Why Change Doesn't Stick: Announcements Bought Six Months

Last week I promised the second argument: why an operation that gets fixed in one place so often breaks again somewhere else, months later.

The short version. An operation runs on three layers at once: the leader, the team, and the strategy. Repair one and leave the other two as they were, and the two you skipped pull the repair back down. The layer nobody touched keeps doing exactly what it did before. Until you pick it up yourself, it stays put.

That happens on a delay, often months after the fix, which is why the relapse so rarely gets traced to its cause. By the time it shows up, the people who made the fix have moved on to the next problem, and the relapse looks like something new.

That delay is my own reading of the pattern, from watching it happen.

The six-month bounce

I inherited an organization whose health survey had collapsed.

The diagnostic surfaced six themes. Every one of them was real.

The first instinct in a situation like that is to announce something. Town halls. Statements of intent. Visible movement that tells everyone the leadership has noticed.

Announcements on their own bought about six months. Then the gains drained back out.

The hard part of those months was being watched. I was the one the organization watched to see whether anything would change. Visible activity holds that kind of attention for a while. Then people stop reading the announcements and start reading the results, and underneath, the results had not moved.

Six themes, two causes

When we took the six themes apart, five of them traced back to two causes, and the second cause was produced by the first.

Leadership had failed. That failure produced a collapse in accountability. Everything else on the list was downstream of those two: the symptoms people could see, complain about, and propose fixes for.

The sixth theme was infrastructure. It was noise, real enough to argue about and unrelated to why the score had fallen.

That changed the problem. Six themes invite six workstreams, and six workstreams are exactly what an organization in that condition cannot absorb. Two causes, one feeding the other, is a problem you can design against.

One loop, three places to break

In the introduction to the book I wrote: "you can't lead a team effectively if you're burnt out (Part I → Part II). You can't think strategically if your team is in chaos (Part II → Part III). And you can't sustain strategic thinking without personal resilience (Part III → Part I)."

And later: "When you lack strategic clarity (poor strategic thinking), you work harder instead of smarter, which drains your personal resilience."

Lay those passages over the six themes and this is how I read them. The leadership failure sat in the first layer. The accountability collapse sat in the second, where the team lives. The third layer, strategy, is what should have told everyone what accountability meant from then on. None of the six themes named it. I read that as the third break, not as the absence of one.

The announcements touched the surface of the team layer. They did nothing to the leadership layer that had failed, and nothing to the strategy. So the two untouched layers kept producing the old outcome, and the surface repair wore off on their schedule.

Repairing one layer well would not have saved it either. As I put it in the book: "No amount of personal resilience rescues bad strategy. No amount of team effectiveness compensates for competitive disadvantage."

That is how the relapse hides. The fix and the failure are separated by months, so the failure gets read as a new problem, and it gets a new announcement.

The delay has a mechanism of its own. A surface repair changes what people do while they are being watched. The layers underneath decide what they do once attention moves elsewhere. Attention always moves elsewhere, because the next problem arrives. The stretch from one moment to the other is the delay, and it is long enough that almost nobody connects the end of it to the start.

What the executives see

Mercer's 2025 Executive Outlook Study surveyed 400 business leaders. 81% of the executives said leaders in their organizations struggle to balance long-term strategic planning with short-term operational needs.

The executives were describing the leaders around them in their own organizations. The figure is what they see in the room, which is a harder thing to dismiss than a self-assessment.

It is easy to read that as a time-management finding. I read it as the loop showing up in a survey. Short-term operational needs are the team layer making noise. Long-term planning is the strategy layer going quiet. Four out of five executives see two layers of the loop pulling against each other in their own organizations, with the leader standing between them.

What held without an announcement

The structural fixes on their own produced nothing anyone could see. The governance changed, the conditions that had let the old behavior run were taken out, and from where most people sat it looked like nothing had happened. After a run of announcements, quiet reads as retreat. People decide whether anything changed by what they can see, and a structure nobody can see gets judged as no change at all.

The announcements on their own had already shown what they could do: six months.

What held was both kinds of work, sequenced. The visible proof, and the structure underneath it. Three pieces.

One visible enforcement action. It was the only visible move that worked, because it proved the rules had changed in a way no statement could.

A governance redesign that removed the structural conditions that had allowed the behavior in the first place. Nobody claps for this one.

A measurement system that made the change self-sustaining. A measure that runs itself is the difference between a fix that needs me in the room and a fix that keeps working when I leave it.

That last one is the move. The way outside the loop is a structure that holds the layers together without you, so the repair stops depending on whether you are watching.

Each piece alone had a weakness the other two covered. The enforcement action alone would have been one more announcement. The redesign alone was invisible. The measure is what kept the new behavior in view after attention moved on.

One move this week

On Wednesday I wrote about a client who went around his field rep and came straight to me. This is the move I skipped.

Pick the next problem that reaches you by going around the person who owns it. A customer who skips the account owner. An issue that someone closer to it should have solved, sitting on your desk instead.

Take the conversation. Staying close to customers and to the work is part of the job, and it matters more the more senior you are. Talking to customers is one way to stay in tune with what is happening in their market, and a leader who stops doing it is making a different mistake.

Then hand the problem back. Give the owner direction, and work through it with them until you both understand why it came to you instead of to them. The routing failure is the useful information. Once you both understand it, the owner can fix the cause, and the next one stops at them.

The line I use with my own managers: "We hire managers to tell us what to do, not the other way around." What I mean by it is that I need them to tell me what needs to be done, and to feel they can execute, make mistakes, and learn from them. The learning is what counts.

The layer you did not touch

When a fix you made fades, look at the layer you did not touch before you repeat it louder. Write down the last change you made that wore off, and next to it the three layers: you, the team, the strategy. Mark which ones the change actually reached. The empty one is usually where the delay is coming from.

Run it on Wednesday's call. The fix I made, solving the client's problem myself, reached exactly one layer: me. The team gave up two people to work that was never theirs. The strategy lost the review that had sat on the calendar for a month. Neither layer was repaired by the fix. Both were spent paying for it.

Do this for one change, not ten. One honest mark on one fix will tell you more about how your operation actually works than another round of announcements.

Next issue picks up what this loop costs, and who ends up carrying that cost when the fix still depends on one person.

Oscar

Want to see which layer your last fix never reached? The Resilience Engine Systems Audit maps it: myresilienceengine.com/systems-audit