I have spent roughly half my career working in agencies.
And I want to say something clearly before I write another word.
Some of the best, most creative, most visionary people I have ever worked with, I met inside agencies.
People who could see around corners. People who could take a confused brief and turn it into something that made a business move. People who could sit in a room with a client’s leadership team and reframe a problem so completely that the whole trajectory of a brand shifted.
I have enormous respect for what agencies can be at their best. And that is precisely why I think this conversation matters.
Because right now, the agency model is under more structural pressure than at any point in my career. And most of it has nothing to do with AI.
AI is not the cause. It is the accelerant.
The problems were already there. AI just made them impossible to ignore.
What I see happening
I am not going to pretend I have a bird’s-eye view of the entire agency sector. I do not. But between my own experience, the conversations I have had, and the research I have done, a picture has formed that I think is worth sharing honestly.
The picture is this.
The agency sector is not dying. But it is bifurcating. And the gap between agencies that are thriving and agencies that are quietly struggling is widening faster than most people realise.
At the top, you can see it in the holding company numbers. Publicis Groupe posted 5.6% organic growth and an industry-leading margin of over 18% in 2025. WPP, the sector’s largest player, saw revenue drop more than 5% on a like-for-like basis, operating margins fell by two points, and the company cut 7,000 jobs in a single year. Their share price hit a 16-year low before being relegated from the FTSE 100.
Omnicom completed its $13 billion acquisition of IPG, creating the world’s largest agency holding company. But the merger has already eliminated roughly 10,000 positions. Three historic creative networks, DDB, FCB and MullenLowe, have been shuttered.
Dentsu swung to a $2 billion loss and replaced its CEO.
These are not small corrections. This is structural reorganisation happening in real time.
And it is not just a holding company story. Gartner’s 2025 CMO Spend Survey found that 39% of CMOs are planning to cut agency budgets. Agency allocations now represent just 20.7% of total marketing spend, down from higher levels in recent years, as paid media continues to absorb a larger share. And 22% of CMOs said generative AI has already enabled them to reduce reliance on external agencies.
At the same time, 82% of ANA members now have an in-house agency. That is up from 58% in 2013. The World Federation of Advertisers reports 66% of multinationals are using some form of in-housing.
So the environment is clear. Budgets are flat. Agency share of those budgets is shrinking. Clients are building internal capability. And AI is compressing the execution layer that many agencies depend on for revenue.
But here is the thing.
I do not think any of this means agencies are finished. Not remotely.
I think it means the wrong kind of agency is finished. And I think the right kind of agency is about to become more valuable than ever.
The question is what separates the two.
The problem that predates AI
The temptation is to make this an AI story. AI is changing everything. AI is eating agencies. AI is the great disruptor.
And AI is genuinely important. I will come back to it.
But if I am honest about what I have observed, the deeper problem is older and less dramatic than that.
The problem is that a significant portion of the agency sector spent the last decade optimising for scalability at the expense of irreplaceability.
Let me unpack what I mean.
The 2010s were a boom era for agencies. Digital channels exploded. Search, social, programmatic, mobile, content, e-commerce. Every new platform created demand for new expertise, new teams, new budgets. Agencies could grow by simply participating in the expansion. Adding capabilities. Covering more channels. Hiring more people.
And many did. Aggressively.
But here is what that expansion rewarded. It rewarded breadth. It rewarded speed of adoption. It rewarded the ability to staff up, build platform partnerships, and manage multi-channel reporting.
What it did not require was a distinctive view of how marketing creates value.
You did not need a proprietary methodology. You did not need deep category expertise. You did not need a clear strategic position. You needed to be present, competent, and scalable.
And for a while, that was enough. Because when the market is growing fast enough, everyone looks like a winner.
But once the tailwind faded, something became visible.
A lot of agencies looked the same.
Same capabilities. Same language. Same “integrated, data-driven, full-service” positioning. Same slide decks with different logos in the corner.
And when agencies look the same, clients treat them like commodities. They compare on price. They compare on headcount. They run procurement-led reviews that reduce the entire relationship to hourly rates and FTE benchmarks.
The IPA’s recent research on agency pricing found that just 27% of agencies believe they are paid a fair price for their work. And 58% reported little to no progress in changing their commercial agreements.
That is not an AI problem. That is a differentiation problem that has been compounding for years.
Many agencies sell strategy but operationally sell motion
I want to say this carefully, because I know how it sounds. And I know from experience how hard agencies work.
But I think one of the most important tensions in the sector right now is the gap between what agencies say they sell and what they actually deliver.
Most agencies position themselves as strategic partners. The pitch decks say strategy. The credentials say thinking. The website says transformation.
But if you look at where the revenue actually comes from, and where the headcount actually sits, the picture is different.
For many agencies, the bulk of revenue comes from production. Content. Campaign assets. Channel management. Reporting. Media buying. Versioning. Adaptation. The machinery of execution.
Strategy is the thin layer at the front of the engagement. It wins the pitch. It frames the relationship. But it is not the core economic engine of the business.
I have seen this from the inside. I have lived it.
And the reason it matters so much now is that execution is exactly the layer that AI is compressing.
If 80% of your revenue comes from production and channel management, and AI can reduce the cost and time of that work by 30, 40, 50 percent, then your economic model is under direct threat. Not because the work disappears entirely, but because the value clients attach to it collapses.
Forrester initially predicted 7.5% of agency jobs would be automated by 2030. They have since revised that dramatically, now forecasting 15% could be eliminated in 2026 alone. That is not a gradual adjustment. That is a step change.
And eMarketer reported that 83% of US marketing leaders would reduce agency spending if they could fully automate content creation. 73% of teams that have already adopted AI agents have cut their agency content spending.
Those numbers should be uncomfortable for any agency leader reading this. Not because automation will replace everything, but because it will replace the thing many agencies are most dependent on.
The conveyor belt was built before AI arrived
There is a critique that gets thrown around, sometimes fairly, sometimes not, that a lot of agencies have become conveyor belts. High-volume output systems that produce campaigns, content, and assets on a production line, with limited strategic originality.
I think the truth is more nuanced than the label suggests. But directionally, I think the critique is fair for a meaningful portion of the market.
And I think the evidence supports it.
Peter Field’s research for the IPA, analysing nearly 600 case studies over 24 years, found that the advertising industry has shifted decisively toward short-term, activation-focused work at the expense of long-term brand building. Creatively awarded campaigns became less effective than at any point in the data, and were ultimately no more effective than non-awarded campaigns. The primary driver was short-termism.
The EACA found that 60% of winning creative concepts from pitches never get implemented. Agencies spend enormous resources on ideas that go nowhere.
And Campaign Brief’s research shows creative agencies take approximately 32 months of revenue to recoup pitch costs. The pitch process itself has become a loss-making ritual.
These are not signs of an industry optimising for strategic impact. These are signs of an industry optimising for throughput. For activity. For billable hours and utilisation rates.
AI did not create the conveyor belt. AI just switched it to high speed.
And when you speed up a system that was already producing undifferentiated output, you do not get more value. You get more noise.
So what actually separates the agencies that will thrive?
This is the part that matters most. Because the story so far sounds bleak, and I do not think it is.
I genuinely believe the best agencies are about to enter one of the most valuable periods in the history of the profession.
But only if they understand what the value is shifting toward.
Here is how I see it.
They have a worldview, not just a service menu
The single most important thing an agency can have right now is a clear, opinionated, evidence-backed view of how marketing creates value.
Not a tagline. Not a positioning line on a website. A genuine operating philosophy that shapes how they diagnose problems, how they frame strategies, how they allocate resources, and how they measure success.
I cannot overstate how rare this is.
Most agencies describe themselves in terms of what they do. We do creative. We do media. We do data. We do strategy. We do everything.
Very few describe themselves in terms of what they believe. What their view is on how brands grow. What they think matters most in a client’s commercial context. What they would argue for in a room even if the client disagreed.
And here is why it matters so much.
In a world where AI can handle execution, the only durable basis for agency value is the quality of the thinking that directs that execution. The judgment. The framing. The ability to see what others miss and say what others will not.
If an agency does not have a worldview, it has nothing for AI to amplify. It is just a faster, cheaper production function. And clients already have access to that.
They have genuine IP, not process theatre
This one is uncomfortable but necessary.
A lot of what agencies call “proprietary” is not. It is standard industry practice with a branded name on it. A process flow with a fancy acronym. A diagnostic framework that is really just a series of common-sense questions in a nice template.
That is not IP. That is theatre.
Genuine intellectual property means something a competitor cannot easily replicate. A unique dataset. A predictive model. A diagnostic tool with demonstrated commercial impact. A codified methodology that consistently produces better outcomes than the alternative.
Publicis’s investment in Epsilon and its CoreAI platform is the holding company version of this thesis. They now have an identity graph spanning 2.3 billion profiles across 109 countries. That is not a slide. That is a structural competitive advantage. And it is a significant part of why Publicis is outperforming everyone else.
At the independent level, you can see similar signals. Known has built Skeptic AI, trained on data from 1.8 million micro-campaigns, and now licenses it as a subscription to clients.
These are not typical agency offerings. They are products. They are assets. And they are very difficult to commoditise.
They sell decisions, not deliverables
I think this might be the most important shift of all.
As AI makes execution cheaper, the scarce resource is no longer the ability to produce work. It is the ability to decide what work should be produced, and why.
Which markets to enter. Which audiences to prioritise. Which creative territories to own. How to balance brand and activation. Where the budget should go. What to stop doing.
These are decisions. And they have enormous commercial value.
But they are not what most agencies currently sell. Most agencies sell deliverables. Campaign assets. Media plans. Content calendars. Reports.
The agencies that thrive next will be the ones that shift from selling output to selling leverage. The strategic clarity, creative quality, and analytical precision that makes a client’s marketing budget produce outsized returns.
The EACA study on CMO expectations found that 49% of CMOs rank trust as the most important agency attribute, and 41% rank deep business involvement above creativity. They are not asking for more deliverables. They are asking for partners who understand their business well enough to improve the quality of their decisions.
That is a fundamentally different value proposition. And it requires a fundamentally different kind of agency.
They speak the language of the boardroom
This one is personal for me, because I have been on both sides of this conversation.
The ability to connect marketing activity to business outcomes, in language that CFOs, boards, and procurement teams understand, is no longer a nice-to-have. It is a survival skill.
Digiday has documented how CFOs have moved upstream in marketing conversations, now attending agency reviews and demanding metrics like contribution margin, payback periods, and customer lifetime value.
Agencies that can only offer activity counts and vanity metrics are going to keep losing that conversation. Agencies that can articulate how their work drives revenue, margin, and market share, and can back it up with evidence, will earn a different kind of relationship with their clients.
Commercial literacy is not a specialism. It is the entry ticket to relevance.
They invest in people, not just platforms
I want to end this section with something I feel strongly about.
The talent conversation in agencies right now worries me.
Ninety-one percent of senior US agency leaders expect AI to reduce headcounts. Fifty-seven percent have already slowed or paused entry-level hiring. The traditional agency pyramid, where senior talent oversees large junior execution teams, is being inverted.
And I understand the economic logic. If AI can do the work that juniors used to do, why pay for juniors?
But one agency CEO described looking at her workforce plan for 2029 and 2030 and realising there would be a crisis. With no pipeline of junior talent developing through hands-on experience, there will be no mid-level or senior talent in five years. She is right.
The agencies that win long-term will not be the ones that cut deepest. They will be the ones that restructure most thoughtfully. Smaller teams, yes. More senior, yes. More AI-augmented, yes. But with deliberate investment in developing the next generation of integrative talent, people who combine strategic thinking, creative sensibility, data literacy, and commercial judgment.
Because that kind of person is going to be the most valuable professional in the marketing industry. And the agencies that develop them will have a compounding advantage that no AI tool can replicate.
The model that emerges
If I pull all of this together, the agency that thrives in the next era looks quite different from the one most of us grew up in.
It is more opinionated. It operates from a clear view of how marketing creates value, not a neutral service menu.
It is more systemised. It has codified methods, proprietary tools, and AI-embedded workflows that create genuine IP and improve over time.
It is more integrated. Strategy, creativity, media, and data are connected by people who genuinely span those domains, not just by organisational charts.
It is more talent-dense. Fewer people overall, but a higher concentration of senior, multi-skilled, commercially literate professionals.
It is more commercially articulate. It can connect its work to business outcomes in language that earns respect from the C-suite, not just the marketing department.
And it is more AI-leveraged. Not AI as a bolt-on efficiency tool, but AI as a fundamental operating layer that amplifies distinctive capabilities.
That is not a fantasy. There are agencies already moving in this direction. They are smaller, sharper, more deliberate, and more valuable per person than the models they are replacing.
A closing thought
I wrote at the start that some of the best people I have met in my career were at agencies. That is true. And it is why I care about this.
The agency model, at its best, does something remarkable. It brings together strategic thinkers, creative minds, analytical talent, and commercial judgment in a way that most organisations cannot build internally. It creates a concentration of capability that, when directed well, can transform a brand’s trajectory.
That capability has not disappeared. If anything, the need for it has never been greater. Marketing is harder. Attention is scarcer. Signals are weaker. The decisions are more complex. The stakes are higher.
But the old way of packaging and selling that capability, billing hours, producing volume, competing on breadth, avoiding a point of view, is running out of road.
The future does not belong to the fastest producer of generic output. It belongs to the agency that has something to say, knows how to prove it, and has built the systems and talent to deliver it at a level that clients cannot get anywhere else.
That is the shift.
From volume to value. From motion to meaning. From scalability to irreplaceability.
If you work in an agency, or you lead one, and you feel the ground shifting beneath you, I hope this is useful. Not as a warning. As a reframe.
Because agencies still have a future. A good one.
Just not this one.
