Distribution Is Not a Department

There is a moment, in the life of nearly every brand that publishes, when someone decides to get serious about distribution. Usually it arrives after a good piece of work lands with a thud — a genuinely sharp essay, a report that took a month, a video someone was proud of — and the numbers come back flat. The instinct that follows is almost always the same, and almost always wrong. The work is fine, the reasoning goes; what we lack is a function to move it. So a function is created. A person, then a team, then a scoreboard. From now on, editorial makes the thing, and distribution gets it seen.

I want to argue against this instinct as directly as I can. Not against distribution — distribution is the whole point; work that no one reads is a diary — but against distribution as a department. The claim I am making is a method claim, an operating stance, not a measured finding: you cannot separate editorial from distribution without corrupting both. This is an editorial read of how publishing operations actually behave, watched over enough cycles to trust the pattern. It is not a survey. It is not a benchmark. It is the shape of a mistake I have seen made in enough rooms to name it.

The mistake is structural, and structure is where I want to keep the argument, because structure is what survives the reorganization. When you split "make it" from "move it," you have not divided labor. You have installed two scoreboards in one building, and pointed them at two different things.

Two scoreboards, one reader

Give a function a name and you give it a number. That is not cynicism; it is how functions stay funded. The distribution team, newly minted, needs to show it is working, and the things it can show are the things that are cheap to count — impressions, reach, posts shipped, the raw arithmetic of exposure. This is the oldest gravity in the business: impression-led operations optimize for what is cheap to count, not for what is hard to earn. A standalone distribution function does not choose this target out of laziness. It chases it because it is the target the function was built to be legible against. The scoreboard was installed before the first campaign ran.

Editorial, meanwhile, keeps a different number, or tells itself it does — some blurry sense of quality, of whether the thing was worth making. And for a while the two coexist politely. The trouble is that only one of the two numbers is easy to report upward on a Monday. Reach is a figure. "Was it worth making" is an argument. When the quarter tightens and someone has to justify the line item, the legible number wins, every time, not because anyone is fooled but because it is the one that fits in the cell.

Here is the part that does the real damage. The distribution scoreboard does not stay in the distribution team. It leaks upstream. Once the organization has learned to reward reach, that reward flows back into the room where the work is commissioned, and it quietly reshapes what gets commissioned. The brief starts to ask, before anything else, how will this travel — which is a reasonable question one beat too early, because asked first it selects for the wrong thing. You begin, without deciding to, to make work that volume metrics select for: content engineered to be forgettable on purpose, shaped to be posted rather than to be finished, tuned to the moment of the impression rather than the minute of attention that would have earned the next one. I have written before about how volume selects for forgettable on the production side. The split is the mechanism by which that selection pressure gets inside the editorial room — it is the pipe.

The split almost always arrives dressed as maturity. Separating the functions feels like professionalizing them: specialists for making, specialists for moving, clean handoffs, clear ownership. This is the reasoning that makes the mistake so hard to argue against, because in most parts of a company specialization genuinely is progress. You do not want the person who writes the contract also running payroll. But publishing is one of the places where the specialization cuts across the grain of the work rather than along it. The decision about what to make and the decision about how it will be found are not two crafts that happen to sit next to each other. They are two faces of one judgment, and the professional-looking move — giving each face its own owner and its own metric — is exactly the move that pulls the judgment apart. The org chart gets cleaner and the work gets worse, and because the two happen on different timescales, nobody connects them.

The axis that splits

Every publishing operation is secretly arguing about one axis, and the argument is usually unspoken. On one end is volume — how much went out, how far it reached, the vanity of the count. On the other is fitness — whether the thing did the job it was made for, whether a reader who started it finished it and came back. Volume is a vanity metric; finish-rate is a fitness metric, and the two are not two readings of the same instrument. They are two different bets about what a publishing operation is for.

When editorial and distribution live in one loop, that axis stays a productive tension. The people making the thing and the people responsible for whether it lands are the same people, arguing it out inside a single judgment. When you split the functions, you do something subtler and worse than picking the wrong end of the axis: you assign the two ends to two teams. Distribution owns volume because volume is what it can show. Editorial owns fitness because fitness is what it can defend. And now the tension that should have lived inside one decision is a turf line between two departments — which means it gets resolved by whichever department has the more legible number, which, again, is not editorial.

This is worth sitting with, because it explains why the fix is never "get editorial more power in the meeting." The seam itself is the defect. You cannot negotiate your way to good work across a boundary that was drawn precisely where the two halves of a single judgment needed to stay joined.

The reader is served by neither

Consider what all of this looks like from the only seat that finally matters — the reader's. The reader does not care about your org chart. The reader gives you attention, the one currency you cannot manufacture, and gives it on a single condition: that the thing be worth finishing. The internet rewards what people finish, not what brands publish — I mean this structurally, as the reason the separation fails, not as a statistic. A reader who finishes has told you the work did its job. A reader who was merely reached has told you almost nothing, and the split-brain operation cannot tell the difference, because the team that measures reach never meets the reader who bounced, and the team that cared about the finish never sees the reach report until the number has already been reported as a win.

So the writer optimizes for the editor, and the distributor optimizes for the algorithm, and the reader is served by neither. That sentence is the whole essay, if you want it in one line. Two functions, each rational inside its own scoreboard, combine into an operation that is irrational about the only thing it was built to do.

I have called the broader version of this problem an org chart that manufactures noise — the way a company's structure produces its output whether anyone intends it or not. This essay narrows that idea to a single seam. Of all the boundaries you can draw through a publishing operation, the make-it/move-it boundary is the one that does the most damage per inch, because it runs straight through the middle of a single decision: what is worth making, given how it will be found. Split that decision and you get the number everyone knows is wrong — not because anyone lied, but because the org boundary hands distribution a scoreboard editorial cannot see and cannot answer to. The wrong number is not an accident of measurement. It is manufactured, reliably, by the seam.

Why 2026 makes the split more tempting

If the separation were merely an old habit, it would be fading. It is not fading; it is accelerating, and cheap generation is why. When variants are free — when you can spin up fifty versions of a post, reschedule them across every surface, and let a model write the ninth headline as fast as the first — distribution starts to look like a solved problem. Mechanizable. Something you could hand to a function and a dashboard and walk away from. The volume the scoreboard loves has never been cheaper to produce, which makes standing up a distribution engine and feeding it output the most natural move in the world. Spin up the function, feed it variants, watch the impressions move.

This is not a claim that anyone wins by doing it, and it is certainly not a leaderboard. It is a claim about why the seam is widening. The cheaper it becomes to produce exposure, the more the split looks like leverage and the less it looks like a wound. But the cost of the split was never the cost of moving the work. The cost was always the corruption of what gets made — and cheap generation does nothing to fix that. It pours volume through the exact pipe that carries the wrong scoreboard back upstream. The more frictionless distribution becomes as a mechanical act, the more tempting it is to sever it from the judgment it was supposed to inform, and the more expensive that severing quietly gets.

Refusing the separation

So what is the alternative, concretely? Not a better distribution playbook bolted onto good writing. The method is the refusal itself: we do not separate the two. Editorial-first, distribution-aware — held as one loop, not a handoff.

In practice this means the question how will this be found is asked in the same room, by the same judgment, as the question is this worth making — asked as a constraint on the work, not as a phase after it. It means the person who commissions the piece owns whether it lands, and the person who moves the piece owns whether it was worth landing. The two are not coordinated across a boundary. There is no boundary.

Concretely, it changes what a planning meeting is about. In a split shop, editorial planning and distribution planning are two meetings, and the second one inherits whatever the first one produced — the work shows up as a given, and the only remaining question is how to push it. In a joined loop there is one conversation, and the question of reach enters as a constraint on the idea rather than a task appended to the finished piece. Sometimes that constraint kills an idea that could not have earned its attention, which is the loop working, not failing. The point of holding the two together is not to make distribution more powerful; it is to let the knowledge of how work is actually found do its job at the moment the work is still cheap to change — before it is written, not after it is published and found wanting.

It means, operationally, that you fix the thing that must not move and iterate the thing that can. Cadence is fixed; format is iterable. The promise to show up — the same day, the same standard, without fail — is the commitment that lets a reader form a habit, and it holds only when editorial and distribution are one loop rather than a negotiation between two calendars. Everything downstream of that promise — the shape of the piece, the surfaces it lives on, the way it is introduced — is free to change, precisely because the thing readers count on is not the format but the arrival. A split operation cannot hold this line, because cadence is an editorial commitment enforced by a distribution reality, and when those live in two teams the commitment becomes the first thing traded away under pressure.

None of this requires that distribution be small or unserious. It requires that it not be separate. You can have people whose whole craft is how work is found — you should — the same way you have people whose whole craft is a sentence. What you cannot have, without paying for it in the only currency that matters, is two scoreboards, two rooms, and a seam between them where a single judgment used to be.

Distribution is not a department. It is a property of how you decide what to make. The moment you can point at the team that owns it, you have already lost the thing you were trying to buy.