Agencies cut writing to one job in thirty. The bill arrives at the next pitch.

In late 2025, writing roles were one in six agency vacancies. By July this year they were one in thirty. The Drum published the collapse on 9 September, and the figure underneath it is the one worth sitting with: across the same twelve months, all agency postings rose 24% while writing roles rose 10%.

Agencies did not stop needing writing. They stopped paying for the judgement about what is worth writing, and kept every obligation that judgement was there to serve.

The decoupling is the story, not the decline

A falling number in a falling market is a recession. A falling number in a recovering one is a decision.

That is what the data from research shop Agency by Agency, drawn from The Data City and Lightcast, actually shows. Writing roles were 17% of agency postings in the six months to January. They were 6% in the six months to July. In July alone they were 3%, the lowest point of the entire period since 2022. Tom Salmon puts it plainly: the agency job market is recovering, and writing roles are not participating in the recovery.

Set that against the backdrop and it gets sharper. Forrester forecast 15% agency job losses across 2026. IPA figures put creative-shop headcount reductions at 14%. Writing did not fall with the market. It fell through it.

What got cut was never the typing

The case for the cut is that drafting is now cheap, and it is. A competent model will produce a serviceable first version of most things an agency publishes, faster than a junior and at a fraction of the cost. Nobody serious disputes this.

The mistake is treating the draft as the job. The expensive part of writing was never producing sentences - it was deciding which argument the agency was prepared to put its name to.

That decision is made by people who have sat in the pitch, read the trade press every morning for a decade, and know which of the agency's three plausible positions is the one a client will actually pay for. It does not come out of a prompt, because the input it runs on is not in the prompt.

Where the bill arrives

Agencies are obliged to explain themselves in a small number of high-stakes places, and every one of them runs on written argument.

The pitch document, where the difference between shortlisted and appointed is usually a paragraph explaining why your read on the category is right. The trade byline, where Campaign and PRWeek will run an agency's opinion and will not run its summary. The AI answer, where the engines cite sources that state a position clearly enough to be quoted, and skip the ones that hedge. The new-business email that has to earn a reply from someone who gets forty.

None of those got cheaper when drafting did. They got harder, because everyone's drafts got better at the same time and the floor moved. When the median piece of agency writing is competent, competence stops being a differentiator and the only remaining one is having something to say.

The forecast is worse than the data

The same research project, run by The Drum with iResearch Services and launching in full at The Drum Live this month, asked global marketing leaders which roles will shrink in scope by 2030. Copywriters and content writers came top, at 42% - double the share who said the same of designers and creative producers. Only 25% expected the role to grow, the lowest vote of confidence given to any job on the list.

I think that forecast describes what marketing leaders intend to do rather than what they will find works. The agencies acting on it fastest are removing the capability that makes them legible to clients, editors and engines, at the exact moment being legible is the scarce thing.

Some of them will be right. Plenty of agency writing was always filler, and filler deserved to go. But the ones cutting to one in thirty are not making that distinction. They are cutting a headcount line, and finding out afterwards which of those people had been quietly deciding what the agency stood for.

That is a hard thing to notice you have lost, because nothing breaks on the day. It shows up a quarter later, in pitches that read like everybody else's.

If your agency's positions are getting harder to write down, that is the signal, and it is worth taking seriously before the next pitch rather than after. We work on exactly that.