It’s Never Too Late: The Culture Work That Makes the Technology Work

Picture this. The email arrives on a Monday: “Copilot is now available on your computer. Have fun exploring!” By Friday, a third of the team has tried it, a third is ignoring it, and a third is wondering whether it’s the first step toward replacing them.

That’s not change. That’s installation.

Versions of this are playing out in organizations everywhere as employees are handed enterprise AI accounts: company-provided tools, like Microsoft Copilot, that the organization licenses, sets up and manages for everyone under its own security and data-privacy controls. In most cases, employees don’t have a say. People are told to “play” with it. Maybe formal training comes with the service package. Maybe it’s mandatory for everyone… maybe not. And leaders may believe that once people are using the tool, the change has arrived and the work is done.

The risk is assuming that because the tool is installed, the change is underway. Organizations make that assumption with every kind of change: they invest in the system, the platform and the how-to training, then wonder why the results stay small. Six months later, the new tool is doing the old work a little faster, and the results are… meh.

Change isn’t going through the motions of using something new. It’s a mental shift: doing things differently, doing completely different things, and NOT doing some of the old things at all. To make that kind of change stick, you have to change the culture.

Here’s my view after more than 20 years of helping teams through change: the culture work is always best done before the technical shift. If that isn’t possible, it must be done during it. It’s never too late, and it’s never optional.

Don’t start so big that it fails

A company that has just rolled out enterprise AI might not be doing big things with it… yet. It’s in the Adoption phase, where people have access and are trying the tool on everyday tasks: a summary here, a first draft there. They’re dipping a toe in the water so they can learn to walk before they run.

That’s a smart place to start. In fact, many AI rollouts fail because they start too big. The risk isn’t starting small; it’s starting so big that people are overwhelmed by a task they aren’t prepared for, in an environment without a fully thought-through strategy.

Know that there is no finish line

Think about the last big system change you lived through: a new ERP, a CRM migration, a reorganization. It followed a familiar pattern. Communicate the vision, train everyone, go live, stabilize, move on. Eventually nobody called it “the new system.” It was just the system.

That playbook was built for change that ends. AI doesn’t end. Whatever version of Copilot a company uses today will behave differently in six months, and the tool you master this quarter gets reshaped next quarter.

People feel this. Gartner research reported in Harvard Business Review found that the average employee went through 10 planned enterprise changes in 2022, up from 2 in 2016, and that willingness to support change dropped from 74% to 43%. That much change could make anyone… cranky. Crankiness leads to fatigue, and fatigue leads to resistance.

So the goal isn’t only a successful launch. It’s a team that can absorb the next change, and the one after that, without falling apart each time. That’s a capability, not an event, and capabilities are built in the culture. In their classic study of more than 200 companies, Corporate Culture and Performance, Harvard’s John Kotter and James Heskett found that organizations whose cultures adapted to change dramatically outperformed those whose cultures protected what had always worked. The data is decades old, but old doesn’t mean untrue. I’d argue that the lesson may be MORE relevant now than when they published it.

Understand what is not being said

MIT’s Edgar Schein, in Organizational Culture and Leadership, described culture as having three levels. The first is what you can see: the tools, the org chart, the new icon on everyone’s taskbar. The second is what people say: the launch email, the values statement, the encouragement to “explore what AI can do for you.” The third is what people actually believe, usually without ever saying it out loud. Most rollouts reach the first two levels and stop, but behavior runs on the third.

The noise in people’s heads is often louder than any other sound in the room. Here’s what they may be thinking, but not saying:

  • “If I use this and it gets something wrong, will I be the one who gets blamed?” So people use it only for things that don’t matter, where a mistake can’t hurt them. The tool never gets near the work where it could make a real difference.
  • “If I get faster, I’ll just be given more work.” So people hide how much time the tool saves them. Leadership sees modest gains and concludes the investment isn’t paying off.
  • “If I ask how to use it, I’ll look like I’m behind.” So people struggle alone, give up after one bad result and decide the tool “doesn’t work for what I do.”
  • “This is how they’ll decide which jobs to cut.” So people don’t share what they figure out. Why would you teach the thing you think might replace you?
  • “Leadership has already decided everything. My opinion doesn’t matter.” So people stop raising problems. You learn what isn’t working only after it has cost you something.

Each of these beliefs shapes how people use the tool, or quietly avoid it. And notice that every one of them makes perfect sense from the inside. That’s what makes them hard to see from the corner office.

Actively engage employees

The company may be at the Adoption stage as a whole. That doesn’t mean every person is.

McKinsey’s research on AI in the workplace found that employees are more ready for AI than their leaders think. Some of your people may be further along than you are. Others haven’t opened the tool and are hoping nobody notices. Others are quietly anxious. The same research found employees were three times more likely than their leaders realized to expect AI to replace 30% of their work within a year.

So don’t assume. Ask. Even if you think you know where everyone is, ask anyway. Asking does more than give you information; it shows you’re interested in, and care about, your people. One of the simplest ways to show empathy, and build trust, is to ask thoughtful questions and really listen to the answers. Listen especially hard when you think you already know what you’ll hear.

Leaders have to go first

I talk about this ALL the time in leadership classes: leaders have to go first. Whether it’s running a good meeting, giving feedback or using AI, the leader needs to show that they’re on a learning curve too, AND that they may be struggling like everyone else. That might mean getting certified in AI, experimenting with a tool like Microsoft Copilot and sharing what happened, or admitting what scares or frustrates you.

A leader who asks the team to adopt something they haven’t tried themselves is asking for compliance, not change. People can tell the difference. A leader who says “I tried this, and here’s where it fell short” does more to support the culture than a leader who sends another encouraging email.

Cultural change is not magic

To be clear, culture work is not the only thing that determines whether a new tool pays off. Processes, structure and the technology itself matter, and in some cases they matter more. But they all rest on the people. You can redesign how work gets done and still watch it fail if the people doing it don’t trust each other, don’t trust their leader and don’t feel safe saying, “This isn’t working.” That’s why the culture work belongs first, or as close to first as you can get.

If the tool is already here, start with a conversation

If your company has already handed out the tool, you haven’t missed your chance. Doing the work during the change is far better than not doing it at all. Start this week with a short, private conversation with each person on your team. Three questions are enough:

  1. “How are you using the new tool so far?”
  2. “What’s one thing it’s done well for you, and one thing it’s gotten wrong?”
  3. “What concerns you about it that you haven’t said out loud?”

Then listen. Don’t defend the rollout, don’t correct anyone’s feelings and don’t rush to reassure. The third question will feel awkward, and you may get “nothing” the first time. Ask it anyway. You’re not just collecting answers; you’re showing that the unspoken beliefs are allowed to be spoken.

Before you have those conversations, answer the three questions yourself, honestly, and share your answers with your team first.

The tool is on the desktop. The more important question is what your team believes about it, and whether anyone has said it out loud.

Sources