They Fired You, Rehired You, and Never Sent the Bill

I think in spreadsheets. Twenty-five years of running people functions and operations does that to you: every decision about human beings eventually becomes a row, and the row has a cost column. I am not apologising for it. It is how a company stays alive, and I have said before that when things get hard the company comes first, because that is where the paycheck comes from.

So when the numbers on AI restructuring started arriving this year, I read them the way I read everything. As rows. And it took me longer than it should have to notice what was missing.

Here is what we know. Careerminds surveyed 600 HR leaders who had run layoffs in the past year. A third of them had already rehired for between a quarter and half of the roles they cut. Another third had brought back more than half. Over half rehired within six months. Only 8.4% said the restructuring delivered exactly what was promised. Forrester puts the share of companies that cut for AI and now regret it at 55%, and Orgvue, working independently, landed on the same figure and added the arithmetic: about $1.27 spent for every $1 saved. Commonwealth Bank of Australia cut 45 customer service roles and reversed within weeks. Klarna's chief executive, after going all in on AI service, now says there will always be a human if you want one.

Meanwhile the next wave is already being planned. A survey of 750 chief financial officers found 44% intending AI-related cuts this year, around 502,000 roles against 55,000 last year, while the same study finds a wide gap between the productivity gains they perceive and the ones they can actually show.

That is a lot of measurement. Read the list again and notice what every single number has in common.

They all measure the company.

The cost of cutting. The cost of rehiring. The skills lost, from the company's point of view. The regret, the company's regret. The productivity that did not arrive, on the company's ledger. The round trip has been costed to the cent from the inside of the building, and I understand why, because I would have commissioned exactly those studies. That is the question a company knows how to ask.

Now think about the person who made the round trip.

Someone spent fifteen or twenty years becoming an expert at something. Then a leadership team, on the strength of a demo and a spreadsheet, decided AI could do it, and let them go. Within six months, more than half of those companies were hiring again for roles they had cut, and a third had already refilled between a quarter and half of them. Sometimes to the same people, I would bet, and sometimes to people just like them. Different desk, possibly a lower band, definitely a shorter memory.

Nobody measured what that did to them.

Nobody measured the six months, or what happens to a person's sense of what they are worth when the people who knew them best decided a language model was a fair swap and then quietly changed their minds without ever saying so out loud, or what it does to walk back into a building that priced you at zero and now needs you at $1.27. There is no survey for that. There is no column.

Hear me out, because the argument here is harder than "companies should feel bad." If you only measure one side of a transaction, you will keep making the same transaction, because your data will keep telling you it was almost worth it. The 8.4% is a company number, and even the company number says this mostly failed. Imagine what the other number looks like.

I have sat in the rooms where these decisions are made, and I can tell you the person is absent from the analysis for a simple reason: there is no row for them. Nobody is cruel. The people around that table are, mostly, decent and tired and under pressure to show a number by the end of the quarter, and the model in front of them was built to answer the company's question because that is the only question anyone knew how to write down. We have a cost for severance and a cost for backfill and a cost for lost productivity, and none of those is the cost of telling someone that what they spent two decades becoming is a line item that can be reversed in a quarter. That cost lands entirely on them, and it does not show up in any report I have ever been sent.

That's the thing. The public conversation has decided this is a skills problem: reskill, adapt, stay relevant. Every number in this piece says the companies were the ones who got the skills question wrong. They cut expertise they turned out to need, and they are paying to get it back. Obsolete is the wrong word for what happened to these people. They were mispriced by someone else's belief, and then repriced again when the belief did not survive contact with the work.

If that has happened to you, or is about to, I want you to hold onto one thing. The regret is real and it has been measured. 55% of them regret it. You were not wrong about what you were worth. They were, and now they have the data to prove it.

What I cannot tell you is what the round trip costs. Nobody has asked. I am starting to think that the fact nobody has asked is the finding.