Answers

We bought Copilot and nobody uses it. What should we do?

Low Copilot usage usually means people were given a tool without a reason to change how they work. Before cutting licenses, check whether people are using other AI tools instead. Then pick one team and one recurring task, show what good use looks like on that task, and measure the result. Renew or cut on that evidence.

Take the free AI Profit Readiness Assessment

First, check what "nobody" means

A usage report counts activity in one licensed tool. It does not count AI use elsewhere. In BCG's 2026 survey, 74% of frontline employees described themselves as regular AI users (BCG, June 2026), and in the 2025 survey 54% said they would use AI tools even if not authorized (BCG, June 2025).

So low Microsoft 365 Copilot numbers can sit alongside plenty of AI use in other tools. Ask a few teams what they use and why. The answer changes the problem from "nobody wants AI" to "our tool is not the one they reach for", and those have different fixes.

The usual reasons a licensed AI tool sits idle

Nobody pointed it at a task. People were told it was available and left to find a use. Most never found one that mattered.
The first try was disappointing. Someone tried it on an obvious task, got an average result and did not come back. Getting past the first use is a skill, and few people were shown how.
The work around it did not change. The tool was added to a process designed without it, so using it feels like an extra step.
People are unsure what they are allowed to do. Without clear guidance on what data can go into it, careful people leave it alone.

A 30-day plan for one team

1. Week one: pick one team and one task they do every week. Record how long it takes and how often it needs rework.
2. Week two: ask the person who uses the tool best to show the others, on that task, in their real work.
3. Week three: change one step of the process so the tool sits inside the work instead of beside it.
4. Week four: measure the task again and decide, in writing, where any saved time goes. Use that result to decide renewal for that team.

One team with a measured result tells you more than a company-wide usage dashboard, and it gives the next team something real to copy.

What to bring to the renewal conversation

Finance will usually see the renewal as a cost line with a usage number beside it. Bring the 30-day result instead. Show the task you chose, how long it took before and after, and where the saved time went. If the test team's work changed, you have a case for renewing that team and running the same test with the next one. If it did not, you have a reason to reduce seats there without guessing about the rest of the company.

Either way, the decision rests on what happened to the work, which is the question finance is asking.

This is rarely a Copilot problem

The same pattern shows up with any AI assistant a company licenses. The AI Profitability Gap™ is the distance between what a company expected AI to deliver and what it is delivering, and it comes down to three conditions standard reporting misses: Direction, Skill and Reinvestment. An idle license is usually a Direction and Skill problem showing up as a usage number. Switching tools without fixing those tends to reproduce the same dashboard a year later.

Frequently asked questions

Should we cancel our Copilot licenses?

Not across the board until you know why usage is low. Run a 30-day test with one team on one task, then renew or cut licenses team by team based on whether the work changed.

Why is nobody using Copilot?

Usually because nobody pointed it at a specific task, people's first attempts were disappointing, the process around it did not change, or people are unsure what they are allowed to use it for. Some may also be using other AI tools instead.

How can we see who is using Copilot?

Your Microsoft 365 administrator can usually pull usage reports by person and app. Treat them as a starting point, because they show activity and say nothing about whether the work changed.

Is low Copilot adoption a training problem?

Partly. Training helps when it happens on people's real work, and BCG found that five or more hours, especially in person and with coaching, makes people significantly more likely to become regular users. Training without a clear purpose for the tool rarely lasts.

Should we mandate Copilot use?

A mandate will raise activity in the usage report. It is unlikely to change the work, and it can push the people who were already skeptical further away.

Next step

Before the renewal decision, see whether Direction, Skill or Reinvestment is holding your team back. Eight questions, about two minutes.