Somebody asked a question once and you answered it.
That is the whole origin. There was no meeting. Nothing went in a policy document, because there was no policy document. A person walked up to you — or caught you in a hallway, or sent four lines in an email — and asked whether they could do a thing, and you said check with me first.
You were right. At the size you were then, you were obviously right. You had the numbers in your head, you knew every customer by their first name and their payment history, and the cost of that call going badly was a quarter you could not afford to lose. Checking with you was the fastest, safest route available, and it was faster and safer than anything else on the table.
Then the company grew, and that answer is still running.
Nobody re-asked the question. Nobody reversed it, because reversing it was never on anybody’s list. It has been doing its job, without interruption, for years — which is exactly the problem, and exactly why you have not gone looking for it.
The decision was correct. That is not the part that went wrong.
Most of what gets written about growth asks you to accept that something you believed was wrong.
This does not.
The decision you made was a good decision. It was made with the information available, at the scale that existed, by somebody who understood the risk better than anyone else in the building. If you had decided the other way you would have been wrong, and the record would show it.
Here is what nobody explains: a decision is an answer to a question about a particular organization. Change the organization enough and the question changes, and an answer that is still sitting there is now answering a question nobody asked.
That is not a failure of judgment. It is the ordinary condition of a company that grew. Organizations keep running answers to questions they no longer have.
There was never a day it failed
Here is what makes these so hard to find, and it is not subtle once you see it.
It stopped working well. It never failed.
Those are different events, and only one of them gets anybody’s attention. There was no morning where the old approach broke and somebody said we cannot keep doing this. If there had been, you would have fixed it — you have fixed a hundred things that announced themselves.
What happened instead is that the arrangement worked a little less well every quarter than it had the quarter before, by an amount too small for anyone to raise. At forty people your answer took two hours. At ninety it took most of a day. At a hundred and eighty it takes three days and the person waiting has learned to start something else while they wait.
No day was the day. There is no incident report. There is no postmortem, because there was no mortem.
And you review what breaks. That is not a flaw in how you operate. It is how anyone operates, and it is the right instinct almost every time. Broken things ask for attention. A decision that never fails never asks to be reviewed. It gets to keep running forever on the strength of never having produced a disaster, which is the strongest recommendation any arrangement can have and the one that tells you the least about whether it still fits.
Why you do not remember making it
The second reason these stay hidden is smaller and stranger, and it is in the title of this piece.
You do not remember making the decision, because from where you were standing you did not make one.
You answered a question. It was Tuesday. Somebody needed to know whether they could move on something, you told them to run it by you, and you went back to what you were doing. It took eleven seconds and you never thought about it again. The person who asked you, though, did something you never saw: they told the next person who asked them. And that person told two more. And within a month it was how things are done here, with nothing written down and no date attached.
That is how most of an organization’s architecture gets built. Not through design sessions. Through answered questions that propagated.
So when you go looking for the constraint, you are looking for a decision — a moment, a rationale, a document — and there is no decision to find. There is a Tuesday that nobody wrote down, including you.
You cannot reverse something you do not remember doing.
Follow the waiting
Which means the search has to run on something other than memory. It runs on evidence, and the evidence is sitting in plain view in your building right now.
Find where work stops and waits.
Work waits for all sorts of honest reasons. Capacity is finite, an outside party has not come back, the input has not arrived, the batch is not full, step four cannot start before step three, a regulator requires the pause, or somebody simply dropped it. None of those is what this piece is about, and a diagnostic that cannot tell them apart is not a diagnostic.
But a surprising amount of organizational waiting is the fossil of an answered question. Somebody, at some point, decided who was allowed to move that work forward. The decision is gone. The queue it created is still there — and a queue you can measure.
So the waiting is not the finding. The waiting is where you dig.
This is not a survey and it does not require anyone to have an opinion. Walk one piece of work end to end. A quote, a hire, an exception, a purchase, a design change, a customer problem that needs a judgment call. Track the calendar time from the moment it was ready to move to the moment it moved.
Then take the difference and ask what it was waiting for. Most of what you turn up will be capacity, sequence or a missing input, and those are ordinary operations doing what operations do. Set them aside.
What is left is the interesting pile. You will find approvals that exist because of a mistake made in 2019 by someone who no longer works there. You will find a routing rule that runs everything through a department that was, at one point, the only department with the software. You will find three people copied on a thing because one of them used to own it. And you will find your own name, more than once, on work you had entirely forgotten you were attached to.
None of that got there by accident. Every one of them was the right answer for the organization that asked, and every one of them is still answering.
How the wrong problem gets expensive
Here is where most leaders have already spent real money.
By the time the waiting is bad enough to notice, it looks like a performance problem, and performance problems have a well-stocked shelf of solutions. So you bought some. Process mapping. A workflow tool. A management training program. Maybe you restructured a department, or hired someone senior to own the mess, or ran an offsite about accountability.
They were not wrong to sell you those and you were not wrong to buy them. Several of them can reach a cause like this one — delegation architecture, governance design and serious operating-system work all can, and good consultants do it. The trouble is what they were aimed at. A structural intervention bought as a performance fix usually gets aimed at the queue you can see, because the queue is what the brief described. It shortens the wait. It leaves the answered question exactly where it is, still routing, still unexamined, now with a faster process wrapped around it.
Nobody dropped the ball here. The ball is sitting exactly where somebody set it down six years ago, and it has been doing its job the whole time.
That is the piece I want you to hold onto, because it changes what you do next. If the slowness is a people problem, you get better people or you train the ones you have. If the slowness is an answered question that outlived its organization, then better people will simply get through the queue faster and the queue will still be there. You can fill a building with excellent people and still route every one of them through a decision made when there were forty of you.
When the queue still earns it
Now the other half, because a piece that turns every wait into a defect is not a diagnostic — it is a crusade with a clipboard.
Some work is supposed to wait, and some decisions should route to one person for as long as the company exists.
The test is not whether something waits. The test is whether the risk the wait protects against still justifies the cost of the wait.
A pricing exception routed to the founder at forty people protected against an unrecoverable margin error at a scale where one bad quarter closed the doors. That is a real risk and the routing was a real control. At a hundred and eighty people, with a finance function and a margin floor and three people who understand the model, the same error is now both less likely and more survivable, while the wait has become far more expensive because much more work is moving through it. The balance inverted. Nobody was watching it invert, because a deal that dies of slowness does not file a report either.
So the question is not why does this wait. It is what is this wait protecting, and does that still buy more than the waiting costs?
Sometimes the answer is yes, and the arrangement stays exactly as it is — now as a decision you have actually made, with a reason you could state out loud, rather than as an answer running on its own recognizance. That is not nothing. An arrangement you have chosen behaves differently from one you inherited, because you can explain it, and the people inside it stop treating it as weather.
What to do with the ones that are not
For the arrangements that no longer earn their queue, there is a move, and it is smaller than a reorganization.
Say it out loud, with a date on it.
The reason these persist is that they were never stated, so there is nothing to argue with. An unwritten routing rule cannot be disagreed with, evaluated, or retired, because the moment somebody questions it the answer is that’s just how we do it, which is true and settles nothing.
So state it. Pricing exceptions under this amount are decided by this person, starting the first of next month, and we look at this again in six months. Now it exists. It has an owner, a threshold and a review date, and every one of those three is a handle somebody can grab later.
You are not making a change so much as converting something invisible into something that can be changed. The retirement is easy. The finding was the work, and you just did it by walking one piece of work across your own building with a watch.
The one that is running right now
There is at least one of these in your company today, and you are the least likely person to see it — for the good reason that you are the one who answered the question, and from where you stood it was eleven seconds on a Tuesday.
It has never failed. Nobody has complained about it, because the people it slows down stopped experiencing it as friction a long time ago. They experience it as the shape of the place.
And it is still answering a question about a company that had forty people in it.
You can leave it running. That is a real option, and it costs you nothing today — the same way it cost you nothing last quarter, which is how it got to be six years old. Or you can go find it and decide whether it still earns what it charges.
Either way, the company is being run by decisions somebody made. The only question on the table is whether you are one of the people who gets a vote.
