Change Management That Actually Takes

Seventy percent of change initiatives fail.

That's the Kotter statistic. It has been quoted at every management offsite for three decades. It is also, at this point, a cliche that lets the people quoting it off the hook. The number is repeated as if it's the weather — fixed, inevitable, something to plan around — rather than a symptom of how change is actually being attempted.

It's not a weather fact. It is a predictable outcome of a specific set of moves that people keep making because they were trained to make them. And the leaders whose change initiatives do take — the thirty percent — are not smarter or luckier. They're doing a different thing.

I have watched enough of both to have an opinion about what the difference is. It's not communication, not urgency, not stakeholder buy-in. Those are the polite answers. The real answer is closer to structural.

The $2.1M restructure that didn't take

A regional human-services nonprofit I worked with spent eighteen months and $2.1M rolling out a new program model. Six months of consultants. Three board retreats. Twelve all-staff town halls. A fifty-page playbook. An implementation timeline with four phases and thirty-six milestones.

A year later, the organization was running both the old model and the new one in parallel, reporting against the new one to funders, and operating on the old one for actual client work. The ED knew. The staff knew. The board had been told the rollout was complete.

The change had not taken. It had been performed. Those are different things, and the difference is the whole subject of this essay.

When I got involved, the ED was exhausted in a particular way. Not the tiredness of hard work. The tiredness of having spent eighteen months and $2.1M on something, and watching the organization absorb the language of the change without absorbing the change itself. The feeling of being lied to by your own team, except nobody was lying — they were all just doing what the structure was asking them to do, and the structure hadn't actually changed.

That's what change management looks like when it fails. Not resistance. Not sabotage. The opposite. Compliance without transformation.

Why the standard change management playbook misses

Kotter's eight steps (create urgency, build coalition, form vision, communicate, remove obstacles, create short-term wins, consolidate gains, anchor in culture) are correct. They're also mostly sufficient if you're changing the easy thing. The announced thing. The thing that lives in PowerPoints and quarterly updates.

They're not sufficient for the hard thing. And the hard thing is almost always the thing that change initiatives are supposed to address.

The hard thing is the structure underneath the behavior. Not the process, not the software, not the org chart, not the training. The incentives that produce the behavior. The power relationships that make certain conversations impossible. The unwritten rules about who can say what. The resource flows that reward the old model and punish the new one. The feedback loops that keep the old system stable even when everyone has agreed, in principle, that it should change.

Schein had the line on this forty years ago. Culture isn't the stated values. It's the assumptions that have become so embedded they're invisible. Change initiatives that focus on behavior without changing the assumptions produce exactly what that $2.1M restructure produced. People doing the new thing on the slides and the old thing in the work.

So the question isn't how to communicate the change better. The question is what structure is still producing the old behavior, and whether the change is actually changing that structure.

Most of the time, it isn't.

What the structure-level move looks like

Back to the nonprofit. Here's what actually shifted in the next four months.

The program model required client intakes to flow through a new triage protocol. On paper, every staff member had been trained. In practice, nobody was doing it, because the old intake system was still the one that produced the reports funders read. The structure rewarded the old behavior. Training rewards nothing.

We killed the old reporting system. Not modified. Killed. The funders were renegotiated to accept the new triage data. It took six weeks of conversations with three program officers. Two of them pushed back hard — one threatened to pull funding, and we had to rebuild the relationship from scratch over the following quarter. The third program officer said something we hadn't expected: she'd been privately frustrated with the old report for two years but felt she couldn't ask the organization to change it. That was new information. It also wasn't conclusive — she left her role four months later and the replacement went back to wanting the old format, which we had to relitigate.

The power structure around case assignments had two senior staff members who had run the old model for a decade. Their personal networks and institutional knowledge made them the informal deciders on every complex case. The new model required case assignment to run through a committee. On paper, the committee existed. In practice, the two senior staff made the calls before the committee met.

We did not fire them. We moved them into explicit authority roles on the new committee and restructured their comp so that the new-model outcomes were what they were measured on. It took a board conversation and three months of negotiation. One of them adapted. The other left. The organization kept its institutional knowledge and also got out from under the power structure that had been blocking the change.

The resource flows around funding moves on case volume. Under the old model, more cases meant more revenue. Under the new model, the right case fit mattered more than total volume, but nobody had redesigned the budget process to reflect that. The program directors were still building their budgets around case counts because that is what the budget template asked for. We rewrote the budget template. It took four weeks. Every subsequent budget conversation got materially easier.

None of that was in the original fifty-page playbook. All of it was the change that actually took. The announced change was a program model. The real change was the three structures underneath it: the reporting system, the power hierarchy, and the resource flow. Change the announcement without changing those and you get performance. Change those and the announcement becomes superfluous. I wrote about why these structures stay invisible until they break in Invisible Architecture: The Four Forces.

Why this is uncomfortable, and why most leaders avoid it

Structure-level change is harder than behavior-level change for a reason that almost nobody names out loud.

Behavior-level change can be done through communication and training. It is knowable. It can be project-managed. It can be presented to a board with a milestone chart. The work is transparent, which means the leader looks competent while doing it, and if it fails, there is a clean narrative: we did the change management, but the team resisted, we'll try again next cycle.

Structure-level change can't be done through communication. It requires the leader to redesign things that other people own. Reporting systems owned by the finance team. Power relationships owned by long-tenured staff. Resource flows owned by program directors. The budget template owned by no one in particular and therefore defended by everyone. Every one of those redesigns requires a conversation that carries actual political weight. Somebody loses something.

The grief of real change management is that the leader has to be willing to take those conversations. Not delegate them. Not communicate around them. Actually have them. With specific people, about specific structures, where somebody is going to feel displaced even if the change is correct.

Most leaders don't want to have those conversations. They want the change to happen because everyone agrees it should happen. Everyone agrees it should happen and still nothing changes, because agreement isn't structural. That's the whole problem.

The thirty percent of change initiatives that succeed are usually led by leaders willing to have the uncomfortable structural conversations. I'm not claiming that's the only variable — I've seen changes take for reasons I couldn't fully explain, and I've seen changes fail despite the leader doing the hard thing correctly. But the willingness to take structural conversations is the closest thing to a reliable predictor I've found, and most of what gets called "change management best practice" is downstream of it.

The question I ask when I'm brought in mid-initiative

When a leader calls me in the middle of a change initiative that isn't taking, I don't ask about the communication plan. I don't ask about buy-in. I ask them what underlying structure the change is supposed to shift — not the announced outcome, but the reward system, power relationship, or resource flow underneath it.

Then I ask who owns that structure. Specific name.

Then I ask whether they've had the direct conversation with that person. Not about the change in general. About the specific redesign that's going to reduce their control, their budget, their informal authority, or their visibility.

If the answer to the third one is no — and it almost always is — we've found the actual work. The change isn't stuck because the team doesn't understand it. It's stuck because the structural conversation that would make the change take hasn't happened yet. Everything else people call change management is downstream of whether that conversation gets had.

I'm not saying the conversation is easy. I'm saying it's the conversation. Most leaders spend years running around it. The leaders whose changes take stop running around it and have it.

That's the thing. Or most of the thing. I've had a few cases where the leader had the structural conversation, did the redesign, and the change still didn't take — usually because a second structure they hadn't named was holding the first one in place. Real systems rarely have a single load-bearing piece. But I've never once seen a change take without at least one of those conversations happening. And the absence of even one is the best predictor of the seventy percent.

The playbook, the training, the communication can follow. None of it's the work.