A British energy company can tell you its customer satisfaction score to the point. It accounts for 1,300 people leaving in one number

If the state has stopped asking what kind of work is going, it is worth looking at what employers publish about it instead. The most useful example this year is not a bad one. It is an ordinary one.

Centrica published interim results for the six months to June on 23 July. Somewhere in them is the reduction of roughly 1,300 roles, about fourteen per cent of its customer operations workforce.

Set against that, what the same document says about customers.

Average contact per customer down twenty per cent year on year. Digital self-serve at around ninety per cent, up two percentage points. Complaints per UK home energy customer down to 3.7 from 4.1. Net promoter score for UK home energy supply up to 37 from 33. Operating costs excluding bad debt and depreciation down three per cent. Transformation investment of ninety-two million in the half, seventy-five through earnings and seventeen capital, against a total cost to achieve of around six hundred million, targeting five hundred million of underlying cost reduction by the end of the decade.

That is a company that knows precisely what it is doing and can prove it.

Now the workforce. Group direct headcount, 21,529, down from 21,881. And one measure of the people inside it: colleague engagement, 76, down from 79.

That is the whole account. One number about how the staff feel, three points lower than last year, in a document that reports customer sentiment to the decimal.

None of which means the company is hiding anything. The figures above are published and more detailed than many. Centrica is not obliged to report role changes by occupation, and neither is anyone else. Its chief executive has said publicly that the reductions follow changing customer behaviour rather than any technology programme, and our own arithmetic supports him: contact volumes fell by around a fifth against a headcount reduction of around a seventh. The demand for the work fell faster than the people did. That is a sufficient explanation and it does not need a machine in it.

The point is about the shape of the instrument, not the conduct of the firm.

A set of interim results is built to tell investors whether a business is working. Customer measurement does that, so it is granular. Workforce measurement does not, so it is one line. Nobody designed it to conceal anything. It was designed to answer a different question.

Which leaves a gap that nothing fills. The company reports to shareholders and reports what shareholders need. The state, until December, asked what kind of work was going and now does not. Neither instrument was built to tell you what happened to 1,300 people, and between them they no longer do.

There is a version of this that gets fixed cheaply. If a listed company reports headcount by segment, it could report role reductions by occupational category in the same table. Some already publish more. It would cost a paragraph.

The reason to want it is not to catch anybody out. It is that in five years somebody will ask which occupations went first in Britain, and the honest answer, on the current arrangements, will be that nobody wrote it down.