Capable people do not add up to a coherent function. They leave before anyone names what was missing.
What This Argues
Every reorganization rebuilds work that was not established.
Nobody can exercise judgment against a target that was never named.
The people doing the work do not pay for it. Everything built on their output does.
Establishing work is not the same as standardizing it.
Governance that sits outside a standard does not hold after the project closes.
A thing that needs its author was never finished.
01 The Pattern
From where you sit, every part of the work looks finished.
One specialist sat on each team, supporting a different part of an enterprise operation spread across many sites, every one of them competent.
Spread that way, into teams that do not practice the discipline, nobody is left who can say what the function is for.
I noticed it before it was mine to fix.
By the time I led it, it was emptying. Nobody called that a crisis, because every person still there was still delivering.
I could see what the function was for. What I did not know was whether anyone else could say it.
So I asked. The specialists first, then about twenty of their leaders, one at a time, the same question to each. What should this function deliver.
The answers did not match.
The leader before me described the role as a step toward leadership. A place you pass through on the way to something else.
I did not agree, and that disagreement is what follows.
A leader's description of a role becomes the role. Said often enough, it tells the people in it what to expect of themselves and the business what to expect of them. Nobody had established what this function could do, so the description stood in for it. People left, because a step toward something is not a place anyone stays.
Somewhere in your organization, someone is doing work they sense is not theirs and has no words to say so. Ask the functions around you what each one owes the others. The answers will not match.
A function nobody can describe does not fail. It empties.
02 The Problem
Nothing is wrong. Every output is defensible, and they do not combine into a view the business can act on.
Work you can watch surfaces its own problems. A wrong part does not fit, so somebody finds it. Knowledge work surfaces nothing. A wrong number fits, and whoever receives it uses it.
How a number goes wrong
One specialist builds a capacity model, meant to describe what their operation can carry in a normal month.
Then it goes to review, and the questions are all what if. What happens when a tool is down. What happens when someone is still learning the job. Every one is a real condition that has to be planned for somewhere. None of them is a normal month.
Each answer becomes a factor anyway. With no requirement to point at, there is nothing to refuse a reasonable question with, and nobody wants to be the one who had not thought of it. Experience does not change that, because experience has nothing to point at either. The model stops describing a normal month and starts describing a bad week.
No other specialist can say a factor does not belong, because no other specialist does that team's work. So nothing corrects it. The model grows with the person who built it, and whoever arrives next learns the work from that person.
Each person judges their own output against their own sense of what it has to satisfy. Taste should differ, and in work meant to be distinctive it has to. The requirement should not, and here it did.
What drifts is the requirement itself, held by every person separately and tested against nobody. Nobody can tell which difference was a choice.
Why nobody catches it
Every other specialist function in that business had a document saying what its role produced. The document spared them nothing. Naming an output establishes nothing about what it has to contain or what it can assume. A definition tells a function what it owns. What it owes the others stays unwritten.
The number is reviewed and approved. Approval records that someone signed it. It does not examine what is inside.
Nothing else would have found it either. Nothing was written about what the work had to deliver or how to build it, so a specialist learning a method asked colleagues and got a different answer from each. Watching the work would need repetition, and there was none: watching one specialist produces one correct version of something that is correct a different way on every team. The question was never how any one person works. It was what every version of the work had to satisfy.
What the number commits
That number leaves the function. It tells planners what the operation can commit to each month, and it tells senior leaders what to invest in years out. Capable specialists were feeding enterprise supply planning and capital decisions by methods that did not match.
When one of those numbers is built on a bad week, the commitment and the capital are sized to an operation that does not exist. Work gets turned down that the operation could have carried.
Demand outside a business sets a rhythm. The number is what says whether the work inside can meet it.
The same condition, different rooms
The same condition appears wherever people produce something new each time and the results still have to add up to one thing.
On a design led platform it is product teams who each define quality their own way, and releases that stop feeling like one product.
In a creative studio it is the senior people who know what the house does and cannot sit in every review, and junior work that is defensible and still wrong.
In a research organization it is scientists whose work is different every time, estimating their own effort against nothing.
In a luxury house it is the codes that have to read the same across continents and channels.
The common diagnosis is that regional teams reimagine, channels diverge, and fragmentation follows. The common prescription is oversight.
Oversight is a person watching. It does not scale and it does not survive the watcher.
The objection is that a house is one person's vision, and that the vision is the point. It is. The houses still standing are the ones that converted it. Their founders are gone and the work is still recognizably theirs, because what made it recognizable was defined and given on.
Four signals
The work cannot be watched and every output looks defensible, so the condition has to be found some other way. Each of these four shows up while the work still looks fine, which is the only point at which it can be fixed without stopping anything.
SIGNAL 01
Learning the work means finding people
Ask a recent hire how they came up to speed. If the answer is a list of names, knowledge is being rebuilt rather than inherited, and every version moves further from the last.
SIGNAL 02
"We are different" ends every discussion
It is true. The teams differ, the constraints differ, the customers differ. Because it is true, it refuses any standard on accurate grounds, and there is nothing to test it against.
SIGNAL 03
The outputs will not reconcile
Nothing compares to anything else, so nothing can be improved or funded on evidence. Several results arrive, each defensible on its own terms, none comparable to another.
SIGNAL 04
There is nothing to fall short of
With no stated reference point, no gap can be named. Nothing is reported as missing, the silence reads as health, and the condition goes on getting older unrecorded.
A business that grows this way gets larger and less like itself, which is dilution before anyone can name it.
A requirement nobody owns is a preference.
03 The Architecture
You cannot establish knowledge work by recording how it is done.
What gets established is who owns the work and what it has to satisfy. The method stays with the specialist, who is the only one who knows where the work meets their teams.
Three levels. What the work must satisfy. Who it answers to. How it gets done. The first two are set once. The third is not set at all. That is the architecture, and I designed it before a word of any standard was written.
The unit is a body of work. One defined area of what the discipline produces, with its own requirement and its own owner. I settled on seven and placed a sponsor on each.
A discipline without a defined body of work is a group of people with the same title.
Four moves
Writing standards before anyone has asked what a new person inherits, or what the work actually requires, produces a document nobody uses. The order is the design. Each move is only available once the one before it has been done.
MOVE 01
Diagnose
Ask what a new person inherits. Ask their leaders too, and expect the answers not to match. If conversations are the only answer, nothing has been established, and every arrival will rebuild the work from scratch. An org chart names who holds a role, not what the role must produce. A function can perform well for years while missing that.
MOVE 02
Design
Stop asking how the work is done and ask what it requires. Steps differ across every team. Requirements do not. Where nothing can be watched, retrieve the reasoning from the person holding it. Do it with enough of them to separate requirement from preference. Someone who produces something new each time can still account for how they decided.
MOVE 03
Orchestrate
Set the requirement for each body of work and how they connect. Put a team on each one and run them at the same time. Each team takes its proposal to the specialists building the others, then to the sponsor already inside it, with the stakeholders and users who will live with it. Nothing moves to build until that loop closes. One person owns all of them until they transfer to the business.
MOVE 04
Conditions
One place every specialist reaches. A champion for each body of work, the person who built it. A second champion alongside, so nothing stands on one person. Seats turn over. The roles do not. The champions meet monthly. The function's own onboarding is staffed the same way. Nothing runs through the designer. That is not a constraint they worked around. It is the specification.
Three decisions sit underneath those moves. Each is a decision right, and each is made once.
Where the flexibility lives
The requirement fixed three points. What the work takes in, what it has to account for, and what it must deliver at minimum. Nothing else was specified. Not the template, not the tools, not the order, and not how anyone reached their answer. That is where the specialist does their job, because the conditions on one team are not the conditions on another, and the only person who can see them is standing in them. The standard carries the flexibility. Nobody has to ask for it.
Where the categories came from
I took the taxonomy from the body that governs this discipline, then translated it into the bodies of work this organization actually performs. The source gave it standing. The translation made it theirs.
The source has to come from outside the function. A governing body is one. A published framework works, or a regulatory requirement, or the codes the organization was founded on. What matters is that nobody in the room has to defend it as their own idea.
Where the sponsor sits
The sponsor is who the standard answers to, so that seat belongs to the leader who already carries the cost of an unreliable output. They require it to work in their own operation, which puts governance inside the standard rather than around it, and keeps it in use after the project closes.
Cost modeling answered to finance and supply chain architecture. Human factors answered to the safety expert. Five more, placed the same way. Each sponsor received a brief before being asked, because a sponsor who does not know what they are governing is a name on a charter.
Put the sponsor where the consequence already lands and governance stops being a meeting.
What the build was for
I ran the building of the standards the way the standards themselves are built. The standards were the deliverable. The judgment to build the next one was the design.
Every team ran a full project, business case through sustainment plan. None of them had run a project before, and the teams they sat in had no reason to build that in them. I could see what they could carry. The build was designed so they would have to.
The people who wanted to lead a team led one, as long as nothing was at risk on the schedule. Every one of them hit a wall, and the wall was different each time.
One could not get a team moving. I coached them to ask each person what they were good at, and the work moved.
Another was doing all of it alone, and nobody around them was learning. Offer the task and see who takes it.
The leads who came through it trained the next cohort.
Taking the work back would have been quicker. It would also have produced seven standards and nobody who could write the eighth.
A standard says what to do. An operating model says what has to hold.
04 The Evidence
All seven were complete in a year. None of that is the evidence.
What it produced
Each body of work was published with its training and its materials, held in one place every specialist could reach, owned by its champion, the person who built it, with a second alongside.
A new specialist's first usable model took about three weeks, most of it deciding what a model should contain. It takes one, and that week goes to learning the teams the model describes. What left the function could be acted on without being checked first.
The capacity standard now carries what the operation commits to each month and what leadership projects years out. The facility record, which had no owner at all, became the controlled source maintenance, quality, safety and regulatory all work from. None of them built either one, and none of them has to ask me what it means.
Ask those same leaders now what this function delivers and they name it. Nobody was trained to give that answer, which is why I trust it.
Why it stopped emptying
The leader before me called the role a step toward leadership. Once the work had a definition it had a trajectory of its own, and people stopped passing through. I found the function at sixteen. It reached thirty five and has held there. Hiring got it there. Hiring is not what kept it there.
What the restructure left
Year two added more bodies of work and kept the open items moving on the first seven. Then the organization restructured. Departments changed shape, people moved, and reporting lines were redrawn.
The foundation held. The method did not, and it should not have. The same work could not run the same way with fewer people and a different shape around them.
The champions put every open item onto one list and scored it three ways. What it is worth to its own body of work, to the discipline, and to the organization's stated priorities. That is a decision rights instrument, and it replaced prioritization by advocacy.
It has been used at every reorganization since, by champions whose seats have turned over.
When the enterprise asked
The proof came from outside the function.
An enterprise system migration arrived, and every function was asked the same thing: how the product is made, and what the work takes. Accurate enough to commit on. Being asked is one thing. Being able to answer is another.
This function could answer. The requirement already existed and so did the champions, so each round asked less of the function than the one before. Functions that had never established their work were still deciding what a correct answer would look like.
Years later, the leader who owns that planning cycle named the step improvement in it and traced it to the capacity requirement. I had not asked. I was not in the room when it happened.
The planning cycle belongs to planning. It took specialists from this function to map where it broke and why. Owning a process and being able to see it are different skills. Establishing their own work is what taught them the second one.
A function that can see work is worth more to a business than a function that can only do it.
Reinforcement is thinner than it should be. It can still say what its work has to satisfy, which is what the other functions cannot.
An established function can change without starting over.
What held was never the method.
05 The Principle
Establishing work is not the same as standardizing it.
Standardizing says how. Establishing says what the work must satisfy and who answers for it. One removes judgment. The other makes judgment possible, because nobody can exercise judgment against a target that was never named.
What never gets standardized
Judgment. What repeats is the requirement and the connection, never the method.
I learned where that boundary sits at a different company, years before this one, on work instructions where every one was different from the last. I fixed what each had to carry and left the writing open.
Structure did not cost anyone creativity there. It settled what should never have been improvised, and gave them back the writing.
Standardize the method in work that is new each time and you get compliance without thinking. Leave the requirement unstated and nothing has changed. One error costs you the judgment. The other costs you the function.
The test
It runs on its own now. What it gets is refinement, in ways I did not design, by people who were not in the room.
Every one of those changes is evidence.
What you are holding
All of that was one function. What sits underneath it is not a function at all.
Look at your own work. There is a requirement in it only you can state. Not a skill, not a relationship. A requirement: what a piece of work has to satisfy before it leaves your hands.
You probably cannot name it, because nobody has asked. It was never written down, and it never had to be, because you have always been there.
That is not a failure of yours. It is what happens when work nobody can watch goes to capable people and nobody establishes what it owes. The requirement moves into the only place left open to it, which is a person.
The cost does not reach you first. It goes to whoever needs your answer next, and to the people who cannot say what their work delivers without you in the room. It reaches you later, when you cannot leave the work without it following you.
A thing that needs its author was never finished.
Institute of Industrial and Systems Engineers. Body of Knowledge taxonomy.
Nonaka, I. (1994). A Dynamic Theory of Organizational Knowledge Creation. Organization Science, 5(1), 14 to 37.
