What ongoing adoption support looks like in practice is described under Services.
