When a Decision Becomes a Commitment

The rule that got you through the decision is the wrong rule a year later

Your team sorted a choice by how hard it would be to undo, found the answer was an afternoon, and decided it in ten minutes. That was correct. How long does it stay correct?

 

The sorting habit is a good one and this piece is not an argument against it. It has an expiry date, and nothing announces the date.

 

Nothing below is measured. It comes from reading scaling companies closely, and the three markers in the middle section are the part a reader can check against their own quarter.

The same item asks two different questions a year apart

At the moment of choosing, the question is how much scrutiny this deserves, and the answer turns on what being wrong would cost. An afternoon of rework is a small number, so the choice is made quickly and correctly.

 

Some time later the company has built on it, and the question has quietly become a different one: what would it cost to lose this. That number has nothing to do with how long the change takes. A commitment can be reversed in an afternoon and cost a year, and the two figures sit at opposite ends of the same item.

 

The habit does not notice, because it was designed around the first question and there is no moment at which anybody asks it again.

Three markers that the handover has already happened

Nothing announces the transition, so the practical answer is to know what it looks like afterwards. Three signs, and one is usually enough.

 

Something else was built assuming it. Another team's work now takes the arrangement as given, so removing it is no longer one change. This is the easiest of the three to check, because the dependency usually has a name somebody can look up.

 

Somebody outside the team relies on it without knowing. A customer, a regulator or a partner has come to expect the behaviour the arrangement produces, without anybody ever telling them it was a decision. Their reliance is real and it is invisible from inside the team that made the choice.

 

Undoing it now needs a conversation rather than a commit. When the person who would reverse it finds themselves thinking about who to warn, the item has moved. The thought usually arrives before anybody can say why, and it is worth writing down and then checking against the other two markers.

 

An item showing any of the three has become something the company relies on, and it wants the other kind of attention from here.

The optionality objection, which is mostly right

There is a genuine position on the other side and it is held by good operators. Keep everything reversible, commit to as little as you can, and a company can never be trapped by a choice it made when it knew less. Where infrastructure is cheap to stand up and cheap to abandon, that reasoning is strong.

 

It is mostly right, and it addresses a different failure. The mistake it protects against is committing too early, which is real and expensive. The failure described here runs the other way: an arrangement the company already depends on, still being governed as though it were a fresh choice, still being waved through on the grounds that reversing it takes an afternoon.

 

The two disciplines are not competing for the same items. A handful of arrangements want the second kind of attention and everything else wants the first, and in most companies the second set is small enough to say out loud.

What changes when an item moves

Before the handover, an item that unwinds in an afternoon should be decided quickly and pushed down, and that instruction is worth defending against anybody who wants ceremony everywhere.

 

After it, the same item wants a named person and a review date, and it wants them precisely because it is still easy to change. Why that follows is the subject of a companion piece in this series and is not re-argued here.

 

The instruction inverts on the same item and on the same property, and the only thing that changed in between is that the company started relying on it.

What to do this week

Take the last three things your team decided fast and delegated, from a quarter or two ago. Run the three markers over each: has something been built on it, does anybody outside rely on it, would reversing it now start with a conversation.

 

Sort the three into two piles: still a decision, and now something the company relies on. That sort is the output, and it is the whole exercise. What each pile then gets is already settled elsewhere, and the reason companies get it wrong is almost never that they picked the wrong instrument. It is that nobody noticed an item had changed piles.

 

Of the choices your team made quickly last year, which one has moved without anybody moving it?