To get employees to use Copilot after rollout, you have to change how people actually work, not just hand them a license. Access is not adoption. Most companies learn this the hard way: the tool goes live, the launch email goes out, and usage flatlines within a few weeks. The seats are paid for. Nobody's in them.
If that sounds like your rollout, you're not behind. You're in the majority.
Here's the part most people miss. The problem isn't the tool. Copilot works. The problem is that nobody owned the messy human work of getting people to change how they do their jobs, and that work almost never sits with the team that shipped the software.
Why Copilot Adoption Stalls After Rollout
Boon sees the same pattern across its client base. IT runs a clean deployment. Licenses provisioned, security configured, a launch announcement, maybe a lunch-and-learn. Then the numbers come back and daily active usage is a fraction of the seats purchased.
The people who did log in used it once or twice to write an email, decided it was fine but not essential, and drifted back to how they've always worked.
That's the thing about a tool like this. It doesn't fail loudly. It fails quietly. People aren't refusing to use it. They just aren't reaching for it, because reaching for it means breaking a habit, and breaking a habit takes more than a login.
Research on workplace technology consistently shows that adoption depends far more on whether people rebuild their daily routines around a tool than on how good the tool is. That tracks with what we see: the companies with the best Copilot numbers aren't the ones with the best deployment. They're the ones where someone owned the behavior change.
Deployment is an IT job. Behavior change is not.
Should IT or HR Own Copilot Adoption?
This is uncomfortable for a lot of organizations, because it means the current owner of the rollout is the wrong owner.
IT is very good at the layer it owns. It gets Copilot into the environment safely, integrates it with your Microsoft 365 stack, and locks down permissions so nothing leaks. That's real work and it matters.
But getting a marketing coordinator to stop drafting from a blank page and start prompting instead? That's not a permissions question. That's a habit question. It lives at the human layer, and the human layer is not IT's job to run. We unpacked this in why IT-led AI rollouts stall. The short version: nothing went wrong with the deployment. The deployment was never the hard part.
Getting employees to use Copilot is a change management problem, and change management at the human layer is exactly what HR and L&D are built for. Not a vendor. Not a Teams channel full of tips nobody reads. When HR steps into the rollout, the whole thing changes shape. Instead of "here's a tool, figure it out," you get someone thinking about how different roles actually work, where the friction is, and what would make a nervous employee try something new without feeling stupid.
That last part gets underrated. A lot of people don't use Copilot because they're quietly worried about looking incompetent, or about being replaced, or about getting the prompt wrong in front of their boss. That's not a technical barrier. It's a confidence barrier, and it's the kind of thing coaching handles well and a rollout email handles not at all. We made the broader case in how HR can lead AI transformation and who owns AI adoption, IT or HR. Sort that ownership question out first. Until someone owns the human layer, nothing you do on the tooling side will move the number.
What Actually Gets Employees to Use Copilot
Boon has watched enough of these rollouts to know what separates the ones that stick from the ones that fizzle. It comes down to a handful of things, and none of them are the tool itself.
- Role-specific use cases, not generic demos. "Copilot can summarize your emails" is not a reason to change how you work. "Here's how someone in your exact role cut their weekly report from two hours to twenty minutes" is. People adopt what they can see themselves doing.
- Managers who model it. If a manager doesn't touch Copilot, their team won't either. Full stop. Adoption follows the manager, every time.
- Permission to be bad at it first. People need a low-stakes place to fumble through prompts without an audience. The fear of looking dumb kills more adoption than any technical issue.
- A reason that connects to their goals. Not the company's productivity metrics. Theirs. What does this free them up to do that they actually care about?
- Someone checking in after week one. The launch is not the finish line. The drop-off happens in weeks two through six, and that's exactly when everyone stops paying attention.
Notice what's on that list and what isn't. There's no line item for "better training video." The things that move the number are human, relational, and ongoing. They're coaching problems, not content problems.
Why Champions Programs Burn Out
Almost every Copilot adoption guide tells you to build a champions program. Pick enthusiastic early adopters, give them a badge, let them evangelize. It's on Microsoft's own adoption pages and it's genuinely not bad advice.
But here's what it misses in practice. Champions are doing this on top of their real jobs, usually with no support and no structure. The enthusiasm is real for about a month. Then the champion has a busy quarter, the Teams channel goes quiet, and the whole thing dies. A champions program without a support structure is really just a volunteer program, and volunteer programs run on goodwill until the goodwill runs out.
That burnout is the whole problem, and it points straight at the fix. The answer isn't to scrap champions. It's to give them what they're missing: someone developing them as they do the work, so the weight of carrying the rollout doesn't fall on enthusiasm alone. That's the piece we cover in how to build an AI champions program. Champions don't need a badge. They need coaching, so they grow while they carry the rollout instead of just spending themselves on it.
Coaching Is How the Behavior Change Actually Happens
So if training videos and champion badges aren't enough, what does the human layer take?
Coaching. Specifically, coaching that lives in the flow of work rather than in a separate program people have to remember to attend.
Here's why it works where training doesn't. Training is one-directional and generic. It happens once, to everyone, in the same way. Coaching is personal and ongoing. It meets a specific person at their specific point of friction and stays with them through the awkward part where the new habit hasn't formed yet.
Boon's program data shows session attendance runs at 89%, which matters here because the whole problem with Copilot adoption is that people don't show up for the change. When people actually engage, behavior moves. Boon's program data also shows a 23% average improvement in competency across engagements, and coaching NPS sits at +87. Those numbers exist because coaching does the thing a rollout can't. It stays.
This is the model behind Boon Adapt, which puts coaching right inside Slack and Teams so it reaches people where the work already happens. If you want coaching to scale to an entire workforce and not just the leadership team, Boon Scale is the 1:1 version built for everyone. And because managers set the ceiling on adoption, coaching for managers is usually where coaching has the most impact. For the mechanics of why this drives adoption, there's a good overview in AI transformation coaching. People change behavior in relationship, not in isolation, and coaching is a relationship built for exactly that.
How to Measure Whether It's Working
One reason Copilot rollouts stall is that nobody measures the right thing. Everyone stares at license count, which tells you what you bought. It doesn't tell you what anyone's doing.
The number that matters is whether people are using Copilot for the work they were doing before without it, and whether that's still true six weeks after launch. That's behavior, not access. If usage spikes at launch and craters by week four, you don't have an adoption problem you can fix with more licenses. You have a behavior problem, and behavior is what coaching changes.
We put together a practical guide to the right metrics in AI adoption metrics for HR, and there's a broader treatment of proving value on the measuring coaching ROI hub. If you're being asked to justify coaching spend on top of the Copilot spend, start there.
Frequently Asked Questions
How are people using Copilot at work?
Most commonly for drafting emails and documents, summarizing long threads and meetings, pulling information across Teams and Outlook, and getting a first pass on reports or slides. The people who get the most out of it tend to have a few role-specific use cases they lean on daily rather than treating it as a general chatbot they visit occasionally.
Why won't my employees use Copilot after we rolled it out?
Usually because access was provided but no one owned the behavior change. People don't drop existing habits just because a new tool appears. Without role-specific use cases, managers modeling the behavior, and ongoing support past launch week, usage stalls. We break this down in why employees don't use AI tools.
Does forcing employees to use Copilot work?
Mandates create login numbers, not real use. People open the tool to satisfy a requirement and then close it, which looks like adoption on a dashboard but isn't. Sustainable use comes from people finding genuine value in their own work, which is what coaching and role-specific onboarding build.
How long does it take to get real Copilot adoption?
Expect the behavior to form over weeks, not days. The critical window is weeks two through six after launch, when the novelty fades and the new habit either sets or doesn't. That's exactly the period when most rollouts stop paying attention, which is why sustained coaching support through that window makes the difference.
The Cost of Leaving It to the Launch Email
You paid for Copilot. Right now, most of those seats are empty, and the longer they stay empty, the harder it gets to justify the spend. Worse, your people quietly conclude that AI at your company is just another tool that showed up and went nowhere.
That's the real cost. Not the wasted licenses. The message it sends about whether change here is real.
This is HR's moment to own the part IT can't. Boon Adapt puts coaching inside Slack and Teams, meets people at the exact point where they're stuck, and stays with them through the weeks when a new habit is either forming or dying. That's how you get employees to use Copilot. Not with another email, but with someone in their corner while they change how they work. Book a demo and we'll show you what that looks like.