Four pieces of business totalling nearly two billion dollars in annual media spend were won in the first quarter of 2026. I’ve looked at each piece of business in isolation to lift insights beyond the trade press reporting and company announcements. Do these deals represent the new era of media orchestration or something more recognisable?
What Was Won, and by Whom
Four deals · Three holding companies
The Business Context Around Each Deal
Estée Lauder was midway through its ‘Beauty Reimagined’ business turnaround plan. Chief Digital and Marketing Officer Aude Gandon, appointed in August 2025, arrived from Nestlé with a mandate to rebuild marketing infrastructure under a new operating model, One ELC. The model has three components: One Team (deployed July 2025), One Culture (introduced February 2026), and One Operating Ecosystem which saw WPP join Accenture and Shopify in April 2026. The media consolidation is the commercial execution layer of a restructuring already underway, not a standalone decision. A review announced January 23 and concluded by February 24: four weeks for a £500 million global account. The industry benchmark for a process at this scale is three to six months.
Wendy’s situation was more acute. CMO Lindsay Radkoski described it directly: “our business performance has not been where we want it to be.” With spend down 40 percent year-over-year and creative and media operating in separate holding companies, the consolidation removed structural friction during a period of commercial pressure. This was not a strategic upgrade. It was a practical fix under duress. A three-week competitive review, with the outcome to expand a 14-year creative relationship already in place.
IBM entered the review under a different kind of pressure. Media spend contracted 42 percent over two years. In March 2026, Ogilvy chose not to participate in IBM’s creative RFP, ending a 32-year partnership. WPP Media made the same call on the media side. Omnicom, which already ran IBM’s EMEA media business through Initiative, was selected over Publicis and Dentsu in the final round, extending an existing regional relationship to global scope.
Microsoft is categorically different from the other three. The media account is a component, not the headline. The core transaction is an enterprise AI transformation partnership: Publicis Sapient’s Slingshot framework helping client organisations migrate legacy systems to Microsoft Azure, Epsilon’s identity layer embedded across marketing workflows, Microsoft 365 Copilot deployed to all 114,000 Publicis employees, and Azure named as Publicis’s preferred cloud provider. The media AOR follows from the infrastructure partnership. Business need: agentic AI deployment at scale, inside a long-term commercial relationship already operating at the board level.
The Trust Component
All four decisions involved pre-existing relationships.
Aude Gandon spent the five years before joining Estée Lauder as Global CMO at Nestlé, one of WPP’s largest global clients. She did not evaluate WPP from a distance; she had spent years working alongside the organisation at scale. Lindsay Radkoski deepened a partnership that had operated for over a decade rather than testing a new one. The Publicis and Microsoft relationship extends back to 2016 and the co-creation of Marcel, ten years of joint development before the media account moved. IBM’s appointment of Omnicom extended an existing EMEA mandate rather than introducing an unknown partner into a period of significant agency transition.
When deploying AI orchestration at scale, under real business conditions, for the first time, three of these brands ran accelerated or no-pitch processes, limiting time for a full review of technically differentiated options. IBM was the exception: a standard review timeline, with incumbents departing rather than competing. In each case, they chose the partner whose capabilities they had already seen perform.
A Question of Compounding Value
Notes to the CFO
The features of the vertical stack involve proprietary agency infrastructure plugging into platform walled gardens. Audience is defined by the agency’s identity layer, media is placed through third-party platforms (Meta, Google, TikTok), and campaigns are executed within those same platform environments. The agency carries its proprietary data and optimisation logic into the platform’s auction system; the platform executes and collects performance data; the agency then validates results through its own measurement layer, creating potential for a double closed-loop where neither the brand nor independent auditors can verify the complete chain from audience definition to revenue attribution.
But. Do brands retain a channel to independently validate the results? Or more precisely, independently verify the relationship between campaign performance and revenue? If not, that represents an asymmetrical understanding of performance between brand and agency.
For some brands this will not register as a concern. Brands have always handed significant parts of their marketing operation to third parties. Trust has always been part of the relationship.
But this is actually where the blind spots I see every day in marketing systems emerge.
There are financial reasons to look more closely.
The ANA documents that only 41 cents of every programmatic dollar reaches consumers as working media, with the remainder absorbed by ad-tech fees and media-quality losses. Combined with a 30% ROAS overstatement in platform self-reporting and a 40–70% content utilisation gap, these are P&L issues: operating inefficiencies that drain EBITDA and inflate customer acquisition costs. Visible in the quarterly close and recoverable through operational intervention.
The balance sheet issue is structural and not visible in period reporting: marketing spend now creates intangible assets that compound outside the brand’s control. In closed-loop architectures, the same capital that the brand expenses as media spend trains AI models, audience segments, and attribution algorithms that appear on the vendor’s balance sheet, not the brand’s.
The Asymmetry
Traditional brand-agency relationships treated marketing as opex-for-talent. These full-stack partnerships are capitalisation arrangements: the brand funds the AI training through media spend; the vendor books the resulting data moats and algorithmic IP. The balance sheet transfer occurs silently, period after period, until contract termination reveals the brand has built intelligence infrastructure in someone else’s castle. The language of competitive moats has become a fixture of advertising services thought leadership of late.
Microsoft is a partial exception: as the owner of Azure, the infrastructure where Publicis trains its models.
The Audit Question
When AI-driven efficiency gains are presented, verify: does the trained model, funded by our media spend and crafted from our 1P data, appear on our balance sheet as a portable asset, or does the economic value compound in the vendor’s infrastructure?
They started with their first-party data. What else do they retain?
If the answer is undefined, the next contract begins where this one began. A new partner, new infrastructure, relearning what the previous partner spent years building. The audience models. The attribution logic. The optimisation weights trained on the brand’s own spend. None of it transfers. The brand funds the rebuild.
Trust, which was the rational mechanism for shortening or skipping the pitch process, is also the variable that removes scrutiny at the moment of signing.
It is what makes derivative data and intelligence portability clauses feel unnecessary. It is what makes nobody stop to define what leaving looks like before the contract starts.
What This Tells Us About the Market
These deals do not represent brands moving confidently into a new era of AI and data-powered media and content orchestration. Rather the CMOs leading the pitches are consolidating with known quantities while the new era is still forming. In return they are banking on ensuring greater speed, smarter more precise campaigns and connected execution in place of silos.
What I am observing across the industry are brands at different stages of a journey, where the 2020’s opex-for-talent playbook is running out of road. Most brands continue to outsource their intelligence almost as a rental agreement, the way they would have outsourced for creative services and talent.
The outliers have started to take a different view where contracts are seen as enabling access to innovation and filling gaps in capability. The brands evolving to a more sovereign playbook instead build their own intelligence machine and stop renting access solely to external closed loop systems. Capability and skills gaps within the brand continue to be outsourced as they always have been but the contracts feed the brand’s intelligence machine as it evolves over time.
#OwnYourIntelligence
