Why Ad Agency Leaders Are Walking Away from AI Data Ownership

A panel of ad agency executives at Programmatic IO New York 2026 signaled a notable shift in how holdcos think about AI: they want to build agentic AI tools for clients, but they want less ownership of the underlying data than before, and some are rethinking owning the tools themselves.

The panel included Sir Martin Sorrell of S4 Capital (Monks), Bob Lord of Horizon Media, Maggie Summers of Dentsu X, and Shane McAndrew of WPP Media. Together they represented a broad consensus that the traditional agency data and tech stack model is under pressure from agentic AI.

Key Takeaways

• Data ownership creates liability. Sorrell said owning client data opens agencies to accusations of "grading their own homework." With Dentsu projecting that algorithms will drive 75% of ad spend by 2028, transparency and neutrality with data partners are now competitive necessities.
• Client data stays with the client. Horizon Media's Lord said its blu. AI platform is structured so brand clients own both the underlying data and any AI models built on top of it. He described agency-owned data used for competing campaigns as "basically like taking a client's ideas and handing them over to a rival company."
• The FTE compensation model is under threat. Sorrell advocated for outcome-based pricing; Lord suggested structuring P&Ls around clients rather than practice silos and moving toward software licensing models.
• Agent-to-agent workflows are arriving now. McAndrew of WPP said agencies are shifting from human-operated DSPs to machine-to-machine interactions, extracting intelligence faster and compressing supply-chain costs.

The discussion flagged Meta's planned end-to-end creative, planning, buying, and distribution platform as an existential risk to the legacy enterprise agency model.

Full Article on AdExchanger