How Much Analysis Is Enough? Sort the Decision by What It Costs to Undo

Most of the time you spend deciding is spent on decisions that did not need it. Not because you are careless, but because the size of a decision, as you feel it, is a poor guide to the size of the process it deserves. A supplier change feels large. A hire feels large. A pricing tweak on a page with forty visitors a week feels large too. The feeling is the same, and the consequences are not.

The error runs in a particular direction. Founders rarely rush the calls that are truly permanent. They slow down the ones that are not, and pay for it in the capacity the permanent ones need.

What is the difference between a one-way and a two-way door decision?

A one-way door is a decision you cannot walk back through, or can only walk back through at ruinous cost. A two-way door is one you can reverse: you step through, look around, and step out again if the room is wrong.

The vocabulary is Jeff Bezos's, from his letters to Amazon shareholders, where he described the same split as irreversible and consequential decisions on one side and changeable, reversible ones on the other. His point was not that speed beats care. It was that the two kinds need different handling, and that a company which applies the heavy process to both will slow to a crawl on the second kind while gaining nothing on the first.

The framing is now so widely repeated that it has been flattened into a slogan about moving fast. That misses the useful part. The useful part is the sort itself, and the fact that you have to do it before you deliberate rather than during.

Why do we treat reversible decisions as permanent?

Because permanence is the default assumption of a tired mind. When a decision arrives without a label, the safe-feeling move is to treat it as though it cannot be undone. Nothing is lost by caution, or so it seems.

Something is lost. Every decision handled at full weight draws on the same finite capacity, whatever its real stakes, and that capacity is not replenished by the ceremony you wrap around a call that will not matter in a month. The cost of over-processing a reversible decision is not the hour you spend on it. It is the hour, or the clear head, that is missing when the irreversible one turns up.

There is a second cause, which is social. In a team, a reversible decision that goes to committee acquires an audience, and an audience makes it feel permanent. Nobody wants their name on a mistake that others watched being made. So the reversible call is defended as if it were a foundation, and the organisation loses the very thing that makes two-way doors cheap: the willingness to walk back through.

How do you tell whether a decision is reversible?

Ask one question: can you undo this within thirty days, at a cost you would shrug at?

Both halves matter. Undoable in principle is not enough, since almost anything can be reversed if you are willing to pay. The test is whether reversal is cheap, in money, in reputation and in your own time. A price change you can revert by editing a page passes. A contract with a twelve-month minimum term does not. An email sequence you can pause passes. A public statement about a partner does not.

Three follow-up questions sharpen it.

Who else is affected by the reversal? A decision that only costs you something to undo is a different animal from one that costs your customers, your team or a third party. Reversal that spends other people's trust is not free, however easy it is technically.

Does the decision generate information you cannot get any other way? Many two-way doors are worth walking through precisely because the room teaches you something no amount of analysis would. That makes them cheaper than deliberating, not merely equal to it.

What is the worst credible outcome, and could you survive it? Not the worst imaginable outcome. The worst credible one. If you can survive it and reverse from it, you have your answer.

Most calls fail the label of one-way once these are asked honestly. The ones that pass are few, and they are the ones that deserve your best hours.

How much information do you need before deciding?

Less than you think, for the reversible kind. Bezos put the working figure at about seventy per cent of the information you wish you had, arguing that waiting for ninety means you are probably being slow. Treat the number as a rule of thumb rather than a measurement. There is nothing scientific about seventy. What matters is the shape of the claim: past a certain point, extra information buys you less than the delay costs you, and for a reversible decision that point arrives early.

For a one-way door the calculation changes. Here the delay is cheap relative to the error, so the extra analysis earns its place. Take the time. Ask the person who disagrees with you. Write down what would have to be true for you to be wrong, and go and check it.

The discipline is holding both postures at once, and knowing which one you are in. Someone who deliberates at length over everything is not careful. They are undiscriminating, which is a different failing, and it is harder to see because it looks like diligence.

What does sorting decisions by reversibility do to a working week?

It changes what reaches you. Consider a plain week for a founder running a small business: perhaps thirty open decisions, of varying size. Run the thirty-day test and the pile splits. A handful cannot be undone cheaply: a long contract, a hire into a senior role, a change to the core offer, anything that touches money you cannot recover. The rest can be tried and reversed.

Those two piles need opposite treatment. The reversible pile is decided quickly, by the person closest to it, with a note of what would trigger a reversal and a date to look again. It does not need a meeting. It barely needs a second opinion. The irreversible pile is slowed on purpose, and given the concentrated attention the first pile no longer eats.

Notice what has happened. You have not become faster in any general sense. You have moved speed to where it is free and moved care to where it is not, and the total effort has fallen because the expensive process is no longer being applied to decisions that cannot repay it.

This is also why the method resists being turned into a willpower exercise. You do not decide to be braver about reversible calls. You sort first, and the sorting makes the bravery unnecessary, because a decision you can undo does not ask for any.

Where does a reversibility test break down?

In two places, and both are worth knowing.

The first is the slow-drip decision that is reversible at every step but not in aggregate. Each small concession to a difficult client, each exception to a standing rule, can be undone singly. Together they set a precedent that cannot. If you notice yourself making the same reversible call repeatedly, it has stopped being a two-way door. It is a policy being formed by accident, and it wants deciding as one.

The second is the reversible decision made by someone who does not know it is reversible. Your team will treat a two-way door as one-way unless you say otherwise, because they cannot see the label you can. A rule that lives in your head does not protect anyone. It has to be written where the decision is made.

That second problem is the practical one, and it is where a structure earns its keep. Decision Throughput is built for the acute version of this: a queue that has outrun your bandwidth, where the fastest way back to control is to sort what is waiting by what it costs to undo and clear the reversible pile first. For the standing version, where you want the sorting to happen without you, Decision Resolution sets the criteria in writing so that they hold on the days you are not at your best.

Either way, the question to ask of the next decision on your desk is not how important it feels. It is what it would cost to take back.