The divergence between WPP and Publicis is easy to see. Explaining it is harder.
WPP was still ahead on the closest available underlying-growth measures in 2021. Publicis moved ahead in 2022. From 2023, the separation became increasingly difficult to dismiss as quarterly noise. By 2025, one group was still growing while the other was contracting sharply. The first half of 2026 preserved the gap, even as WPP's second-quarter decline moderated.
That establishes a performance divergence. It does not establish a single cause.
1. The gap is real, but not timeless
The closest directional comparison is between WPP's like-for-like revenue less pass-through costs and Publicis's organic growth in net revenue.
The measures are conceptually close, but not identical. Publicis defines organic growth in net revenue after excluding acquisitions, disposals and currency effects. WPP calculates like-for-like performance at constant currency while adjusting comparison periods for acquisitions and disposals, and reports revenue less pass-through costs separately. They are useful for tracking direction across time, not interchangeable accounting KPIs.
In 2021, WPP grew 12.1% on its measure against Publicis's 10%. The ordering reversed in 2022: 6.9% for WPP against 10.1% for Publicis. By 2023, the gap had widened to 0.9% versus 6.3%. It widened further through 2024, then became especially large in 2025, when WPP declined 5.4% and Publicis grew 5.6%.
The divergence persisted into 2026. In the first half, WPP declined 4.7% while Publicis grew 4.7%. But WPP's Q2 decline moderated to 2.8%, against 4.8% growth at Publicis.
Profitability points in the same broad direction. WPP reported a 13.0% headline operating margin for 2025, down from 15.0% a year earlier, while Publicis reported an 18.2% operating margin. Their margin definitions are not perfectly standardised, so the exact gap should not be treated as one common industry measure. As secondary evidence, however, profitability reinforces the growth pattern.
The chronology matters. WPP was not always behind, and its latest quarter was less weak than its 2025 low point. What needs explaining is a crossover followed by a widening divergence—not a permanent difference between a winner and a loser.
2. The clearest recent contributor evidence is commercial
For the latest stage of the divergence, client and account dynamics provide some of the most direct evidence.
WPP's top 25 clients declined 4.1% like-for-like in 2025, including the effect of client assignment losses earlier in the year. WPP Media declined 5.9%. Those losses continued to affect the first half of 2026: revenue less pass-through costs among WPP's top 25 clients fell 6.3%, while WPP Media declined 5.4%. WPP has explicitly said prior client losses remained a drag on reported performance.
That does not mean the current trajectory is static. WPP's Q2 2026 decline was materially smaller than in Q1, even while prior client losses continued to weigh on reported performance.
The timing matters because account decisions and reported organic revenue do not move together instantaneously. A large loss can depress revenue after the pitch itself has disappeared from attention; a new account can take time to become meaningful in reported growth.
Publicis has been moving in the opposite commercial direction. Publicis estimates that its H1 2026 new-business performance had secured roughly 200 basis points of growth on a full-year basis. Publicis also says Comvergence's 2025 New Business Barometer ranked it first globally for new-business performance.
That is significant contributor evidence, but not a complete explanation. Public disclosures do not reveal comparable account profitability, onboarding economics or the precise revenue attached to individual wins and losses. More importantly, the WPP–Publicis crossover began before WPP's most recent major loss cycle.
Client dynamics therefore provide the clearest proximate contributor evidence for the recent gap. They do not explain the entire 2022–26 divergence.
3. Different business configurations, no clean causal test
The deeper explanation is harder to observe.
Publicis assembled much of its current business mix before the present AI boom. Its January 2024 strategy described a six-year transformation built partly around Sapient and Epsilon, a country-led organisation and a common operating backbone. Epsilon, acquired in 2019, expanded Publicis's data capabilities; in 2022, Publicis reported organic growth of 12% at Epsilon and 19% at Publicis Sapient.
That history makes the group's mix of data, connected media and transformation capabilities a plausible part of the explanation. It does not prove that those assets caused Publicis's overall outperformance.
Owning capabilities is not the same as integrating them effectively. Integration is not the same as commercial effectiveness. And none of those observations, on their own, establishes company-level causation.
WPP presents a related problem from the opposite direction. Its current management has diagnosed recent underperformance partly in organisational terms. In February 2026, CEO Cindy Rose attributed it to excessive complexity, the absence of an integrated operating model and inconsistent strategic execution. Elevate28 is intended to simplify the group around four operating units and a more integrated client proposition.
That diagnosis is relevant because it comes from the management team now restructuring WPP around it. It remains management's diagnosis, not independent proof of why the performance gap emerged.
Nor is WPP's weakness confined to one business. In H1 2026, WPP Media declined 5.4% and WPP Creative 4.9%, alongside weakness across several regions and client sectors.
The narrower conclusion survives. Publicis's longer-established business mix is a plausible background contributor. WPP's organisational and execution difficulties are another. Public evidence cannot tell us how much either matters—or whether the more important distinction is what each company owns, how its businesses work together, how it goes to market, or how clients respond.
4. The easy explanations fail
AI becomes useful here—not as the governing explanation, but as a test of it.
Publicis formally put AI at the centre of its next strategy in January 2024. By then, its relative outperformance was already established: it had moved ahead of WPP in 2022 and widened the gap substantially in 2023. Publicis itself presented AI as the next stage of a transformation already built around Sapient, Epsilon and organisational changes made over previous years.
WPP, meanwhile, cannot reasonably be characterised as a company that simply failed to invest in AI. In January 2024 it set out plans for around £250 million of annual cash investment in proprietary technology supporting its AI and data strategy. By June 2025, WPP said roughly 85% of client-facing staff were using WPP Open. During the same half-year, revenue less pass-through costs declined 4.3% like-for-like.
Investment and deployment did not mechanically translate into stronger group performance.
Publicis's own 2026 reporting complicates the simple technology story further. It groups 87% of net revenue into a company-defined “AI-powered marketing services” category, which grew 6.5% organically in Q2. Connected Media grew at a high-single-digit rate. But Publicis's separate Technology practice, representing 13% of net revenue, declined at a mid-single-digit rate as clients delayed large transformation projects.
The 87% figure describes Publicis's reporting taxonomy. It does not show that AI generated 87% of group revenue.
None of this establishes that AI is commercially irrelevant. It establishes something narrower: investment, deployment and economic impact are different things, and public evidence cannot isolate AI's incremental contribution to either company's revenue or margins. Chronology does not support AI as the origin of the divergence.
Acquisitions fail as a singular explanation too. The relevant organic and like-for-like measures substantially remove the direct scope effects of acquisitions and disposals. Historical acquisitions can still reshape what a company is capable of selling and how its business mix develops; they do not mechanically manufacture the organic-growth gap.
The same applies to market conditions. They matter, but peer performance is mixed. Havas was growing organically in early 2026; Dentsu's group result was positive but weak, with contraction in some major regions. Those companies use different measures and have different mixes, but the variation is enough to weaken a pure market-cycle explanation.
AI, M&A and macro conditions all belong in the story. None carries it alone.
5. What the comparison can—and cannot—tell us
Both companies are organising around combinations of capabilities that extend beyond traditional agency categories. Publicis has spent years building across data, media and transformation, while WPP's current restructuring is explicitly pushing towards a simpler, more integrated proposition across media, creative, production and enterprise solutions.
It is therefore reasonable to ask whether changing client demand increasingly rewards organisations that can combine these capabilities effectively.
The comparison does not prove it.
That boundary is important. Publicis and WPP have different portfolios, acquisition histories, client exposures, management decisions and recent account trajectories. Peer results are mixed. Public evidence also cannot separate the value of owning particular capabilities from the quality of integrating or commercialising them.
The useful structural possibility is therefore narrower than “Publicis has the winning model.” Client demand may be increasing the value of combinations that cut across traditional agency categories, while company-specific execution determines who captures that value. Those two explanations can coexist.
What the WPP–Publicis comparison shows is that a structural hypothesis deserves investigation. It does not validate a universal agency model.
6. A layered answer, not causal arithmetic
The divergence is more certain than its explanation.
For the recent period, client and account dynamics provide the strongest proximate contributor evidence. Behind them sit plausible background contributors: Publicis's longer-established mix across data, connected media and transformation, and organisational or execution problems that WPP's own management is now trying to address.
Historical acquisitions and market conditions matter but cannot explain the gap alone. AI may matter to current and future economics, but the evidence does not support it as the origin or primary explanation for a divergence that predates Publicis's 2024 AI strategy.
Public evidence cannot responsibly assign causal percentages to accounts, business mix, integration, organisation, geography or macro conditions. Nor can it yet establish whether the gap will persist.
That is the clearest conclusion available: WPP and Publicis have materially diverged. The evidence supports several plausible contributors, but not a single cause—and not yet a universal theory of how agency groups win.