Commercial aviation defines, for every take-off, a speed after which the aircraft will fly rather than stop. Before it, an abort. After it, you go, and you deal with whatever is wrong in the air.
The rule exists because the worst outcome is neither continuing nor stopping. It is starting to stop after the point where stopping fits in the runway remaining. Landing has an equivalent: criteria that require the approach to be stable by a defined height, with a go-around if it is not — a decision forced before the moment when both options stop being available.
The point of both is not decisiveness for its own sake. It is that at a certain moment the two options stop being alternatives and become sequential, and the second one now starts from a worse position than it would have.
Organisations have the same boundary and almost never name it
Consider the ordinary cases.
A migration that is half-executed, with data in two systems and staff trained on one. Reverting is not returning to the old system; it is running two, badly, having burned the goodwill that made the change possible. An acquisition where integration has begun and the acquired team has been reorganised. Unwinding does not restore the company that was bought. A restructure announced and then partly retracted, which leaves the organisation with neither the old structure nor the new one and with everybody now aware that announcements are provisional.
In each case there was a moment when continuing and abandoning were both genuinely available, and a later moment when they were not. Almost nobody identified either one in advance.
Why late reversal feels like the responsible choice
This is the part worth sitting with, because the instinct is a good instinct arriving at the wrong time.
New information appears during execution. It usually appears during execution, because execution is what generates it. A leader who receives evidence that a course is wrong and continues anyway is being stubborn — that is the failure mode everybody is trained to avoid, and it has a name and a literature.
So the reversal is made in the spirit of prudence, by someone doing what good judgement appears to require. The problem is not the judgement. It is that the option being selected no longer exists in the form it had when the decision was open, and nobody said when that changed.
What the rule actually buys
Not better decisions. Earlier ones, and a shared understanding of when the window closes.
The value of a defined boundary is that it moves the hard thinking to a point where thinking is cheap and information is adequate, and it removes the possibility of a well-intentioned reversal at the moment of maximum cost. It also does something political: it makes continuing a pre-agreed outcome rather than a personal choice, which is what allows the person in the seat to continue without it looking like stubbornness.
This desk has written about the adjacent failures — a successful pilot that will never ship because organisations are good at running trials and bad at ending them, and the reorg announced because a specific uncomfortable decision was available and nobody wanted to make it. Both are the same absence: no named moment at which the question must be answered.
How to install it, concretely
Before a significant commitment, write down two things. The point at which reversal stops being cheap — expressed as a milestone, not a date. And what has to be true at that point for the course to continue.
Then hold a scheduled review at that milestone, with the authority to stop, and treat the decision as made if it passes.
That is all of it. It is not sophisticated and it does not require a framework. What it requires is that somebody name the boundary in a document before execution begins, because the one thing that is certain is that it will not be identified in the middle, by people under pressure, who are trying to do the right thing.
This is a column. The aviation practices described — the take-off decision speed and stabilised-approach criteria requiring a go-around decision by a defined point — are standard in commercial flight operations and are described here in general terms. The NTSB's factual findings regarding the 6 September 2026 accident at Miami International Airport, referenced as an illustration, are as reported this week and covered by this publication; no cause has been determined and no inference about this crew's conduct is drawn here. The organisational argument is the author's own.
Topics leadershipdecision makingmanagementriskoperations
