This Isn’t a Blip. It’s a Reset for Agencies.

The recent IPA agency census figures landed heavily for a reason. Creative and non-media agency headcount fell 14.3% in 2025, while total headcount across IPA member agencies fell 6.8%. Open roles were down sharply too, with vacancies across all seniority levels falling 40.8% year on year. (The Drum)

It is tempting to read numbers like that as a rough patch. A correction. A difficult year. Something that will bounce back once confidence returns.

I do not think that is what this is.

The Drum’s piece framed it as “a structural shift, not a blip,” and that feels much closer to the truth. This looks less like a temporary downturn and more like an industry being forced to confront business models, delivery structures and talent assumptions that have been under pressure for years. (The Drum)

The headline is job loss. The real story is model change.

The most important thing here is not just that jobs have gone. It is where the pressure is landing and what that suggests about the shape of agency work going forward.

Creative and non-media agencies appear to have taken the hardest hit, and that is significant. These are often the parts of the market most exposed to project volatility, scope creep, margin compression and the acceleration of executional work through automation. At the same time, advertised roles have fallen sharply, especially on the creative side, which suggests many businesses are not simply pausing hiring but actively redesigning what kinds of roles they need. (Campaign Live)

That points to something more structural than a cyclical slowdown.

AI is part of the story, but not the whole story

AI is clearly a factor. Coverage of the IPA findings linked the scale of job losses to the speed with which AI tools are changing production, workflow and cost expectations, particularly in execution-heavy roles. (The Guardian)

But blaming AI alone lets the industry off too easily.

A lot of agencies were already operating with fragile commercial models before generative AI became mainstream. Many were over-reliant on selling time, overexposed to constant pitching, and carrying delivery structures that made margin hard to protect. AI has accelerated that pressure, but in many cases it is revealing weaknesses that were already there rather than creating them from scratch. This is not just a technology story. It is an agency model story. (The Drum)

The tension is that demand has not disappeared

What makes this moment more complicated is that the market is not simply collapsing. The IPA’s Q1 2026 Bellwether Report, published today, said UK companies revised their marketing budgets upwards by a net balance of 7.3%, the highest level in almost two years. Events and PR were among the areas driving that uplift, and main media also returned to growth territory. (ipa.co.uk)

So this is not a neat story of brands no longer valuing marketing.

If anything, it looks more like brands still want marketing investment, but are becoming much more selective about what they buy, how they buy it and which parts of the value chain they believe should be paid for at legacy agency rates. In other words, money is still moving — just not necessarily toward the same structures, teams and service models agencies built themselves around a decade ago. That is an inference, but it is strongly supported by the gap between rising budget revisions and shrinking agency employment. (ipa.co.uk)

What agencies may need to confront now

For agency leaders, I think the uncomfortable question is not just how to ride this out. It is whether the business is still built for the version of demand that actually exists.

That includes some fairly fundamental questions. Are you still selling inputs when clients want outcomes? Are you carrying teams shaped around old production assumptions? Are you differentiating around capability lists that now feel interchangeable? Are you relying on volume and over-servicing to make the numbers work? And do you have a commercial model that holds up when clients expect more flexibility, more speed and more visible value? These are the kinds of structural issues the current employment figures seem to be exposing. (The Drum)

For some agencies, the answer will be repositioning. For others, it may be a delivery redesign, a sharper specialism, a different staffing mix or a more productised offer. But the broader point is that incremental tweaking probably will not be enough if the underlying model is the thing under strain.

The talent question worries me most

One of the most worrying signals in the wider reporting is what this means for the pipeline. The Guardian’s coverage of the IPA data highlighted a steep decline among younger workers and a sharp pullback in graduate and entry-level recruitment. (The Guardian)

That matters because industries do not stay healthy when they hollow out the bottom of the ladder.

If agencies cut junior opportunities too deeply, they may protect short-term cost bases while weakening the long-term talent system that creative industries rely on. The danger is not just fewer jobs today. It is a smaller, less diverse and less sustainable future workforce tomorrow. (The Guardian)

A reset does not have to mean decline

None of this means agencies are finished. It does mean the old assumptions are under serious pressure.

There will still be strong agencies in this next phase. Probably very strong ones. But they are likely to look different: leaner in some areas, more specialist in others, clearer on what they solve, more disciplined commercially and less dependent on labour-heavy models that confuse effort with value. That conclusion is not stated directly in the reports, but it is a reasonable inference from the mix of falling headcount, reduced vacancies and rising marketing budgets. (The Drum)

Final thought

The agency market has been talking for years about transformation, future-proofing and reinvention. What the latest IPA findings suggest is that the reset is no longer theoretical.

It is happening now.

The question for agencies is not whether the old model is under pressure. It plainly is. The more useful question is what they are building next — and whether it is designed for the market that exists now, rather than the one the industry got comfortable in. (The Drum)