A Four-Layer Battle

Your biggest threat is unlikely the competition next door. In marketing services, the real battle is between layers and the middle is being eaten from both ends.

Last May, I wrote about the $179.5 billion tollbooth.

The argument was simple: retail media, programmatic infrastructure, and the platforms sitting between brands and consumers had constructed the most expensive rent-extraction system in marketing history. Brands were paying to access their own customers. The tollbooth operators were getting rich. And most CMOs were funding the construction while debating which lane to use.

Just nine months later, the tollbooth is under siege.

I've spent the last several months analysing holding company strategy in some depth. Evaporate28. The Omnicom integration. The gap between what the press releases say and what the balance sheets reveal. That body of work matters, and I'll continue it. But there is a trap in holdco analysis that I want to name clearly: it can make the inter-agency competition look like the main event. WPP versus Publicis. Omnicom absorbing IPG. Market share moving between legacy players.

It isn't the main event.

The holding companies' most dangerous competitors are not each other. They never were. The real pressure is structural and it's coming from clients who are inhousing, platforms who are disintermediating, and a technology layer that is being squeezed into irrelevance from both directions simultaneously.

When I analyse WPP's decline or Omnicom's merger rationale, the underlying question is always the same one: how does any of these organisations survive a world in which the value chain they were built to manage is being redistributed by forces they didn't create and can't fully control?

The Trade Desk's Q4 2025 earnings, which landed on February 25th, make that question visible in a way that no holdco earnings report can. Because The Trade Desk sits in the layer that everyone is currently trying to eliminate.

Not from regulation. Not from client revolt. From the four-layer battle that the marketing industry struggles under.

Revenue grew 14% year-over-year to $847 million. On the surface: solid. Beneath the surface: a company in structural deceleration. Full-year 2024 growth was 26%. Full-year 2025 was 18%. Q1 2026 guidance: 10%. That's not a rough quarter. That's a trend line pointing in one direction.

The stated culprit was CPG and automotive spending (also very interesting for the impact on the agency layer, given the importance of these industries to the marketing-industrial complex). Together, these categories represent more than a quarter of The Trade Desk's $13.4 billion in gross platform spend. CEO Jeff Green was explicit on the earnings call: without those two category headwinds, TTD's growth rate would have been at least 5 percentage points higher. The macro forces (tariff uncertainty, consumer cost-of-living pressure, input cost volatility) are "bigger than us and even bigger than them."

That admission matters. The Trade Desk has no owned inventory. No first-party commerce data. No ability to inflate its own attribution numbers. When Green says CPG brands are pulling back on open-internet programmatic spend, he is the most reliable datapoint in the industry about where the money is going, and where it isn't. Meta posted 24% revenue growth in Q4. Google 14%. Amazon Ads 23%.

The open internet is losing the budget war to the walled gardens. In real time.

But that's only one half of what the TTD earnings revealed. The more instructive half is what Jeff Green said about agencies.

On the same earnings call, Green addressed the agencies pulling out of OpenPath, which is TTD's direct-to-publisher supply channel, that routes ad dollars to publishers at a 4.5% fee. The most transparent supply path in programmatic. And certain agencies are quietly exiting it.

Green's explanation deserves to be read carefully. At a moment, he said, when many agencies are "focused on principle-based buying" and "not doing as good of a job of representing their clients as they could," it's no surprise that a product designed to create supply chain efficiency "ruffles feathers and bothers people."

That is a public company CEO, on a recorded earnings call, telling investors that some agencies are structuring media buys to protect their own margins rather than client outcomes.

But here's the point Green was actually making, beneath the OpenPath defence: The Trade Desk is being squeezed from the layer below: agencies who benefit from opacity are routing around the most transparent option in the market, at exactly the same moment they are being squeezed from the layer above by platforms whose owned-inventory moats are compounding.

The vice is tightening from both directions.

The marketing value chain has four layers, each fighting to expand its share of the total spend:

Layer 1: Clients. Inhousing accelerates. First-party data infrastructure, in-house trading desks, direct publisher relationships. The goal: reduce dependence on every other layer. Every dollar they reinternalise is a dollar extracted from the layers below.

Layer 2: Agencies. Caught between a client layer that is building capacity above and a middle layer that is being disintermediated below. The strategic response from the holding companies has been to build proprietary platforms (like Epsilon/CoreAI at Publicis, Omni at the new Omnicom entity) that route spend through their own data infrastructure rather than through independent vendors. If your DSP is the agency's own platform, the independent DSP's take rate disappears into agency margin instead.

Layer 3: The Middle Layer. The Trade Desk. Adobe. Salesforce. The independent tech vendors who positioned themselves as neutral infrastructure between agencies and media owners. This is the layer currently being eaten from both ends. Platforms are growing 20%+ while the open internet decelerates. Agencies are building proprietary alternatives and exiting transparent supply paths. Clients are inhousing. Every layer above and below is in motion and the middle is the space everyone is trying to shrink.

Layer 4: Platforms and Media Owners. Meta, Google, Amazon, retail media networks. Growing. Consolidating. Automating. Meta Advantage+ and Google Performance Max are not just media buying tools, they are slowly making the entire decisioning layer above them optional. When a platform's AI can optimise a campaign better than an agency trading desk, the question of which middle-layer tech vendor you're using becomes moot.

Four layers. All moving simultaneously. All trying to capture more of the same value chain. This is a structural redistribution of who owns the economics of marketing.

Which brings us back to The Trade Desk's actual strategic question and the one that matters for every vendor occupying the middle layer.

Jeff Green's argument is that in an AI-powered, supply-abundant world, objectivity is the moat. The Trade Desk processes 20 million ad decisions every second. It owns no inventory. It has no channel conflict. And Kokai, its AI platform, now reaches nearly 100% of clients, claiming 26% better cost-per-acquisition and 58% better cost-per-unique reach versus its prior platform.

This is the "toll booth learning to drive" thesis: TTD is not a passive fee-extractor sitting on a road that already exists. It is actively building the decisioning infrastructure that makes the open internet more valuable than the walled gardens, if you believe that independent, objective AI outperforms captive, incentive-compromised AI at scale.

I find the argument intellectually serious. I also find the evidence mixed.

The Kokai performance numbers are vendor-reported. Not independently verified. That does not make them false. But it is the question every CMO should be asking before treating them as gospel.

The Trade Desk's Q4 earnings are not a company story. They are a structural signal.

When the most transparent platform in independent programmatic decelerates while every walled garden accelerates, when agencies exit transparent supply paths to protect margin, when CPG brands simultaneously cut spend and inhouse capability. These are not separate events. They are the same event. The middle layer is being compressed. The question is which parts of it survive by transforming, and which parts are simply too slow to notice what's happening.

I wrote in May that the toll booth was the problem. The update is more precise: the problem is that there are four toll booths now, stacked on top of each other, and every one of them is trying to become the last one standing.

Over the coming months I'll be mapping each layer in detail, from Adobe's orchestration play, the holdco data platform land-grab, the client inhousing ceiling, the platform endgame. The four-layer battle is the strategic frame for understanding every major move in this industry in 2026.

This is Act One.

Brandflow is written by Justin Billingsley, who has spent his career on all three sides of the industry's table: senior client, global agency leader, technology founder. First published 5 March 2026 in the Brandflow newsletter on LinkedIn.