The promotion that comes with fewer seats

Last week I spent an hour reviewing a financial model I didn't build. One of the agents I run built it.

The maths were right. All of it, every row, checked to the penny.

The summary line at the top of the tab, the one line anyone opening the file reads first, pointed at the wrong cells. It displayed 2 numbers that had nothing to do with the answer.

You wouldn't catch that by skimming. You'd catch it by already knowing what the number was supposed to say.

That hour is a small example of something happening across a lot of professional work right now.

Let's dive in.

The phrase going round in AI adoption is “moving people from doers to reviewers”.

Accounting is a good place to observe this in, because the work is credentialed and someone has to vet it. Firms are running agents through multi-step workflows now. Document intake, data extraction, reconciliation, first-draft audit testing. The output arrives review-ready. The qualified professional moves across to judgement, interpretation, and makes decisions that carry professional liability.

A product director at Thomson Reuters described the review stage as the new bottleneck in the whole process. Firms are having to build fresh habits around checking what a system used as its source material and how it reached its answer.

It's presented as a promotion. And review work does sit above production work. Nobody can argue about that.

A reviewer covers more ground than a doer. One person can review the output of many. That is the arithmetic, and that is the point of the change we are seeing.

Which means the number of reviewers a function needs is smaller than the number of doers it replaced. Not slightly smaller.

The management layer has been moving this way for a while already. Live Data Technologies put manager headcount down 6.1% between May 2022 and May 2025. Executive roles down 4.9% across the same period.

Gartner's forecast for this year is that 20% of organisations will use AI to simplify their structures, potentially removing more than half of existing middle management positions.

So when your organisation tells you your role is moving toward oversight, 2 things are happening at the same time.

Your work is getting more senior in description.

Your layer is getting thinner in headcount.

Both are true. Only one of them goes in the promotional announcement.

Back to the model on my screen.

The reason I caught the error is that I'd built the assumptions myself, months earlier, by hand. I knew what that header was supposed to say because I'd worked out the answer the slow way first.

Take that away and the word for what I'm doing is ‘approving’.

Review is more concentrated work, and it runs entirely on judgement you built by doing the thing you're now reviewing. Supervising a process well is harder than performing it.

Which creates an awkward problem for organisations, and a more interesting one for you. The people best placed to review are the people who spent years doing. And the doing is what's being removed.

Here's why I'm telling you this in a career newsletter.

The reviewer seat is real and it's worth having. Take it if it's offered.

But it's still one seat, inside one organisation, and the number of those seats is falling in your layer as you’re reading this. If your entire plan is to be the reviewer where you currently work, you have a plan with very few possible outcomes.

The more useful question is whether your judgement is portable.

Judgement is the input being priced now. That's the whole logic of the change. And judgement has never had to be sold to a single buyer.

That's the reason I built Zelova. Working out what your judgement is worth outside one employer, and who else would buy it, is work most senior people only start once the decision has been taken for them.

Not a quick exercise. Worth starting this month rather than the month you need the answers fast.

1) Where is your judgement actually load-bearing?

Not your job description. The specific decisions that would go wrong if you weren't present. Most senior people can name 3 or 4 once they stop listing responsibilities and start listing decisions they’ve made.

2) Who else buys that judgement?

Same sector. Adjacent sector. Smaller companies who can't afford it full-time. A board seat. A fractional arrangement across 2 or 3 businesses. Write the list of options before deciding what you want or what's realistic.

3) Can you demonstrate it outside the organisation that currently vouches for you?

Internally, your judgement is backed by your title and years of people working with you.

Extenally, it's backed by what you can show a stranger in 20 minutes. Those are different assets, and only the latter travels with you.

4) How many reviewers does your function need in 2 years, and what makes you one of them?

You probably know the answer to the first half.

The reason to do this while employed is that you can't do it well while reacting.

Once the announcement has gone out you're answering these questions in a fortnight, under pressure, with a mortgage in the background. The answers are worse because they’re rushed.

I've observed and spokent to enough senior people going through it to be confident about that, and I've been through my own experience of it too.

Has your role been described to you as moving toward oversight in the last year or so? If it has, did anyone explain to you the bigger picture and what was happening to the headcount in your layer at the same time?

David