WPP vs the Field: Who Is Winning the AI Advertising Race

WPP, Publicis, IPG, Omnicom, and Dentsu are all racing to embed AI into their core advertising and media offerings. The differences between them are not cosmetic. The strategic bets each holding group has made reflect genuinely different theories about where AI creates value in advertising, and those choices will have commercial consequences for clients over the next three to five years.

This is a comparison of where the major holding groups actually stand, what their AI infrastructure looks like in practice, and where the gaps between the marketing and the reality are most visible.

I spent the better part of two decades inside agency structures, including a period growing iProspect from around 20 people to over 100 and taking it from loss-making to a top-five performance agency. One thing I learned in that environment is that the gap between what agencies say they can do and what they can actually deliver for clients is often wider than anyone admits in a pitch room. That instinct shapes how I read the AI claims coming out of the holding groups right now.

What Is WPP Actually Building With AI?

WPP’s flagship AI initiative is WPP Open, a proprietary operating system designed to connect the group’s agencies, tools, data, and production capabilities into a single platform. The pitch is that a client brief can flow through WPP Open and emerge as a coordinated campaign, with AI handling everything from audience modelling to creative production to media allocation.

The ambition is real. WPP has invested heavily in partnerships with Nvidia, Adobe, and Google Cloud, and the group has been explicit that WPP Open is intended to be a competitive moat rather than a bolt-on feature. CEO Mark Read has framed it as the infrastructure layer that will differentiate WPP from both other holding groups and from technology companies trying to disintermediate agencies entirely.

The honest question is whether the integration works as described at the client level. Large holding groups have a structural tendency to announce platforms and then deliver them unevenly across agencies and geographies. The holding group model is inherently federated, which creates friction between the ambition of a unified AI platform and the reality of dozens of semi-autonomous agency brands with their own tooling, cultures, and client relationships.

WPP’s creative AI capabilities are genuinely interesting. The partnership with Stability AI and investment in generative creative tools means WPP can produce and iterate creative assets at a scale that was not possible two years ago. For clients with high-volume production needs across multiple markets, that is a meaningful capability. For clients who need one exceptional campaign idea, it is less obviously relevant.

How Does Publicis Groupe’s AI Position Compare?

Publicis has a structural advantage that the group does not always get enough credit for: Epsilon. The 2019 acquisition of Epsilon gave Publicis access to one of the largest first-party data platforms in the industry, covering hundreds of millions of consumer profiles with genuine purchase and behavioural data. When Publicis talks about AI, it is talking about AI trained on and applied to real audience data, not synthetic proxies.

The group’s CoreAI initiative, announced in early 2024, is built on top of this Epsilon foundation. The claim is that Publicis can connect audience data, media planning, content personalisation, and measurement into a single AI-driven workflow. If that connection is as tight as described, it represents a genuinely differentiated capability, because most AI in advertising is still operating with impoverished data inputs.

Publicis has also been more aggressive than WPP in quantifying the productivity impact of its AI investments. The group has publicly attributed significant headcount efficiency gains to AI tools, which is either a sign of genuine operational transformation or a way of managing margin expectations, depending on how cynical you want to be. Probably some of both.

The risk for Publicis is that Epsilon’s data advantage erodes as the broader industry gets better at building first-party data infrastructure. Privacy regulation and the decline of third-party cookies have pushed every major advertiser toward first-party data strategies. Publicis has a head start, but the gap will narrow.

Where Does Omnicom’s Omni Platform Stand?

Omnicom’s AI story is centred on Omni, a data and planning platform that has been in development since 2018. That longevity matters. While WPP and others have been scrambling to build AI platforms in response to the generative AI wave, Omnicom has been iterating on Omni for several years. The platform covers audience identification, media planning, content personalisation, and performance measurement, and it is genuinely embedded in how Omnicom’s media agencies work rather than being a parallel system.

The pending merger with IPG, announced in late 2023, adds a significant data layer through Acxiom, which IPG acquired in 2018. If the merger completes and the data integration works as planned, the combined entity would have a data and AI infrastructure that rivals Publicis Epsilon in scale and arguably exceeds it in certain verticals.

Omnicom has also been thoughtful about where AI actually improves media outcomes versus where it creates operational efficiency. The group’s investment in AI-driven audience modelling and programmatic optimisation is grounded in media planning logic rather than technology enthusiasm, which I find more credible than holding groups that lead with the AI story and follow with the use case.

What Is IPG’s AI Strategy Before the Omnicom Merger?

IPG’s AI position is complicated by the merger situation. Before the Omnicom announcement, IPG was building its AI capabilities around Acxiom’s data infrastructure and a series of agency-level tools across McCann, FCB, and Mediabrands. The group had invested in AI-powered creative tools and was developing an AI governance framework that was more explicit than most competitors about responsible AI use.

The Mediabrands intelligence platform, which combines Acxiom data with programmatic capabilities, is the most commercially coherent part of IPG’s AI story. Media is where AI has the clearest and most measurable impact on client outcomes, and Mediabrands has been building toward that for several years.

The merger uncertainty has created a talent and client retention challenge that is probably more damaging to IPG’s near-term AI development than any technology gap. The best people in AI and data tend to have options, and holding group mergers create exactly the kind of organisational uncertainty that accelerates departures.

How Is Dentsu Approaching AI in Advertising?

Dentsu’s AI narrative is built around its dcx (Dentsu Customer Experience) and dentsu.AI initiatives, with a particular emphasis on the Japanese group’s strength in data and technology consulting. Dentsu has historically been stronger in Asia-Pacific markets and has used its technology consulting capabilities to build AI solutions that are more integrated with client business systems than the typical agency AI offer.

The group’s investment in AI-powered creative production and media optimisation is real, but Dentsu’s global footprint is less even than the Western holding groups, which creates execution risk when delivering AI capabilities across markets with very different digital maturity levels.

Dentsu has also been more willing than its competitors to talk about AI as a threat to traditional agency revenue models, which is either honest or strategically useful, depending on how you read it. The group has been explicit about the need to rebuild its commercial model around AI-driven efficiency rather than defending the status quo.

If you want to go deeper on how AI tools are being evaluated across the marketing industry more broadly, the AI Marketing hub at The Marketing Juice covers the practical dimensions that holding group press releases tend to skip.

What Are the AI-Native Competitors That Holding Groups Should Be Worried About?

The comparison between holding groups is only half the picture. The more structurally interesting competition is coming from AI-native companies that do not carry the overhead of a federated holding group model.

On the media side, companies like Quantcast, The Trade Desk, and a growing number of AI-first DSPs are offering planning and buying capabilities that are genuinely AI-native rather than AI-augmented. The Trade Desk’s Kokai platform, for example, is built around AI-driven audience modelling in a way that legacy agency trading desks are struggling to match.

On the creative side, the tools available to independent agencies and in-house teams have improved dramatically. AI copywriting and creative tools that were novelties eighteen months ago are now genuinely production-capable. The holding groups’ creative AI advantage is real but narrowing, because the underlying models are increasingly available to everyone.

I remember when I was at lastminute.com, running paid search campaigns that were generating six figures of revenue within a day of launch. The tools were relatively simple by today’s standards, but the speed of feedback and iteration was significant for how we thought about campaign management. What AI is doing now is compressing that feedback loop further, and the organisations that will benefit most are not necessarily the largest ones. They are the ones with the clearest connection between AI output and commercial decision-making.

For a broader view of how AI tools are being evaluated in practice, Moz’s analysis of AI SEO tools and Ahrefs’ AI tools webinar series are worth reviewing, not because SEO is the whole picture, but because the evaluation frameworks being developed in that space are more rigorous than most of what passes for AI assessment in the advertising world.

Where Is the Gap Between AI Marketing Claims and Actual Client Value?

Every holding group is now leading with AI in its new business pitches. That creates a specific kind of problem: when everyone claims the same capability, the claim stops being differentiating and starts being table stakes. The actual differentiation is in the details that pitch decks rarely cover.

The questions worth asking are specific. What data is the AI actually trained on? Is it proprietary client data, licensed third-party data, or public data that every competitor has access to? How is the AI output connected to campaign measurement? Can you trace an AI-generated audience model to a business outcome, or does the chain of accountability break somewhere in the middle?

I judged the Effie Awards for several years, which gave me an unusual window into how agencies and clients frame marketing effectiveness. The entries that won were almost never the ones with the most sophisticated technology. They were the ones with the clearest line from insight to idea to outcome. AI changes the speed and scale at which you can execute, but it does not change that fundamental logic.

The holding groups that are most credible on AI are the ones that can answer the measurement question. Publicis’s Epsilon infrastructure gives it a genuine advantage here because the data exists to connect AI-driven decisions to actual consumer behaviour. WPP’s creative AI capabilities are impressive, but impressive creative production at scale is only valuable if it connects to campaigns that move business metrics.

Semrush’s research on generative AI adoption among marketers is useful context here. Adoption is high, but the gap between using AI tools and integrating them into workflows that produce measurable outcomes remains significant. That gap is exactly where the holding groups’ AI promises are most likely to disappoint clients who do not ask the right questions upfront.

How Should Marketers Evaluate Holding Group AI Capabilities?

If you are a senior marketer evaluating holding group partners, or reassessing an existing relationship in light of AI claims, there are a few frameworks worth applying.

First, separate the platform from the people. WPP Open, Omni, CoreAI, and the rest are platforms. Platforms require skilled operators to produce results. Ask to meet the people who will actually work on your account and understand what AI tools they are using day-to-day, not what the holding group’s central AI team has built.

Second, ask for case studies where AI drove a measurable business outcome, not a production efficiency. Faster creative iteration is useful, but it is an operational benefit, not a marketing effectiveness story. The case studies worth scrutinising are the ones where AI-driven audience modelling or media allocation produced a demonstrable improvement in campaign ROI.

Third, understand the data provenance. The quality of AI output in advertising is almost entirely determined by the quality of data input. A holding group with access to genuine first-party consumer data is in a fundamentally different position from one running AI models on licensed or synthetic data. Ask the question directly and listen carefully to how it is answered.

Fourth, consider the independent alternatives. The range of AI tools available to independent agencies has expanded significantly, and some specialist shops are delivering AI-driven results that holding groups cannot match on specific use cases. The holding group model has advantages in scale, data, and integrated capability, but it is not automatically superior on every dimension.

For teams working through how to evaluate and implement AI tools in their own workflows, Semrush’s guide to AI optimisation tools and Moz’s work on LLM-based competitive research offer frameworks that are more practically grounded than most holding group white papers.

What Does the Competitive Landscape Look Like in Three Years?

The holding group AI race is not going to produce a single winner. The more likely outcome is that different groups establish genuine advantages in different areas: Publicis in data-driven personalisation, Omnicom in media planning and buying, WPP in high-volume creative production, Dentsu in technology consulting and system integration.

The bigger structural question is whether holding groups retain their position as the primary AI partners for large advertisers, or whether technology companies and AI-native specialists disintermediate them in specific areas. Google, Meta, and Amazon are all building AI advertising capabilities that reduce the need for agency intermediation in performance media. That is not a new threat, but AI accelerates it.

Early in my career, I was refused budget to build a website and ended up teaching myself to code and building it anyway. The lesson was not about stubbornness. It was that the people who figure out how to use new tools without waiting for permission tend to move faster than the institutions that are still debating the business case. The holding groups are the institutions now. The question is whether their AI investments are genuinely transforming how they work or just updating the language of the pitch.

The answer varies by group, by agency within each group, and by the specific team you end up working with. Which is, in the end, exactly how it has always worked in this industry.

There is more on the practical application of AI across marketing disciplines in the AI Marketing section of The Marketing Juice, covering everything from workflow integration to where the genuine limitations lie.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions