Every framework has a case it does not want you to raise.
For this one it is not a hard question about mechanism or evidence. It is a question anybody could ask in the first five minutes, and it goes like this: if organizational failure is structural, what do you do about the organization that is well-designed and led by somebody who should not be leading it?
The reinforcement architecture is sound. Decision rights are clear. Escalations return answers. Somebody built it carefully — and if you have ever inherited a well-run organization and watched the person above you undo it, you already know where this goes.
A framework that has an answer to that is lying, and a framework that has no answer to it is incomplete. What follows is neither, which is the most either of us should expect.
Can good structure fix a bad leader?
For a while. Not indefinitely, and the reason is the whole limit.
Structure does constrain behavior, and a leader inside a well-built architecture will make better decisions than the same leader inside a poor one. A threshold that routes a decision to committee routes it whether or not the person likes committees. A review cadence that puts outcomes on the table puts them there whether or not anybody wanted to look. That is not theoretical — it is the entire case for building the architecture in the first place, and it holds.
What it holds against is drift, pressure, and ordinary human limitation. It does not hold against intent.
Because every governor in an organization has an override, and the override belongs to the person the governor was built to constrain. That is not a design flaw somebody failed to catch. It is structurally unavoidable, because an organization with no capacity to change its own rules cannot respond to anything its rules did not anticipate.
So the constraint is real and it is conditional. It holds as long as the person subject to it declines to remove it, and not one day longer.
Architecture is maintained by the person it governs. That is the load limit, and no amount of design engineering gets around it.
What does that look like when it happens?
Slowly, and with a defensible reason at every step.
A leader who wants around a structure rarely dismantles it. Dismantling attracts attention. What happens instead is a sequence of individually reasonable exceptions: this decision is time-sensitive, this review can be deferred until after the quarter, this threshold was set for a smaller company. Each one is arguable. Each one is granted. And the organization, which reads what happens rather than what is stated, records that the structure is negotiable and learns who it is negotiable with.
If that sounds like the Execution and Momentum tension, it is the same mechanism operating on governance instead of process — controls eroding through successful exceptions, with real evidence supporting every step.
The difference is intent. Process erosion happens to people who are trying to do the work. Governance erosion, in this case, happens to a structure that is in somebody’s way.
A structure can be built to make that expensive. It cannot be built to make it impossible, because the authority to maintain the structure and the authority to suspend it are the same authority. You can split them — a board, an outside director, an ownership arrangement — and that helps, and it moves the problem up a level rather than solving it. Somebody, somewhere, is still the last person who can decide, and no architecture reaches above its own ceiling.
What about the opposite case?
The exceptional leader inside terrible architecture, who succeeds anyway. That happens, it is real, and it is more common than the first case.
Somebody with unusual judgment, energy, and personal credibility can hold an organization together across structural failures that should sink it. They catch what the escalation path drops. They know which of their people to check on and when. They carry the institutional memory the systems never captured. You have worked in one of these, or for one of these people. They perform well, and from outside they look like proof that structure is secondary.
What is true about them does not show up on a results page.
The cost is paid by one person, and it is not sustainable at scale or over time. The load that architecture would have distributed sits on somebody’s evenings.
It does not transfer. You cannot promote it, teach it, or hire a second one. What that leader does is not a method, it is a person, and the organization has no way to reproduce it.
And it does not survive them. The most reliable prediction the framework makes about this case is what happens eighteen months after that leader leaves — and it is the same prediction every time, because the thing holding the organization together was never installed anywhere.
So the exceptional leader does not disprove the framework. They demonstrate the cost of not having one, paid in a currency that does not appear until they are gone.
What can architecture not reach at all?
Some things are outside the boundary entirely, and stating them is the honest part of this piece.
Direction. Architecture governs how decisions get made, who makes them, and what reinforces them. It does not tell you which market to enter, which product to kill, or whether the thing you are about to commit to is a good idea. Nothing in your design will answer that for you. That is judgment, it is irreducible, and no reinforcement design produces it. An organization can be superbly built and pointed at the wrong thing.
Care. A governor can force a decision through review. It cannot make somebody want the organization to succeed after they are gone. Indifference in the center is not a structural condition and it does not have a structural repair.
The single instance. Structure operates on distributions, not on Tuesdays. A well-designed organization will have meetings that go badly and decisions that turn out wrong, and any individual case proves nothing about the architecture in either direction. That matters to you directly, because the temptation after any structural change is to judge it on the first case that follows — and the first case is noise.
Conditions with no precedent. Architecture is built out of past consequence — it encodes what an organization has learned. Something with no precedent has no reinforcement history to draw on, and in that moment the organization is running on judgment, which is the thing architecture cannot supply.
Does organizational design replace leadership judgment?
No. And this is where the framework’s position sits, rather than where its critics usually put it.
Architecture does not replace judgment. It decides what your judgment gets spent on.
In a poorly built organization, a leader’s judgment is consumed by things structure should be handling: which escalations to trust, whether the number in front of them is real, who needs to be checked on, which of five competing accounts of the same situation is accurate. That is judgment being spent on reconstruction — on rebuilding, by hand and daily, information the architecture should have delivered.
In a well-built one, the same judgment is available for the things only judgment can do. Direction. The unprecedented case. The decision where the evidence is balanced and somebody has to choose. Whether this is the moment to move.
The framework is not an argument that leaders matter less. It is an argument about what leaders should be spending themselves on, and the claim is that most of them are spending it on drag that a design would have removed.
That is a smaller claim than the framework is sometimes read as making. It is also the one it can defend.
Where this leaves the framework
With a boundary, stated rather than defended.
Leadership by Design holds that a great deal of what gets attributed to people is architectural, that the architecture is legible if you know where to look, and that it can be redesigned. All of that is what the practice sells and all of it survives this article.
What it does not hold is that architecture is sufficient. A well-built organization run by somebody who should not be running it will degrade, and the framework’s contribution is to tell you roughly how long you have and what the degradation will look like on the way down. That is not nothing. It is also not a solution to the problem you have, which is the person.
If you are looking for a framework that removes the need for you to be good at this, this is not it, and you should be suspicious of the one that says it is.
What this offers is narrower and more useful: your judgment applied to the things that need judgment, rather than to the drag your own design has been generating.
The framework, in full → Eight phases, four structural tensions, and the diagnostic move that connects them.
Is this just pop psychology with animals attached? → The other objection, taken seriously.
