18 June 2026 ยท 7 minute read
What a department was holding up
The cost line is visible from the first meeting. The load bearing part takes three months to show up and nobody connects it back.
A department costs a known amount. That amount appears on a report, every month, next to its name. It is the most legible thing about it.
What the department produces is much harder to see, and in the cases that matter most it does not appear on any report at all, because what it produces is the absence of problems elsewhere. Nothing went wrong, so nothing was recorded, so the value shows up as a cost with no visible return.
Then somebody with a spreadsheet and a mandate looks down the cost lines for something to remove, and the least legible department is the most attractive target. Not because it is the least useful. Because it is the least defended.
The three month lag
The pattern is consistent enough to be worth naming.
Month one is clean. The cost is gone, the savings are real, and the people who argued against the cut look like they were being sentimental. Month two is also clean, and this is when the decision gets described as the right call.
Month three is when the work the department was doing starts arriving somewhere else. Not as a crisis, as friction. Things take longer. Something that used to be checked is no longer checked, so a mistake gets through, and the mistake is attributed to the person who made it rather than to the check that was removed.
By month six the cost has reappeared, distributed across four other teams as overtime, rework, slower delivery and a quality problem that customers have started to mention. None of it is labelled. There is no line item on any report that says this is the department you cut.
Where the load went
What gets counted and what does not
| Counted, in month one | Not counted, and arriving later |
|---|---|
| Salaries and on costs removed | The work itself, which still needs doing and now has no owner |
| Severance, as a one off | A second wave of leavers, from the teams that absorbed the load |
| Floor space and systems, sometimes | The institutional knowledge about why a process is shaped the way it is |
| A headcount number that reads well | Morale across the whole company, not just the part that was cut |
| Nothing else | Quality, which follows capacity strain with a lag of about a quarter |
| Nothing else | Churn, which follows quality with a further lag of about two quarters |
The bottom two rows are the ones that turn a cost saving into a revenue problem, and they arrive far enough after the decision that the connection is hard to see from inside.
The morale part, which is not sentimental
People treat the morale argument as the soft objection, raised by somebody who does not want to make hard choices. It is not soft. It is the mechanism by which a one time cost saving becomes a recurring cost.
A cut is observed by everybody, not just by the people cut. The ones who stay do two things. They work harder for a while, which looks like a productivity gain and is a loan. And a fraction of them start looking, quietly, and leave over the following two or three quarters.
Those leavers are not random. They are the ones with options, which means the better ones. Each of them costs money to replace, takes months to replace, and takes their knowledge on the way out. The shock to morale decays over time, which is the good news, and the attrition it triggered does not undo itself, which is not.
Before cutting
Three questions that are answerable in a week
None of this is an argument that departments should never be cut. Some of them should have been cut years ago, and the honest version of that conversation is worth having.
It is an argument that the cost line is a tenth of the information and the rest of it is obtainable if you go and ask.
- What arrives at this team that would have to go somewhere else, and who specifically would receive it
- What has not gone wrong in the last two years that this team is the reason for, which is a question you have to ask people rather than a report
- Which customers touch this team, how often, and what they would notice in the first quarter
The real failure is the attribution
The reason this keeps happening is not that people are stupid about costs. It is that the saving is attributed and the damage is not.
Whoever made the cut gets credited with a number, immediately, in a review they attend. The delivery slip in month five gets attributed to the delivery team. The quality complaint in month seven gets attributed to the plant. The two resignations in month eight get attributed to a competitive labour market. Every one of those attributions is locally reasonable and collectively wrong.
A model that runs the year forward with the department removed does one useful thing above all others: it puts the saving and the damage in the same document, with the months attached. Whether the numbers are exactly right matters much less than the fact that they are now in the same place, where somebody has to look at both.
Run the question instead of arguing about it
Build a twin from whatever record you have, and see the range rather than the one number somebody liked.