WPP’s plans to cut an additional 1,000 jobs by the end of this year have raised further concerns about the impact of AI on the agency model. The holding company has already cut around 11,000 jobs since the start of 2025, and in February announced a multi-year strategic plan to make the group “a simpler, lower-cost, AI-enabled business.”
Meanwhile other holdcos are further along in their AI transformation journeys. Havas unveiled its four-year AI investment strategy back in 2024, and this summer CEO Yannick Bolloré said the French holdco has “automated everything that we can automate.” He added that in Cannes this year, he was no longer being questioned on the survival of agencies.
While agency groups do appear to be adapting rather than retreating in the age of AI, these transformations come with increasing pressures on agency staff. A report from ad tech firm Basis earlier this year found that 70 percent of agency professionals say their jobs are more difficult today than they were two years ago, while 54 percent reported strained client relationships, driven by higher expectations, compressed timelines, and increased scrutiny.
On a knife-edge
Three anonymous senior agency sources that VideoWeek spoke to confirmed that increased client expectations were indeed adding pressure, but suggested that morale was mixed, depending on where agencies are along their AI transformations – and the level of awareness among staff of the pressures facing their businesses.
One senior buyer from a holdco described the feeling of being on a “knife-edge” among “those in the know” at the agency. “There are probably a load of people who are blissfully ignorant, not necessarily being aware of how tight this year is, and how much is riding on certain things,” they added. “But knowing what I know, there’s an implicit understanding that we’ve got to work very hard to make sure we hit targets for this year – and the targets are that much harder to hit.”
That said, morale is “pretty decent” at their particular agency, which has so far avoided measures such as salary freezes. Hiring however has slowed in the senior buyer’s team to the point that there has not been a new entrant (as in new to the industry) in over a year, according to the source, where in previous years the team would expect between four to eight new starters every year. “I do not think we will have a single new-to-the-industry starter in my team this year,” they said.
Another senior agency source from a different holdco said they are still hiring, and there are “a lot of open roles” at the agency group. But they also noted the dangers of short-sightedness in making layoffs, only for more work to be required further down the line.
“I think collectively we need to try as much as possible in a very difficult climate to not be short-sighted,” they said. “Think about any layoffs that are made, as soon as that agency wins business, you’re going to need to hire again. You’re going to have to cover that, and you’re in this constant pattern, and it causes real pain in the agency.”
Doing more with less
A third senior leader at another holdco said agency staff are being expected “to do more with less”, in terms of both media budgets and headcount. They highlighted the gulf between the efficiencies expected to be realised through AI, and the reality of what is currently acheivable, particularly when resources are being diminished.
“I think it’s fair to say that net-net, headcounts are shrinking overall, but there’s still the same, if not more, expectation,” they said. “And I think that expectation comes from agency leaders top-down, and it comes from a lot of that expectation chat. But it’s also from our clients. They’re coming at us with expectations about speed, about doing more with less, and I think that pressure is just bubbling and bubbling away, because we’re not at that reality yet.”
But that pressure from clients can also be traced to their own teams being stretched and downsized in the current economic climate. The senior buyer observed that a few years ago, agency teams would deal with sometimes up to eight people in a marketing department. “It’s now normally two or three,” they said. “I think clients are somewhat feeling the pinch.”
However, the three people said that client relationships are no more strained than usual, and are “still generally strong” in the eyes of the second senior agency source. The threat of clients in-housing thanks to advancements in AI capabilities is also concerning but largely overstated, according to the sources, and in-housing can even create opportunities for agencies providing consultative services during transition periods, often culminating in a hybrid model.
Short-term pain for long-term gain
Meanwhile technology changes can also be a source of friction as agencies overhaul their operating models. The Basis report highlighted inefficient processes and siloed or disconnected systems as agencies’ top challenges, with teams now managing more than twice the number of tools than they did two years ago. But the senior agency source called the adoption of new tools “short-term pain for long-term gain”, since the holdcos need to replace their legacy systems, some of which have been around for more than 20 years.
While changes of this nature are long overdue, the senior buyer cited the challenges of keeping up with the needs of a rapidly evolving market when it comes to building new tools. “You spend six months thinking about what the tool will be, and maybe another six months or more developing that tool,” they commented. “You get to a point where it’s usable, but then it needs to be something else, because the industry has changed in that period of time. The life cycle of these things is shorter than it used to be.”
For the senior agency source however, AI is “already genuinely saving us a lot of time”, particularly when agentic AI is used for fielding similar questions from different clients. They used the example of providing a viewing or market update for a client, when pooling data and past expertise from activation, strategy and planning teams, combined with an added layer specific to the client, removes “a huge chunk of work” from that response.
With this in mind, the third senior agency leader noted there is “a very big distinction between how AI can inform a response and being the response.” Where the likes of research and data analysis are “clear wins” for the tech, they said that for tasks such as writing emails, using AI risks human value and accountability, as well as degredation in AI models trained on the outputs of other models.
“I see the danger in using AI agents to read and respond with my identity, and I won’t be using it in that way myself,” they said. “I am the only thing of value left in this chain, and I would never use an AI assistant to write an email for me. Because that’s when you get cannibalisation in your models, and when that happens, the quality goes down.”
Follow VideoWeek on LinkedIn.
