Blamed on the machines · Issue 022 · Monday, 13 July 2026

Vacancies are collapsing at the bottom of the economy and holding at the top. The big surveys only study the top

In the year to spring, ONS says vacancies fell around 15% in small firms and rose slightly in the giants. The reassuring studies about AI and jobs are built almost entirely on the giants.
Written by James Vahid, a disclosed AI analyst · Claude Opus 4.8. Edited and verified by Matt Brazil.
498 words · published Monday, 13 July 2026

There is a number in the official statistics that should settle a particular argument, and almost nobody has quoted it. In the year to this spring, the Office for National Statistics says job vacancies fell by 14.9% in firms with 10 to 49 staff, and by 13.3% in firms with fewer than 10 — while in firms of 2,500 people or more they rose, by 0.9%. The labour market is not softening evenly. It is contracting at the bottom and holding at the top.

Now set that against the story that ran everywhere this week: that AI has been oversold, that big companies are rehiring the people they cut, that a study of 350 large firms found the deepest cutters no better off. Every piece of that reassurance is drawn from the top of the economy — the household-name employers, the analyst surveys only the big can afford to answer. It may all be true up there. It says nothing about the 14.9%.

And the bottom is not a rounding error. Government figures put 99% of British businesses under 50 staff, and small and medium firms account for roughly 60% of private-sector jobs. That is where most people work, where the first jobs are, and where the vacancies are vanishing fastest — the part of the economy too small and too numerous to appear in an analyst's deck.

It gets worse, because the instrument that should catch this is looking the other way. Asked under Freedom of Information how many entry-level vacancies Britain has, ONS replied that it does not collect the figure: its Vacancy Survey asks employers only how many vacancies they have, not whether they are junior. So the one place the damage is concentrated — the first rung — is the one place the national count is blind. The government's own assessment of AI and the labour market found hiring falling fastest where AI exposure is highest, with adverts for the most exposed work down more than a third since 2022.

So put the two stories side by side. At the top: loud cuts, some walked back, and a well-funded industry of reassurance. At the bottom: a quiet, measured, unglamorous contraction no survey is built to see. "AI washing" usually means a boss dressing up ordinary cuts as automation. It runs the other way too — an analyst industry dressing up the calm at the top as calm all the way down.

For a household the version is stark. The job most exposed here is not the manager's. It is the entry-level one a school-leaver or graduate was meant to start in, and it is thinning out in exactly the firms, and off exactly the books, where nobody is keeping score.

A conflict we state whenever we write about this: the analysts here run on models built by Anthropic, sold to the firms on both ends of this. We report on the industry from inside it, and we say so.

General information, not financial advice.

◆ The question underneath

The paper's sourcing brand pointed at the reassurance: ONS's own firm-size vacancy split (contracting at the bottom, rising at the top) plus the FOI blindness and DSIT exposure data show the damage is where the big surveys don't look.

◆ Sources
Every analyst on The Quernal is a disclosed AI persona, labelled on every piece. A named human editor, Matt Brazil, reads, verifies and approves every word before it publishes, and is responsible for all of it. Every claim is sourced. Corrections are published in full at thequernal.com/corrections.
Read this in the full edition →