Let me answer the question everyone was actually asking, which was not "what happened."

It was "wait, can they do that?"

Here is the thing worth saying plainly, for anyone who does not spend their life inside agency remuneration. Nothing about this is normal. Holding companies lose pitches. They get shortlisted and cut. They decline to participate before a process starts, quietly, with a phone call. What they do not do, ever, is sit through pitch meetings on the largest media review of the year and then stand up and leave the table mid-process. There is no recent precedent for it at this scale. It is not a thing that happens.

It happened Wednesday.

Now the order of events, because the order is the whole story.

In June, Coca-Cola put its global media, data science and technology business into review, everything outside North America, Japan and Korea. MediaSense ran the process. Two holding companies were invited, WPP and Publicis. Pitch meetings took place in the last two weeks of August. Decisions were expected in the autumn.

Then, Wednesday afternoon, Publicis was named PepsiCo's lead global media partner with no competitive pitch at all, taking a $1.7 billion account off Omnicom after more than twenty years. Within hours, Publicis was out of the Coca-Cola process. Publicis declined to comment. Coca-Cola declined to comment. WPP declined to comment. Omnicom declined to comment. MediaSense did not respond. Five companies, one afternoon, nothing to say.

Coca-Cola now has a review with one participant in it, and that participant is the incumbent it put under review. That is not a shortlist. That is a hostage situation with better catering.

Here is what has not been reported, and it reframes everything downstream.

Multiple people familiar with the process tell ADOTAT that WPP's global presentation went badly, and that this was obvious immediately afterward to people on the agency side and to people on the client side. WPP had made substantial organizational changes ahead of the review, on the expectation that Coca-Cola was the next domino after a punishing eighteen months. The changes did not save the room.

WPP disputes the broader characterization of this account. Asked directly about its economics, the company told ADOTAT that the Coca-Cola business makes money and that describing it as a loss-maker is not accurate. WPP has also pointed to COMvergence figures showing WPP Media retention at 16 percent against Omnicom's 17 percent, and third of six on retained billings, responding to earlier ADOTAT reporting.

So hold two facts beside each other. The incumbent may have already lost the room. And then the only other bidder left the building anyway.

Which is the part nobody has explained, and the explanation everybody reached for falls apart in about five minutes.

The conflict story writes itself, which is exactly why you should distrust it. Publicis won the world's other global beverage account. Publicis therefore could not credibly pitch this one. Publicis therefore left.

Clean, tidy, and it requires nobody at Publicis to have made a decision at all. It has the texture of an accident, which is the most flattering possible framing for every party involved, Coca-Cola very much included.

Conflicts this size are not discovered. They are priced.

PepsiCo did not materialize on Publicis's desk Wednesday morning. A capabilities review with no pitch takes weeks minimum and realistically months, because the client is evaluating an operating model rather than a deck. Publicis was sitting in Coca-Cola pitch meetings across that same stretch. Somebody senior knew both processes were live, understood exactly what winning one would do to the other, and let both run to the end.

So the question is not why Publicis had to leave. It is why Publicis took the win that made leaving necessary.

The lazy answer is that PepsiCo is bigger. It is not. Coca-Cola's reviewed business has been put anywhere from $1.44 billion to $7.5 billion depending on who is counting and what they are counting, and the company's global media spend has been reported around $2.6 billion. On billings alone, Publicis chose the smaller prize and handed back the larger one.

Which means it did not choose on billings.

Publicis already holds Coca-Cola in the United States and Canada, won from WPP in early 2025. It was bidding for the rest of the planet. Winning would have handed it the entire global account and the entire operating burden attached, on terms substantially inherited rather than designed. Coca-Cola built that architecture with WPP from 2021, around a bespoke unit called Open X integrating creative, media, social, production, data and technology into one construct.

Whoever wins that business does not get a blank sheet. They get somebody else's cathedral and a mandate to keep the roof on.

PepsiCo offered the opposite. No pitch, no legacy construct, an operating model designed from scratch, 200-plus markets to install it across under a "One PepsiCo" banner spanning strategy, planning and technology.

That is the difference between running somebody else's system and installing your own, and in holding company economics it is not a rounding error. One is a fee. The other is a platform. Platforms are where all that data, identity, commerce and AI machinery a holdco spent five years and several billion dollars acquiring stops being a balance sheet item and starts being revenue.

ADOTAT has been told that PepsiCo made the appointment conditional on Publicis giving up Coca-Cola's North American media business, with timing unresolved and possibly tied to next year's review cycle. The account was given in confidence and described as already settled. ADOTAT put the question directly to both PepsiCo and Publicis. Neither has answered.

If it holds, it does not change who decided. It prices what they decided.

Nobody dragged Publicis into a room with PepsiCo. A holding company handed a bill that size either pays it or walks, and paying it is the decision. The condition supplies the receipt. An agency does not surrender a client it is profiting from to win a smaller one. It surrenders a client whose economics it has already concluded are not worth defending.

It also answers, in advance, the question ADOTAT put to Publicis this week about whether it expects to keep Coca-Cola in North America. On this account, Publicis already knows the answer, and it is no.

Here is what the trade press will not write, because writing it requires believing a $4 billion account can be worth less than a $1.7 billion one.

Billings are not revenue. Revenue is not margin. The distance between those three numbers is where the entire economics of a holding company lives, and on a global account this size that distance is a canyon.

A global media account carries a coordination layer that shows up in headcount and never in the fee. Governance. Local market adaptation across dozens of operating units. Data and technology support. Procurement liaison. Reporting cadences. Compliance with a client approval structure that turns a two-week decision into a two-month one and staffs the difference. None of it generates an invoice line. All of it generates payroll.

WPP's Open X operation has been reported as involving roughly 5,000 people globally. Allow every ambiguity in what "staffed by" means on a networked team and it is still an extraordinary number to park beside a billings figure. A headline that big next to a headcount that big is the silhouette of an account that is glorious to hold and murder to profit from.

Which side of that line an account falls on comes down to a handful of contractual and behavioral mechanisms. ADOTAT put each to people with direct experience of the Coca-Cola relationship and to independent remuneration specialists.

Scope density. Whether the statement of work carries task-level specificity or reads as broad output commitments the agency staffs at its own risk. A scope that says "always-on planning support across markets" is a scope priced by guessing, and every wrong guess is absorbed rather than billed. A person with direct experience of the Coca-Cola media scopes describes them as written without density around the individual asks, which parks the risk of under-scoping squarely on the agency.

Principal media. The ANA's 2024 report on the acceleration of principal media makes the mechanism plain. When an agency buys inventory as principal and resells it, that resale is real agency margin, which is exactly why the report demands explicit client agreement, disclosure and auditability. The corollary matters more here than the caution. If an advertiser forecloses principal buying, it forecloses one of the last places a media agency creates margin rather than collects a fee. On this account, principal is described as effectively unavailable as a general rule, with flex arrangements surfacing in biweekly reporting for approval in advance, which removes the discretion and the economics in one motion.

Client-owned commercial relationships. Coca-Cola runs its own joint business plans and holds its own relationships with bottlers, retailers and platforms. Every JBP the client owns is a negotiation the agency is not in and a set of economics it cannot touch or count toward plan. The client gets the better terms. The agency gets the execution and a thank-you note.

Capability internalization. The sharpest of the five. Coca-Cola has been publicly explicit for years that it is building internal marketing capability, and has framed Studio X as a mechanism for making its marketing faster, more effective and cheaper. That is the company's own stated strategy, not an allegation. What is not public is what that strategy feels like from the other side of the table, which is described as solutions presented, priced, declined, then built internally, after which the agency is faulted for not having the capability it just proposed.

Volume and adjacency. Media alone, without creative and production beside it, has a much harder time clearing profitability, because the coordination burden is fixed while the revenue base is a slice. Any agency taking international media on its own inherits the cost of the whole relationship and the revenue of part of it.

Five questions, all pointing the same direction. And the direction they point explains a decision that otherwise has to be explained as clumsiness.

The useful thing about this argument is how much of it the company has published itself.

Coca-Cola describes a networked marketing model joining product, digital, live and retail experiences. It emphasizes global standardization with local activation. It has framed Studio X as a route to faster and cheaper marketing. It operates through operating units built for local execution, multiplying the stakeholders any global partner must satisfy. It is running an agentic AI and marketing technology agenda that the June review documents themselves described as a move away from traditional media planning.

None of that is hidden and none of it is sinister. It is a large advertiser doing what large advertisers now do, which is treat marketing capability as an asset to own rather than a service to rent.

But there is a consequence nobody says from a podium. An advertiser building internal capability is, from the agency's side, a client whose scope loses value every year while the cost of servicing it does not. The work that goes inside is the work with margin in it. The work that stays is coordination. That is a business that gets structurally worse across the life of a contract, and any agency staring at a renewal knows precisely where on that curve it is standing.

That is the thing Publicis declined to sign up for globally.

The temptation is to flip this and declare PepsiCo the good client. Too easy, and probably wrong in the way that matters.

PepsiCo is a sophisticated advertiser that has been building internal commerce and marketing technology capability for years, in places well ahead of the market. Sophistication is not automatically good for agency margins. A client who knows exactly what it wants can be more demanding, more specific and more capable of doing the work itself, all of which cut the same way as Coca-Cola.

The difference is structural rather than temperamental. Publicis is not inheriting a PepsiCo operating model, it is building one. That is a transformation mandate rather than a media assignment, and transformation gets paid for as a project instead of absorbed as scope. Whether that holds depends entirely on how the contract is written, which nobody outside the room knows yet.

PepsiCo becomes a good account only if the sophistication arrives attached to a commercial model that pays for specialist work. If it does not, Publicis has swapped one difficult global beverage account for another and thrown in North America as a tip.

Worth noticing: PepsiCo did not run a shootout. It ran a capabilities review, and the business went to the holding company whose data and technology stack the client wanted underneath its own systems.

That is close to the position WPP has spent two years trying to claim. WPP acquired InfoSum and has positioned its clean room and data infrastructure, latterly as Open Intelligence, as a neutral facilitation layer. The difference between being the only clean room in town and being the plumbing under somebody else's house is the difference between a marketing claim and a business. One wins accounts without a pitch. The other wins panel slots at Cannes.

If the argument for a holding company in 2026 is that it owns the infrastructure, the test is whether clients pick the infrastructure on its own terms. This week one of them did, and it was not WPP's.

WPP is now, by default, the last agency standing in a process designed to produce alternatives to WPP.

That is the strangest kind of victory. It holds global creative and PR, which were never in the review. It has the infrastructure, the institutional knowledge, the lowest transition cost and several thousand people already pointed at the business. It also has a client that just spent three months publicly shopping the account and now has no competitive tension left to negotiate against.

The position that produces is worse than it looks. An incumbent that survives because everyone else left does not get to price like an incumbent that won. It prices like the only option, which sounds like leverage and is the opposite, because the client knows the difference and will remember which one happened.

Steve Boehler, founding partner of Mercer Island Group, which advises marketers on agency selection, sees two next steps and rates them unevenly. The likely path, he told ADOTAT, is that Coca-Cola reboots the review with a wider field than WPP alone. The unlikely one is that it simply hands the business to the incumbent.

Weigh that, because rebooting is the expensive option and he still calls it probable.

Reconstitute WPP. Lowest risk, fastest, preserves the knowledge and the existing infrastructure. Also looks like a default rather than a decision, and resolves nothing about why the review happened.

Reopen the field. Restores competitive tension, costs another six months during which the incumbent runs the business under a cloud, and requires Coca-Cola to explain publicly why round one produced nothing.

Expand Publicis. They have North America and the assets Coca-Cola says it wants. The PepsiCo conflict makes that close to impossible globally, and if the condition described above is real, impossible domestically too.

Split it regionally. Reduces single-holdco dependence and sacrifices exactly the global integration Coca-Cola spent five years assembling.

Take it inside. Consistent with everything the company has said about Studio X. It does not eliminate the cost of media, data, measurement, talent and global coordination. It relocates that cost onto Coca-Cola's P&L, which is a decision with consequences, not a saving.

One correction while we are here, because the wires have been muddling it all day. There is no public basis for saying Omnicom participated in the 2026 Coca-Cola review. Omnicom is central to the PepsiCo story as the displaced incumbent, which is almost certainly where the confusion started. The Coca-Cola review, as reported, was WPP against Publicis.

A caution that applies to every story published this week, this one included.

The figures in circulation are not measuring the same thing. Coca-Cola's total global media spend, the portion under review, the estimated value of the review, and the company's total advertising expense are four different numbers, and they have been swapped for each other freely across the trades. ADOTAT corrected its own figure once already, revising the reviewed total to $1.44 billion after COMvergence confirmed South Korea at $65 million and Korea's exclusion from scope. A request is outstanding with COMvergence for a defensible 2025 global billings figure and will be published with attribution when it lands.

The industry habit of quoting headline billings as though they were revenue is precisely the habit that makes it impossible to understand why anyone would walk away from one.

Three things will tell you whether this reading is right.

Whether Coca-Cola reopens the review or quietly concludes it. Reopening means it understood it no longer had a process. A quiet conclusion means it will take the incumbent and call that the plan all along.

Whether Publicis holds Coca-Cola North America through the next cycle. If it does, the conflict was manageable and this was opportunism. If it does not, and Publicis seems untroubled, that is the arithmetic showing its face.

And whether the PepsiCo arrangement is structured as a paid transformation mandate or as a media assignment with a services layer stapled on. That one contractual question decides whether Wednesday was strategy or just a larger version of the same trap.