The Delegation Test: Why Operators Hit a Ceiling They Can't See

Here is a specific failure mode, watched from close range in dozens of companies between $10M and $100M in revenue: the operator promotes someone, hands them a functional area, and spends the next six months doing the job through them anyway. They rewrite the brief, edit the deck, reshape the hire. They tell themselves the person is ramping. They tell themselves the business is at a sensitive stage. Eventually the person leaves and the operator hires again, convinced they just need to find the right one.

This is not a hiring problem. It is a metacognition problem.

What delegation actually requires

To hand a decision to another human being, you have to be able to describe three things about how you would make that decision. First, what counts as relevant information , and what, critically, doesn't. Second, what you are actually weighing when you weigh it, in terms you could explain without using the word "feel." Third, what threshold moves you from one action to another , in numbers, or in conditions specific enough to be testable.

If you cannot describe those three, you have not delegated a decision. You have delegated the performance of a decision, and you remain the decider. This is why your reports keep bringing the same call back to you. They are not incompetent; they are reading the room correctly. The authority was never moved.

Most operators under-describe these because their own decision process is implicit. They know what to do when they see the situation, and they have never had to translate that knowing into language. Tacit skill is fast, adaptive, and essentially un-delegatable. Explicit skill is slower, sometimes wronger, but it is transmissible, testable, and it is what your company runs on above a certain scale.

The ceiling has a specific shape

The operator who cannot explicate their own decision process hits a predictable wall around 40–60 employees. At that size, the number of decisions daily exceeds the number of hours in a day. Something has to give. Three things typically happen, in order. The operator starts sleeping less. Decisions at the periphery degrade, first quietly, then loudly. Good people leave because the authority they were promised is being reclaimed weekly. The operator, watching all three, concludes the company needs to hire better people.

In the best case, someone , a board member, a coach, a very honest spouse , says the sentence that can hurt to hear: the bottleneck is not the people. The bottleneck is that the operator is still the only person in the company who knows how the operator thinks.

What the training actually looks like

The practice here is not complicated, but it is effortful, and it runs against the grain of how most high-performers learned to win.

Before the decision, write out what matters , the three or four variables you are actually weighing, and what ranges of each would shift your call. This is not a document for anyone else. It is an instrument for extracting your own model.

During the decision, notice what you are actually doing. If you are pattern-matching to a past situation, name the pattern and name the disanalogy that almost certainly exists. If you are feeling something strong, name the feeling with precision , not "worried" but "worried about signaling weakness to the board on this specific line item."

After the decision, write the outcome and score your call against your pre-registered probability. Over time, you develop a library of decision patterns with known failure modes and known heuristics. That library is what you delegate.

Why this works faster than it should

Explicit description does two things at once. It improves your own decisions by forcing disconfirmation , the act of writing it down surfaces the part of the reasoning that didn't want to be said out loud. And it produces, as a byproduct, exactly the artifact you need to give your senior people authority they can actually use.

Operators who have done a version of this , what Tetlock's superforecasters do almost automatically, what strong generals do, what the best investors make their analysts do , will tell you they hit the ceiling later, push through it without the two-year stall, and keep the people who would otherwise have left. The compounding is not subtle. A company where the top ten decision-makers have each explicated their own reasoning produces different outcomes than one where the operator is the only one who knows how.

The private version of this discipline is keeping a decision journal. The cohort version is sitting in a room with twenty other operators and being asked to defend your reasoning out loud in front of strangers who will disagree. Both work. The second one works faster.

The test

Here is the test, if you want to run it on yourself. Pick a decision you delegated in the last quarter and that went badly. Open a document. Try to write, in two pages, the decision procedure you wanted the person to follow , with the three items above: what's relevant, what's being weighed, what the thresholds are. If you can do it in thirty minutes, you probably did delegate it and the person made a mistake you could not have anticipated. If you cannot do it without realizing you would have made the call differently in six out of ten possible situations, you did not delegate , and the mistake is upstream of the person you blamed.

Either answer is useful. One of them is where the ceiling is.