Rolling out Microsoft 365 Copilot: a 90-day adoption plan

Switching on Microsoft 365 Copilot takes an afternoon. Getting it into daily work takes months. In between, many companies pay for licences few people open. Public estimates of weekly use range from about 20% to 55% of seats depending on the source and the definition, and there is no audited figure. What we do know is that change management weighs heavily: in Prosci’s research, 88% of projects with excellent change management met or exceeded their objectives, against 13% when it was poor.

Here is the plan I follow over 90 days. It uses the broad stages of Microsoft’s adoption guide (get ready, onboard teams, deliver impact, extend) and adds what the guides say less about: who does what, each week, and how to decide at the end.

Before you start: three decisions

• A sponsor who uses the tool. An executive who shows a meeting summary drafted with Copilot does more than a launch email.
• A definition of done. Before the first licence, write down what will make you scale, adjust or stop. I cover this in define done before AI.
• Clean access rights. Copilot finds everything a user can already open. A SharePoint folder shared too widely becomes visible in one question. IT has to go before the users.

Days 1 to 30: a focused pilot

Start small and real: 30 to 100 people, in two or three teams whose tasks suit the tool (drafting, meeting summaries, searching documents).

1. Week 1. Half-day workshops with each team to list their repetitive tasks. Keep five per team, no more.
2. Week 2. Short training on those five tasks, with the team’s real material. Show a Copilot mistake live too: that is what builds trust.
3. Weeks 3 and 4. A 30-minute drop-in twice a week, and a Teams channel for questions. Log every question: it becomes your FAQ.

Measure from day one. The Copilot usage report in the Microsoft 365 admin center shows active users per app. Add a two-question survey at the end of the month: which task do you use it for, and how much time does it save you?

Days 31 to 60: champions, then the next wave

The pilot has surfaced people who already help their colleagues. They are your champions. Do not appoint them, spot them: they are the ones answering in the Teams channel.

1. Give them a clear role. One hour a week agreed with their manager, a monthly check-in with the project team, and early access to new features.
2. Open a second wave of a few hundred people, built on the use cases that worked. Each new team gets a champion from its own line of work.
3. Publish the rules on one page: which data never to enter, what to check before sending, who to ask. That is also a large part of what Article 4 of the AI Act expects from you.

Days 61 to 90: measure and decide

After two months you have enough data to decide. Look at four numbers, which I detail in measure usage, not licences:

• weekly active users, not licences assigned;
• task coverage: how many of the chosen use cases are done with Copilot;
• quality: a sample of outputs reviewed by managers;
• drop-off: who tried and stopped, and why.

Then compare them with your definition of done. There are three outcomes, and all of them are acceptable: scale, move licences to the teams that use them, or stop. The only bad decision is to renew without looking.

What makes this plan fail

• Launching for the whole company on the same day, with no use cases.
• One two-hour training on "the features", unrelated to real work.
• No follow-up after week three, when the real questions arrive.
• Counting licences as proof of adoption.

To see where you stand before you start, the readiness checklist takes five minutes. If you want support over these 90 days, that is the core of my services.

Sources: Microsoft, Copilot adoption guide; Microsoft Learn, Microsoft 365 Copilot usage report; Prosci, the correlation between change management and project success; Worklytics, 2025 adoption benchmarks.