Which Golden Girl is WPP most like?

Full analysis

Publicis took PepsiCo's global media account away from Omnicom without a formal pitch, after roughly 25 to 30 years with Omnicom. That one move forced Publicis out of the Coca-Cola global review (you cannot hold both), left Omnicom and Dentsu fighting over Coke North America, and gave the Digiday editors a Golden Girls bit to hang a year of holdco positioning on.

Here's what's actually being decided for ad-tech operators: which holdco data and buying stacks win the next budget cycle. Michael Burgi says a raft of accounts goes into review in 2026. When a $25-billion-a-year relationship moves without a pitch, the preferred-vendor list underneath it moves too. That is the story for anyone selling into these agencies. The deadline is the cycle itself: Coke resolves soon, and the 2026 review wave follows.

Hard to undo? For the agencies, very. A client that assigns business without a pitch is not coming back next year. For vendors, the contract churn is the undo-able part, and it will be messy.

The Market Analyst — Publicis won without pitching, and that is what the whole game has come down to. Arthur Sadoun bought Epsilon in 2019 and is closing LiveRamp (a data-connectivity platform that stitches consumer data across systems) by year-end. Twenty-one straight quarters of growth, roughly 17.5 pence of profit per pound of revenue, a company record. Clients are now buying the data layer rather than the creative. Omnicom just became the largest group by revenue after swallowing IPG, but revenue rank is a vanity metric when Pepsi walks out the door weeks after close. For a general reader: the agency that owns the customer data gets handed the account; the one that owns the most offices has to fight for it.

The Skeptic — Steelman the bear case on Publicis and it gets thin fast. That is the problem. The non-pitch win is real, but PepsiCo is one data point, and conflict rules did half the work on Coke. The genuinely shaky claim is Omnicom as "largest in the world by revenue." Seb Joseph states it as settled; the fact check calls it unverified and currency-dependent. A CMO told Joseph the Dentsu strategy was great but the talent depth behind the pitch team was insufficient, so they walked. That is the signal worth tracking: winning the room and keeping the account are two different things, and that gap sinks more agency wins than any data stack.

The Operator — Tuesday morning, you are a DSP or measurement vendor with contracts routed through legacy IPG entities. Omnicom retired the DDB brand and booked over half a billion in divestitures. Your contract is now sitting inside an org chart that is still being redrawn. The thing that breaks first is renewal: nobody on the new combined side owns your relationship, and the person who signed it may be gone. The S4 Capital saved reading drives that point home. Monks is running AI agents unsupervised for 36 to 48 hours doing production work. That is a different cost structure attacking the same deliverables, and it does not need your seat license.

The Customer / End User — The customer here is the CMO, and the Coke CMO reportedly wants creator marketing as a buying criterion. That tells you where budget pressure lands next: holdcos that can execute influencer and creator work at scale, measured, win. The deeper customer anxiety is principal media, where agencies buy inventory on their own account and resell it to clients. WPP is contending with the Richard Foster suit and allegations it kept vendor rebates instead of passing them back. For a general reader: clients are starting to ask whether their agency is a buyer working for them or a reseller marking them up. That question reprices trust across all four.

The CFO — Follow the margin. Publicis holds 17.5 pence per pound because the data layer carries higher margin than media buying, which is a pass-through business. Omnicom's half-billion in divestitures is one-time cash, with no durable earnings behind it. Dentsu is flat overall: Japan up 5%, the rest going backwards, so 40% of the business is carrying the other 60%. The real cost nobody is pricing is integration drag. Omnicom-IPG talent alignment is the open question, and talent that leaves takes accounts with it. Pepsi may be the first account out the door.

Where they disagree. The Market Analyst says the data layer is now structural and durable. The Skeptic says one non-pitch win plus conflict rules is a headline, not a thesis. That is the real fight. The second split: the Operator and CFO both see the Omnicom-IPG integration as the opening for churn and defection, while the positioning narrative treats "largest by revenue" as a win. Size and strength are pointing in opposite directions here. Third, the Customer section raises the thing the whole episode skates past: principal media could reprice the holdco model faster than any account review, and the panel treats it as a WPP footnote rather than an industry-wide trust problem.

What it hinges on. Three things. One, whether Publicis's data advantage actually produces more non-pitch wins or whether PepsiCo stays a one-off. Two, whether Omnicom can integrate IPG without bleeding accounts and the vendors attached to them. Three, whether principal media scrutiny stays a WPP story or spreads. The council leans toward Publicis's advantage being real but oversold in the telling, and toward the Omnicom integration being the live risk for vendors right now. Before anyone rebuilds a go-to-market plan around holdco hierarchy, verify the revenue-rank claim against actual full-year 2025 numbers, and watch which accounts move in the 2026 review wave rather than trusting the Golden Girls seeding.

Prediction: Publicis Groupe will report full-year 2026 organic revenue growth higher than Omnicom's, when both report full-year 2026 results by March 2027.

Confidence: Medium — The data-layer margin gap is real, but integration timing could surprise either way.

Why: Publicis has posted 21 straight quarters of growth on the strength of Epsilon and the closing LiveRamp deal, which let it win accounts like PepsiCo without pitching, and data-driven work carries higher margin than Omnicom's media-buying pass-through, so its growth compounds while Omnicom's leans on one-time divestitures. Omnicom is absorbing IPG, the largest agency takeover ever, and the Pepsi loss weeks after close shows integration periods bleed accounts and the talent attached to them, which drags organic growth. The opposite outcome, Omnicom outgrowing Publicis, would require the IPG integration to add net new business faster than it loses legacy accounts in its first full year, which almost never happens in the first year after a merger this size. The comparison is Omnicom specifically because it is the only one of the four that just became larger than Publicis by revenue, so organic growth is the cleaner test of which model actually works.

Revisit by 2027-03-31: We're right if Publicis reports higher full-year 2026 organic growth than Omnicom. We're wrong if Omnicom matches or beats Publicis on full-year 2026 organic growth.