It Isn’t the Org Chart That Decides a Good Organization — It’s the Performance of People

Peter Drucker, Management: Tasks, Responsibilities, Practices — Chapter 48: Organization Conclusions

Introduction

Every December, we redraw the org chart. A division appears, a team splits off, someone slides one box to the left. And yet, when the new year begins, last year’s friction tends to return unchanged. The chart has changed; whether the work has gotten any easier is another question.

As a consultant, I took pride in designing the “ideal” organization — principled, airtight, clean to look at. Looking back, I rarely asked what would have to be true for that chart to actually work. Drawing a clean diagram and building an organization that functions, it turns out, are two entirely different things.

Over the past eight weeks, we have worked through Drucker’s chapters on organization design. Chapter 48 closes them, and in it Drucker delivers a cold verdict on the very charts we labor over. The test of a good organization is not the beauty or clarity of its structure. The only test that matters is the performance of the people inside it.

The Assumptions Hidden Behind the Org Chart

Every design rests on assumptions — that under this structure, people will work, communicate, and perform in a certain way. But what holds at the drawing board does not always hold in practice, and some assumptions, when they fail, harden an organization’s arteries. That is why testing them is not optional.

Drucker points to GE. In the early 1950s, its leadership treated every unit with product responsibility as a “manufacturing business.” For some units the assumption simply did not fit: they were innovation units, built to develop a new process or product line. Dropped into the functional structure of a manufacturing business, they suffocated. Yet at the time, the assumption was never even a candidate for testing.

The lesson is simple but heavy. The most dangerous assumptions are the ones that seem most plausible and self-evident. They are so obvious that no one questions them, and so they govern the whole organization, quietly and untested. Drucker, who was himself deeply involved in that GE work, concedes the point: what looks obvious now had yet to be learned then.

This is not a mistake confined to one American company seventy years ago. It is the one we repeat every year, in the name of “benchmarking.”

The Service Got Lighter. The Infrastructure Did Not.

Consider a Korean telecom and media company that ran broadband networks and a pay-TV service delivered over those networks — what’s known locally as IPTV. The service was once a powerful differentiator. But as OTT platforms like Netflix became the norm, paid video-on-demand consumption fell and the pay-TV subscriptions themselves came under threat. A stable infrastructure operator suddenly faced pressure to ship new services fast — and its heavy, slow decision-making could not deliver that speed.

So the company imported the agile model pioneered by a global streaming company. It turned the service-, customer-, and growth-facing areas into squads and tribes, pushing decisions close to the front line, while leaving infrastructure and support in the existing functional divisions. A hybrid. On paper it looked almost like a textbook: teams for innovation work, functional structure for large-scale operations.

The trouble was the premise beneath it — that a lighter service organization would make for faster service decisions. The service side did get lighter, and decisions did move closer to the ground. But most new screens and features are only finished once they connect to the infrastructure, and that infrastructure stayed heavy, slow, and necessarily cautious. An organization responsible for network stability cannot decide at the speed of a service team.

That is where the collision began. A fast call from the lighter service side stalled, again and again, in front of the heavier infrastructure. The side that believed it could move fast and the side that could not meet inside a single decision. Employees, meanwhile, could barely grasp how the new structure was supposed to work. In its first year, the organization fell into confusion.

This is exactly what Drucker flags in Chapter 48. Using teams where teams belong is right. But dragging unchanging, large-scale operating work along at the same pace pushes a design principle past its limits, and the result is only confusion. “A lighter service makes faster decisions” was half true: calls independent of the infrastructure sped up; calls that had to join with it never could.

Copying a structure and verifying that the structure makes people perform are two entirely different things. The form — squads, tribes — can be transplanted, as can the Lego-like flexibility of snapping teams together and apart. But whether that form actually lets decisions flow, inside the weight of this company’s infrastructure, was not something to copy. It was something to test.

There was, to be fair, a reasonable intention behind the change. But in the process, the veteran managers who had spent years moving between complex organizations and untangling collaboration had to step aside — the people the company had trusted with its hardest knots, vacating their seats on the faith that the new structure would do their work for them. Whether that work was actually redesigned into the structure — who would do it, at what level, with what authority — is the open question. I left before I could see how it finally settled.

Eight Chapters, One Story

Drucker names one more principle here: the simplest structure that does the job is the best one. A good structure is judged by the problems it does not create, and the simpler it is, the less can go wrong. There is no universal design principle; each has its limits, and principles are merely tools, used well or badly.

Looking back over eight weeks, the failures we examined converge on one thing. Building the organization before defining the key activities (Chapter 42). Assigning authority by the size of the seat rather than the nature of the decision (Chapter 43). Forcing work that demands different design principles into a single frame (Chapters 45 and 47). Each one mistook structure itself for the goal. Structure is a means to an end, not an end in itself.

Conclusion

As a consultant, the first thing I usually did was benchmark the org chart of a global leader. But a chart, seen from outside, never reveals the assumptions that shaped it — why it was drawn that way at all. And because every company performs and behaves differently, lifting another’s chart wholesale is dangerous.

The question Drucker poses comes earlier than “which structure is better.” What would have to be true for this one to be right — and have we tested that? Designing toward an ideal and testing the assumptions beneath it, he insists, must proceed together. The more elegant the new form, the more carefully you must weigh the confusion it will introduce.

Drucker closes Chapter 48 — and the eight chapters on organization design — with a single line. The test of a healthy business is not the beauty, clarity, or perfection of its structure. It is the performance of people.

So the question to ask after a reorganization is not whether the chart looks cleaner. It is whether people can now carry a decision all the way through — whether, compared with last year, one person’s decision reaches farther and faster. The chart you just drew: is it making people work, or making them relearn how to work?

From our Monday morning Drucker reading group, 7:00 AM.