A restaurant group has an opening playbook with forty checkpoints in it.
The playbook was not written in a conference room. It was assembled the hard way across the first four locations, and most of its items exist because something went wrong once and somebody wrote down what would have prevented it. Checkpoint nineteen is there because of a health inspection at location two. Checkpoint thirty-one is there because of a supplier who did not deliver at location three and a manager who did not have a fallback.
Location five opens strong. The playbook runs, all forty, and the opening goes better than any of the first four.
At location six the team skips three checkpoints. Not out of laziness — they are ahead of schedule, they have done this before, and three of the items look like belt-and-braces for a market they now understand. Location six opens strong too.
At location nine they skip eleven. By now there is a shared understanding, never written down, about which parts of the playbook are the real playbook. Location nine opens strong.
The group is now opening faster than any competitor in its market. Internally, that reads as the system working. Somebody says so in a board meeting, and they are not wrong about the speed.
Location twelve opens badly.
And here is what happens next, in every organization this happens to — including, if you have lived through one of these, yours: they investigate location twelve. The general manager, the market, the buildout, the hiring, the training calendar. They will find something, because there is always something to find at the site of a failure.
What they will not find is the decision that caused it, because that decision was made at location six by people who were right at the time.
How does an organization get faster and worse at the same time?
Because speed and repeatability are two different capabilities, and improving one can consume the other without anybody noticing the trade.
Leadership by Design calls this the tension between Execution and Momentum. Execution is the phase that runs on discipline — the same work, the same way, at a standard that holds when conditions are unfavourable. Momentum is the phase that runs on accumulated movement: the organization is already going, the going supplies its own confidence, and the confidence lowers the perceived cost of going faster.
Both are real capabilities and an organization needs each of them. An organization with Execution and no Momentum is careful and slow, and it loses to people who move. An organization with Momentum and no Execution is fast and it is building something that will not hold.
The tension is not a problem to solve. It is a permanent structural relationship, and what you get to decide is where the threshold sits between them.
What goes wrong is that the threshold moves on its own, in one direction, without you deciding anything.
Nobody in that company chose to stop running the playbook. Every individual choice was to skip three items on a schedule that was ahead.
Why can’t you tell a working control from an unnecessary one?
Because they produce the identical observable: nothing bad happens.
That is the mechanism underneath the whole pattern, and it explains something that otherwise looks like carelessness: why intelligent, experienced operators erode their own controls while paying close attention.
When you run checkpoint nineteen and the opening goes fine, you have learned nothing about checkpoint nineteen. It might have prevented a problem. There might have been no problem to prevent. From inside the result, those look the same.
When you skip checkpoint nineteen and the opening still goes fine, you have generated something that feels like information: evidence that the checkpoint was unnecessary. And that evidence is real evidence. It is an honest observation, honestly obtained, and it points in the wrong direction.
Controls are insurance. Insurance that does not pay out has not been proven unnecessary — it has been unused, which is what you were paying for. But an organization under schedule pressure does not experience it that way. It experiences a cost that was incurred and a benefit that never showed up.
So every successful skip strengthens the case for the next skip, and the case gets stronger the longer nothing goes wrong. Which means the strongest argument for abandoning your controls arrives right before you need them, assembled entirely from true statements.
Why does success erode discipline when failure tightens it?
Because consequence teaches, and success removes the consequence.
Failure is a fast, unambiguous teacher. Something goes wrong, everybody sees it, and you tighten the process — that is where the forty checkpoints came from in the first place. Every item on that list is a scar.
Success teaches too, and it teaches in the opposite direction. The skip that produced no consequence gets reinforced, because nothing punished it and it saved time, and saving time is visible and immediate. The behavior that gets rewarded is the one that removed the control.
That is not a failure of the people involved. It is reinforcement operating the way it operates everywhere else. Behavior that produces a favourable consequence becomes more likely, and faster with no downside is about as favourable as an operating environment gets.
And it compounds, because each successful skip lowers the bar for the next one. Location six skipped three items that looked defensible. Location nine skipped eleven, and the eleven included items nobody at location six would have touched — but location six had established that the playbook was a guideline, and once that is established the argument is no longer about which items are essential. It is about how far ahead of schedule you are.
That’s the erosion. Not a decision. A drift, with a reward attached to every step of it.
What is the difference between momentum and progress?
Progress is movement toward something. Momentum is movement that continues because it is already underway.
They feel identical from inside the vehicle, and that is not a figure of speech — it is the problem itself. An organization moving fast in a sound direction and an organization moving fast in an unsound one report the same experience: energy, pace, a sense that things are happening, a comparison to competitors that flatters you.
The distinguishing question is not how fast are we going. It is what would have to be true for us to notice we were wrong?
If your Execution is intact you have an answer, because your checkpoints are also detectors. Running the playbook does not only prevent problems; it brings conditions into view that would otherwise stay invisible until they became failures. Strip the checkpoints and you have not only removed the prevention. You have removed the instrumentation — and now the first thing you learn about a problem is that it has already happened.
Momentum does not feel like risk. It feels like competence — and it feels most like competence in the period immediately before it stops feeling like anything at all.
Why doesn’t the post-mortem find this?
Because the post-mortem investigates the failure, and the failure is not where the decision was made.
Location twelve gets the full treatment, and the review will produce findings, because a struggling opening always contains something to criticise. A manager who was underprepared. A hiring plan that slipped. A market that was harder than the model assumed. Every one of those will be true, and none of them is the cause.
The cause was a defensible decision made six locations earlier by people operating on real evidence, and there is nothing at location twelve that points back to it. The distance between the decision and the consequence is what makes this survivable as a pattern — if skipping three checkpoints had produced an immediate failure, the group would have stopped skipping checkpoints. The delay is not incidental. It is the enabling condition.
So the review concludes that location twelve had execution problems, which is accurate and useless, and the group tightens up at location twelve while continuing to run a nine-item playbook everywhere else. If you have ever fixed the site of a failure and watched the same failure appear somewhere else eight months later, you have run this loop.
This is the diagnostic descent in miniature: the fracture is not where it hurts, and the phase that produced the failure is upstream of the phase reporting it.
What holds when the organization is moving fast?
Not resolve, and not a reminder. Speed does not create discipline any more than pressure creates readiness — it reveals what was built into the process before the speed arrived.
What survives contact with momentum has a few things in common.
Skipping has to be visible rather than silent. The most useful change available to most organizations is not enforcing the playbook harder. It is instrumenting it, so a skipped item is recorded as a skipped item at the moment it is skipped. Not to punish anyone — to make the drift legible while it is happening, since nobody at location six believed they were starting anything.
Someone has to own the process separately from the schedule. When one person is accountable for both fidelity and speed, speed wins — because you measure speed weekly and you measure fidelity after something breaks. Separating those two is a decision-rights change and it is the highest-leverage structural move here.
The exceptions need a cost. A skip that requires a signature is not bureaucratic friction — it is a governor, and its whole job is to be mildly expensive at the moment when the argument for skipping is at its most persuasive. If it is easy to override, it will be overridden, and it will be overridden by intelligent people with good reasons.
And the review has to reach the decision, not the failure. When something breaks, the useful question is not what went wrong here. It is: when did we last change how we do this, and what evidence did we have at the time? That question is uncomfortable, because the answer usually implicates a period the organization remembers as a success.
Where this leads you
To a suspicion you should take seriously if your organization is currently moving well.
The four structural tensions in the Leadership by Design framework are the places an organization gives under load, and this one has a particular signature: it does its damage during the good stretch. The other tensions announce themselves through friction. This one announces itself through the absence of friction, which is why it is usually diagnosed in retrospect, from the far side of a failure that seemed to come from nowhere.
If you are faster than you were two years ago, the question is not whether that is good. It probably is. The question is which controls you are no longer running, when you stopped, and whether anybody made that decision on purpose.
Most organizations cannot answer the second part. That is the finding.
The four structural tensions → Where organizations give under load, and how to read yours.
How to trace a failure back to its source → The diagnostic descent, worked end to end.
