Split decisions by reversibility, not by seniority

Ask a founder of a company of fifteen people what the bottleneck is and you will often get an answer about hiring, or tooling, or a process that needs writing down. Sit with the calendar for a week and the answer is usually simpler. Every decision routes through one person, and that person is the constraint.

This is not a failure of delegation in the character sense. It is what happens by default. In the early years the founder genuinely was the only person who could decide anything, because there was nobody else and no basis on which anyone else could judge. The company then grows past that point without anyone noticing that the arrangement no longer fits, because nothing breaks loudly. Things just get slower.

Why the usual fix does not hold#

The standard response is to delegate by seniority. Give the head of delivery authority over delivery decisions, the head of sales authority over sales decisions, and so on.

It works for a few weeks and then leaks. The reason is that seniority does not track the thing people are actually anxious about. A junior person can safely change the format of a weekly report. A senior person signing a three-year lease is a different category of act, and everyone knows it, so the escalation quietly resumes for anything that feels weighty, which is most things.

The result is a delegation policy on paper and the original bottleneck in practice, plus a layer of ambiguity about which is in force.

The split that does hold#

Divide decisions by how hard they are to undo.

Reversible. If it turns out wrong, you can change it back within a week at a cost you would not need to discuss. A named owner decides now, without asking, and does not report it as an exception. Most decisions in most companies are in this group, including a good number that currently escalate.

Irreversible. Money committed for a long period, anything with a customer’s data or a legal obligation attached, anything that would be embarrassing to reverse in public, anything affecting a person’s employment. These escalate, and they are written down before they are decided: what is being decided, what the options were, and what would make us revisit it.

The founder sees the second group only. The first group runs without them.

Two things that make it stick#

Name the owner, not the committee. A decision assigned to a group is a decision assigned to nobody, and it will find its way back to the founder within a month. One name per class of decision, written down where people can see it.

Protect the mistakes in the reversible group. The first time a delegated decision goes badly and is met with a post-mortem in the founder’s office, the delegation is over. Everyone in the room learns that the authority was conditional on being right, which is not authority. Reversible decisions are supposed to be wrong sometimes. That is the entire basis on which they were delegated.

The corollary is that if a decision cannot be allowed to go wrong, it was not reversible and it was classified incorrectly. Fix the classification rather than the person.

What changes#

The visible change is speed, and it arrives quickly, because a large volume of small decisions stop queuing behind one calendar.

The less obvious change matters more. Writing down the irreversible decisions creates a record of reasoning, and the value of that shows up eighteen months later when somebody asks why the company works this way and there is an answer other than a shrug. Most of what gets called institutional knowledge is just the rejected options nobody wrote down.

This is not a restructure and it does not need new software. It needs a list of decision types, a name against each, and a founder who holds the line the first three times something comes back that should not have.