This edition was a bit in the making. It’s not an everything is sh*t edition. This week we throw a bit more meat on the bone. Many things I knew of (like everyone else) but didn’t know the details (like everyone else.) Also a hamfisted Bruce Lee reference.
You probably suspected it. Most people who spent enough time inside one of these companies eventually did: that quiet, nagging feeling that the actual product, the advertising, was almost incidental to whatever the real business was. That the work you were doing mattered less than you’d been told. That the meetings were long and the decks were elaborate and the clients were managed rather than served, and that something in the whole arrangement was slightly off in a way nobody would name out loud.
You weren’t wrong. You just didn’t have the sentence for it yet.
The holding company was built on the arbitrage of client ignorance.
If Leo Burnett (the man) were here today, he might change “client ignorance” to “client trust.”
That’s the business. Not creativity. Not strategy. Not the transformative power of great advertising, which is real, and which was always more likely to come from a twenty-person shop with a chip on its shoulder than from a publicly traded entity managing forty-seven agency brands across six continents. The business, the actual underlying financial engine of the thing, was the systematic monetization of what clients did not know.
They didn’t know what media cost. You did. The margin between those two numbers, for decades, was enormous and largely invisible and considered by everyone involved to be a perfectly reasonable arrangement. The 15% commission wasn’t a fee for expertise. It was a toll. A toll on not knowing. It persisted as long as it did because clients, many of them, preferred not to know. The not-knowing was comfortable. The lunches were good. The arrangement was stable.
And if they knew, they’d probably have to deal with it.
Agencies had every incentive to keep it that way.
-
Media rebates that weren’t disclosed.
-
Production markups that bore no relationship to actual production costs.
-
Research billed at rates that assumed the client had no idea what research actually cost.
The whole beautiful machine, examined line by line, looked considerably less beautiful.
The 2016 ANA transparency report told clients what insiders had known for years. The agencies thanked the ANA for their thoroughness and changed as little as possible.
Download it here or the image above. It’s free. Only 62 pages cover-to-cover with bonus block diagrams.
Procurement officers arrived in the 1990s and early 2000s with spreadsheets and bad news. They were not romantic figures. They did not win Cannes Lions. They did not give inspiring speeches about the power of storytelling at industry conferences. What they did was look at what something cost and compare it to what they were being charged for it, and the gap, when they found it, was not small.
Clients started building in-house. Why pay for your agency’s infrastructure, overhead, margin, and the attention of talent split across twelve other accounts when you could hire the talent directly? The talent, it turned out, was available. The tools were available. The knowledge was not, in fact, proprietary; it had just been presented that way, and presented well.
By the early 2020s, most major brands had some version of an internal creative capability. The work that stayed at agencies was the overflow, the specialized, the stuff that genuinely required outside perspective. It was still a lot of work. It was a fraction of what it had been.
The expertise was always somewhat abstract. The knowledge was not proprietary; it had just been presented that way, and presented well.
Google and Meta built self-serve dashboards specifically designed to make the agency unnecessary in the media-buying transaction. They did it for their own reasons. The effect on agencies was the same: the expertise that had made agencies indispensable had been packaged, automated, and sold directly to the client for a monthly subscription fee.
Two companies shake hands and call it a vision. The vision: being very large.
Faced with a model losing its foundation on every side, the holding companies made a decision.
They decided to get bigger.
This is the logic of the cornered. If you cannot compete on product, compete on leverage. Acquire more companies. Aggregate more spend. Get large enough that your bulk purchasing power gives clients rates they cannot replicate on their own. Stop being a service business and start being infrastructure. Stop selling talent and start selling access.
WPP bought companies the way other people buy lottery tickets: constantly, hopefully, with the creeping sense that the last one hadn’t quite worked out but the next one might. Publicis purchased Sapient and Epsilon and remade itself as a ‘data-driven transformation company,’ which is a phrase that means something different every time someone says it and has never quite meant what the investor deck implied. Dentsu expanded globally. All of them got larger. But better?
The Cannes Lions kept going to the independents and the in-house teams and the upstarts, because that is where the people who actually wanted to make things had gone.
The Omnicom-IPG merger, which closed in late 2025 and created the largest advertising holding company in history, is not a new idea. It is the old idea, made very large, on the theory that large enough is a strategy. The antitrust review, and this is the detail that tells you everything about the political moment we are in, focused primarily on whether the combined entity might withhold ad dollars from conservative media outlets. The structural conflicts, the competing client accounts under one roof, the media buying leverage: largely unaddressed. The deal went through.
It is not a growth story. It is not a creative vision. It is two companies who tried everything else deciding to try being each other.
The layoffs that followed were announced in the language all of these things are announced in: efficiencies, streamlining, positioning for growth, the evolving landscape. Thousands of people. The combined entity had significant overlap, as combined entities always do, and the overlap was resolved in the traditional manner. You, if you were in the overlap, were handed a box.
Here is what they will not say in the press release: you were not collateral damage. You were the proof.
Our positions existed because the model needed bodies:
Bodies to fill the hours on the SOW
Bodies to justify the fees
Bodies to staff the infrastructure that clients were paying for whether they needed it or not.
When AI collapsed the cost of producing the work, when clients pulled more in-house, when the platforms cut out the middleman, the bodies became the most visible line item. The expertise was always somewhat smoke and mirrors. The hours were always somewhat inflated.
When the gap between what was being billed and what the work actually required became too large to hide, the bodies took the hit.
The merger didn’t create this situation. It confirmed it. Two companies that had been slowly, quietly, structurally failing at the thing they were supposed to be for decided to “try something else” together at greater scale. More of everything. What could be better? The synergies are real; you can absolutely cut costs by eliminating duplicate functions, which is why the duplicate functions are being eliminated, which is why you are reading this.
The business was never what it said it was. Now you have the sentence for it.
The business was never what it said it was. That is not a small thing to have worked inside of, and it is not a small thing to have suspected and never quite had the language for. Now you have it.
The arbitrage of client dependency. That was the product. It had a good run.