Agency Headcount Trends After the AI Shift

One agency workforce survey found that 8% of agencies had reduced staff directly because of AI in the previous year, while a separate filing showed a major agency group cutting its workforce by more than 9,000 people in one year while also reporting lower revenue, client spending cuts and account losses. The numbers look similar on a spreadsheet. The reasons do not.

That distinction matters when looking at agency headcount trends. Headcount is falling in parts of the industry, but AI is only one part of the story. Client budgets, changing scopes, account losses and work moving in-house were already putting pressure on agency teams before AI became a major explanation for workforce reductions.

Key Takeaways

• Agency employment fell, but the decline was much sharper in creative and non-media agencies than in media.
• Revenue pressure and client spending cuts were already reducing agency capacity before AI became a major workforce explanation.
• AI is changing repetitive work, but it has not removed the need for strategy, media, data, technology and senior client decisions.
• Younger employees are seeing some of the sharpest declines, raising a bigger question about how agencies train future talent.
• Clients need to know where the work went when an agency reduces the people assigned to an account.

What do the numbers actually show?

The headline is a 6.8% decline, but that number hides a much bigger split. IPA member agencies employed 24,963 people on 1 September 2025, down from 26,787 a year earlier. Creative and other non-media employment fell 14.3%, while media employment increased 2.4%.

WPP shows the pressure at holding-company scale. Its year-end workforce fell from 108,044 in 2024 to 98,655 in 2025, a reduction of 9,389 people. Revenue less pass-through costs also fell 5.4% on a like-for-like basis.

The important point is not that every agency is shrinking. It is that the mix of agency work is changing. A workforce can fall overall while some functions continue to add people.

What was falling before AI arrived?

Client scopes were already getting tighter:

The commercial pressure came before the AI explanation. WPP’s 2025 results show the connection clearly. Revenue less pass-through costs fell 5.4% on a like-for-like basis, while the company reported client assignment losses and reduced spending across most client sectors.

That changes how a redundancy should be read. If a client removes part of a scope of work, an agency can need fewer people even when no new technology has entered the process.

The agency economics were changing too:

Agency remuneration determines how much work an agency can support with a given client relationship. Retainers, project fees, hourly charges, commissions and other models put different pressures on staffing.

The pressure becomes harder when clients expect more output without increasing the budget. The agency can cut the scope, change the team, automate part of the work or accept lower margins. Headcount is the result of those decisions, not always the original cause.

Some work moved closer to the client:

In housing is another part of the story. When clients take activities into internal teams, the work has not necessarily disappeared. The agency simply loses some of the people-hours attached to it.

Principal media adds another layer. When an agency acts as a principal and purchases media inventory before reselling it to a client, the economics are different from a traditional agency acting only on the client’s behalf. The point is simple: where the work sits affects how many agency people are needed to deliver it.

What AI genuinely changed?

AI made some work cheaper and faster:

The clearest AI effect is not that entire agency jobs disappeared overnight. It is that some tasks now take less manual effort. Versioning, adaptation, first-draft copy, basic research, some reporting and parts of production can all be accelerated by AI tools.

The latest industry census puts the change in context. Some 88.3% of agencies said AI was having a considerable impact on how they work. But only 8% said they had reduced their workforce in the previous 12 months directly because of AI.

That is a big difference. AI is changing how agencies work faster than it is explaining why agencies are cutting people.

What still needs people:

A faster production process does not remove the need for someone to decide what should be produced. Strategy still requires judgement. Senior client leadership still requires decisions and accountability. Media still needs planning and measurement. Data and technology work still needs people who understand what the numbers mean and what action should follow.

The employment split makes this visible. Media agency employment increased 2.4% while creative and other non-media employment fell 14.3%. Technology can change the amount of work one person can handle. It does not automatically remove responsibility for the result.

Why the cuts are hitting a narrow band:

The younger workforce is showing a much sharper decline than the overall agency workforce. Employees aged 25 and under fell 19.2% in the latest census, while total employment fell 6.8%. Vacancies also dropped from 1,149 to 680.

That does not prove AI is replacing junior employees. Junior roles are exposed to commercial pressure for another reason: they often sit closest to repetitive execution work and are also the lowest-cost layer of the agency structure.

So the same role can face pressure from two directions at once: tighter budgets and better tools.

The Roles that are growing

Media is already showing a different pattern

There is one part of the workforce where the latest numbers move in the opposite direction. Media agency employment increased 2.4% in 2025, from 12,012 to 12,304 employees, while creative and other non-media employment fell 14.3%.

That does not mean every media role is growing. It does show that the workforce shift is not a simple reduction across the entire agency business.

Data and technology are becoming more central:

The agency model is becoming more dependent on people who can connect creative, media, data and technology.

That does not mean every agency is hiring large teams of engineers. It means the value of a person increasingly depends on what they can do with the tools around them, not just how many manual tasks they can complete.

The agency staffing model is changing:

The old staffing model relied on layers of people moving work from a brief to execution. A newer model can put fewer people around the client while giving each person more tools and a wider scope.

That creates a different kind of agency team. The question is no longer simply how many people sit on an account. It is what each person is responsible for and how much work technology now handles around them.

What this does to the pyramid

The junior layer is getting thinner:

The sharpest workforce decline is happening at the bottom. IPA member agencies had 2,936 employees aged 25 and under in 2025, down from 3,632 in 2024. Graduate and apprenticeship participation also fell, with 43.4% of responding agencies reporting those programmes compared with 56% a year earlier.

That makes the agency layoffs in 2026 story bigger than the people leaving today. Fewer junior employees entering the industry means fewer people moving into experienced roles later.

The training problem comes next:

Junior work was never only about getting tasks done. It was also how people learned. A junior planner learns by handling smaller pieces of a plan. A junior account person learns by sitting in meetings and taking responsibility for increasingly important work.

A junior creative learns by producing, receiving feedback and trying again. If AI removes too much of that first layer, agencies need another way to create the same learning path.

The next five years will be about the mix:

The evidence does not point to every agency employee disappearing. It points to a different workforce mix.

There will be more pressure on roles built around repeatable execution. At the same time, agencies still need people who can make decisions, manage clients, interpret data, use technology and take responsibility for outcomes.

That means headcount could continue to fall without the agency business becoming smaller in every capability.

What it means for a client

A smaller agency team is not automatically a problem. If the scope has become smaller or technology has removed low-value manual work, fewer people may be enough. The problem comes when the team gets smaller but the client still expects the same amount of work.

That is when clients need to ask where the work went. Use this five-question account check:

1. Who is doing the work now? Ask for the actual people and roles, not just job titles.
2. What work changed? Separate work that disappeared from work that moved to another team or technology.
3. What did AI actually replace? Ask for the specific process that changed.
4. Did the scope or remuneration change? A smaller team should be considered alongside the work and agency remuneration being provided.
5. Who owns the senior decisions? Make sure strategy, measurement and accountability still have clear owners.

Clients should also understand whether principal media is part of the relationship and how it is handled contractually.

The issue is not whether an agency uses the model. The issue is whether the client understands what it means for the account.

The Read

The agency workforce is not simply being replaced by AI. The numbers point to something more complicated. Client budgets, account losses, changing scopes and different commercial models were already putting pressure on headcount. AI then arrived with a second effect: it made some types of work faster, cheaper and easier to standardise.

That combination is likely to keep changing the agency pyramid. The biggest question is not how many agency jobs AI will remove. It is which work will still require a person, and what happens to the people who used to learn by doing the work AI now handles? For agency leaders, the answer starts with rebuilding the training path around the new workflow. For clients, it starts with one simple question: If the team gets smaller, where did the work go?