They Costed The Role. Not The Person.

When I’m brought in after the fact, after the automation is live and something isn’t working the way it should, the first thing I ask is: what do we have to work with? Every time, the room moves toward the same things. The system architecture, the data, the governance documents, the codebase. People start pulling up diagrams. Someone mentions the platform vendor. Someone else opens a laptop.

That’s not what I’m asking about.

I’m asking about the people who know how the job used to be done. The ones who held the edge cases, the exceptions, the patterns that arrived twice a year and never needed to be written down because someone was always there to handle them. And when I ask that question, and the room goes quiet in a particular way, I already know part of the answer.

This is the mistake the fastest-moving organisations are making right now. Not a technology mistake. Not a strategy mistake. A sequencing mistake, with a specific shape. They identify what can be automated, they calculate the headcount saving, and they use that saving to fund the transformation. The logic is clean. The arithmetic works. And somewhere inside that sequence, before anyone has looked carefully enough at what the people actually carry, the people leave.

What they carry is not in any document. It accumulated over five years of contact with the work, specifically with the parts of the work that didn’t follow the rules. The exception that arrives with a shape the system has no answer for. The case that needs someone who has seen it before. That knowledge never needed to be written down because the person was always there. The role map doesn’t show it. The efficiency case certainly doesn’t show it. It shows up exactly once, when the system hits its edge and hands the problem to a human, and the human has less context than the system does. That is the moment the costing missed.

In the engagements where I get to shape the sequence from the start, this is what I protect. You run the new system alongside the people, not instead of them. You watch what the system proposes and what the people actually do, and somewhere in that gap, usually quietly, in a moment nobody anticipated, something becomes visible that was invisible before. A judgment call made in seconds. A pattern recognised before the flag fires. Knowledge that accumulated over years of contact with the edge cases, never written down because it never needed to be. When a team sees it for the first time, something shifts. The people who hold it understand their own value differently. They want to go further. The work gets better and they know why. That experience is one of the best parts of this work.

Then there are the conversations at conferences, or the ones that reach me through colleagues, where someone describes what they are doing and I listen and I say “oh wow” and they take it as enthusiasm. Sometimes I mean it that way. Sometimes I am looking at the shape of what they are building toward and I cannot say plainly what I see, because they did not ask, and because they are genuinely proud of moving fast, and they should be, it is hard to move that fast. If you are reading this and thinking about your own timeline, your own efficiency case, your own announcement already made, that is not a coincidence.

What I see is this. They are financing the transformation with the thing the transformation needs to work.

The people who left, the headcount saving that funded the investment, some of those people carried knowledge the system will need when it hits its edge. The system will handle the routine. It was always going to handle the routine. The eight percent, the fifteen percent, the high-volume straightforward work that was always going to go first. That part is fine. What is not fine is making the irreversible decision about the people before anyone has looked carefully at what sits next to the routine in the same role, invisible, uncosted.

The timeline comes from above the room. Three, four levels above the person who holds the knowledge, someone has decided the pace, and they have never been close enough to the work to know what is inside it. The room that might have slowed down doesn’t.

And so the sequence continues. The system runs. The metrics hold. The exceptions haven’t arrived yet in sufficient volume to show the gap.

What do we have to work with?