Publicis Groupe’s CEO, Arthur Sadoun, told analysts last month that it had become clear that his company and WPP were now pursuing very different strategies.
While Publicis is throwing its resources into tech that allows it to identify consumers on the internet at an individual level, WPP is exploring federated learning technology, which targets people probabilistically, without user-level data.
Sadoun is supremely confident that Publicis’ approach is the right one, saying, ‘leading in identity gives us a unique competitive advantage in new business, but also in client retention.’
But do advertisers need to know people’s identity in order to market to them, or is it enough to understand behaviour and context in order to communicate effectively?
We asked our network how they think both the Publicis and WPP approaches will play out going forward.
Chris Camacho — CEO at Cheil UK
Both have merit. But if you’re asking which feels smarter for the future, probabilistic, confederated learning edges it. Publicis’ investment in ID tech is logical short-term thinking. It shores up addressability in the face of cookie deprecation and increasing data regulation. But it still hinges on identity resolution at the individual level, which is only getting harder and riskier. WPP’s move towards probabilistic, federated models feels braver and more future-proof. It accepts the inevitable: that scale, privacy and compliance will be the pillars of the next era of digital marketing. Building targeting models without clinging to PII [personally identifiable information] aligns better with where consumer sentiment and regulation are headed. The brands and platforms that win long-term will not be those that know everything about you, it will be those who know enough to deliver relevance without crossing the line. WPP’s direction feels more in tune with that reality.
Claire Elsworth — strategy director at Impression Digital
Framing this as a technology arms race misses the point: our role as agencies is to deliver consistent, sustainable growth for our clients. Technology is the facilitator not the solution, and growth must be measurable, incremental, and resilient. Both identity-based and probabilistic approaches have strengths and vulnerabilities. ID-based might be an easier short-term sell, but potentially brittle, in a world where regulation varies enormously by market, and is shifting near-constantly. The real question is not which will ‘win’, but how we deploy evolving capabilities to best serve each client’s ambition; finding the balance between precision-led gains and durable business growth over time. Businesses need strategies that endure, supported by technology that evolves as regulation, consumer sentiment, and platform capabilities shift. The smart client money isn’t on any single method, it’s on an agnostic approach that keeps results at the centre of the strategy, however the tech landscape evolves.
Chris Andrews — head of marketing technology at Wake The Bear
Publicis have bought a good business in Lotame, and WPP have done the same with Infosum. They are both perfectly reasonable bets. I’d personally lean towards clean room tech as the stronger long-term option, but there’s a solid logic either way. But the fact remains that this is just the latest in a long — and not especially storied — history of big agency groups buying up tech, to shoehorn a one-size-fits-all solution onto advertisers dressed up as ‘proprietary technology’, that they’re handcuffed into for time immemorial, paying undisclosed margins into the yawning void, for fear of losing access to what should be their own data. After all, for any agency group, it’s easy not to lose your seat at the table if you’ve bought up all the seats and tables in the first place.
Niki Chana — programmatic strategy director at SBS
The conversation we’re having now isn’t just about tech, it’s about building long-term resilience in the industry. Publicis’ ID-led approach still delivers results, but it’s built on identifiers like cookies and device IDs that are steadily losing ground. It’s effective in the short term, but in the long run it’s like building on sand — vulnerable to ever-evolving regulation, privacy concerns and platform restrictions. WPP’s tilt toward probabilistic models reflects an acceptance that change is already here. It’s not about pinpointing exactly who someone is, but understanding how they behave, what they care about, and how to meet them in the right moments. It reminds me of the early shift to mobile-first design. Bold at the time, but now the standard. The same shift is coming to how we target and measure. As media traders, we don’t need to know who someone is, we need to understand behaviours and context. That’s what drives sustainable performance and builds trusted client outcomes over time. Both strategies carry uncertainty. But if we’re thinking beyond the next quarter, towards lasting brand value and long-term effectiveness — probabilistic, privacy-first thinking is the shift worth making.
Elja Polak — chief growth hero at SuperHeroes Amsterdam
The big networks have spent years chasing data and missed the real shift. Publicis is buying ID tech. WPP is building prediction engines. But both are still playing catch-up with Alphabet, Meta, and Amazon, platforms that already own direct audience relationships at scale. But the real battle isn’t for data; it’s for attention. That’s where the networks keep falling short, throwing money at tech to cover a creative deficit. In 2025, attention isn’t bought — it’s earned. No one cares how precisely you target them if the message is forgettable. The future belongs to earned-first thinking: ideas that move because they’re worth talking about, not because you paid to push them. Data is a weapon, but creativity wins the war. Platforms are sprinting. Audiences are tuned out. Perfect targeting of a boring idea? That’s failure in high resolution. If you’re not building something worth sharing, you’re invisible.
John Clarvis — data and insight director at The Kite Factory
While they are logical solutions, both are rooted in replacing an old system (cookies) rather than creating a new solution. ID tech is the safer bet as far as convincing clients and advertisers is concerned, as they already understand it. However, it faces the challenge of remaining accurate within the constraints of the law, which globally becomes very complicated very quickly. I prefer WPP’s approach. Probabilistic approaches have incredible potential for doing smart, business-savvy work, and until now have found limited uses. It’s a clever way of targeting without violating privacy. It will also improve as it learns, making it a great investment. The challenge will be convincing clients — who are used to hard numbers — of its accuracy when, by definition, it doesn’t conform to traditional methods.
Chris Pitt — CEO at Gain Performance
The power lies in uniting the two types of technology to create a holistic strategy. The magic really happens when you apply algorithmic analysis to the data, alongside your own proprietary technology and secret sauce AI-driven capabilities, to gain the complete data picture. Using these technologies together is the smartest approach, because a considered strategy really needs both.
Mattia Fosci — founder and CEO at Anonymised
Amidst all the uncertainty created by privacy, competition and the rise of AI, Publicis is prioritising short-term revenue uplifts over long-term investment. This will be a smart move only if the payback period for their investment is relatively short, which is a tough ask. Buying Lotame seems more of an attempt to fix Epsilon than a long-term strategic investment. WPP is playing the long-term game with privacy-focused federated learning, but like all those who’ve been waiting for data protection to have an impact on industry practice they have been left waiting. And unless their product can deliver results in a short time, they run the risk of being surpassed by a much more substantial restructuring of the ecosystem driven by AI and LLMs. Perhaps they too will need M&A to plug the gaps still affecting their new product…
Featured image: Mauro Mora / Unsplash