Institute for Human and Artificial Work‹ Essays
The Chief Workforce Officer · Essay 11 of 23

Headcount is the wrong number

Every workforce report an executive sees counts people. None counts productive capacity. That blindness is now expensive.

Iman Abrishamchi··5 min read

The most important number in any workforce report is headcount, and headcount has just stopped meaning what it meant. A department with forty people and twelve agents is not a department of forty. It is not a department of fifty-two either. It is something the current reports cannot describe, and the inability is not a detail.

Consider what headcount was for. It was a proxy for capacity: how much work the organisation could do. It was a proxy for cost, since people were the largest expense. And it was a proxy for responsibility, since every unit of capacity was a person who could be asked what happened. Agents break all three proxies at once. They add capacity without adding to the count. They shift cost from salaries to subscriptions, which sit in a different budget and are approved by different people. And they add capacity that has no one to ask.

The result is that executives are steering by an instrument that no longer measures the thing. A board that sees headcount flat and output up concludes the organisation has become more productive. It may have. It may also have deployed agents whose output nobody checks, displaced the junior roles it will need in five years, and shifted eight hours a week of unbudgeted verification onto staff who are now quietly overloaded. The report cannot distinguish these cases.

What would a better report contain? For each major task family, the share of output produced by people, by agents, and by the two together. The human hours spent verifying agent output, as a line of its own. The incident rate per thousand agent outputs. The number of agents in service, with a named accountable human against each. And the one forward-looking measure that matters most: how many people in the organisation could still do each automated task by hand, which is the organisation's insurance against the day the agent fails or the vendor changes it.

None of this is hard to collect. All of it is currently uncollected, because no one owns the question. HR counts people. IT counts systems. Finance counts money. The thing that needs counting, productive capacity across both kinds of input, falls between them, and so it is not counted, and so the board does not see it.

The Chief Workforce Officer owns that report. It is, in a sense, the role's most concrete deliverable: a quarterly statement of the organisation's whole capacity, human and artificial, presented beside the headcount the board already knows how to read, until the board learns to read the new one instead.

There is a historical parallel. Companies once reported only tangible assets, and the growing share of value in brands, software and know-how was invisible until accounting caught up with intangibles. The gap mattered because decisions were made on the visible numbers. The same is true now. An organisation that manages what it measures, and measures only its people, will manage its people and let its agents run. That is not a technology risk. It is an accounting choice, and it has an owner the moment the organisation decides it should.

The question this series asks. Should organisations create a Chief Workforce Officer, accountable for people and AI agents together, and what must be thought through before they do? Every essay takes one angle on it. Write to the Institute; every message is read and answered.