The industry has spent years debating third-party cookies, but AI has settled the debate. AI decision engines optimized for outcomes (sales, retention, lift) require deterministic identity, clean feedback loops and governable data lineage.
First-party data isn’t just preferred; it’s structurally necessary. And the capital is already moving: IAB’s State of Data report found that 71% of brands, agencies and publishers are currently or planning to grow their first-party data sets, nearly double the rate from two years earlier.
Data architecture is the differentiator in the agentic AI era. Environments that can prove who the buyer is buying from, what was delivered and what signal was used, without turning governance into a tax that kills performance, will win the day.
The budget shift is real, but it’s architectural
Ad budgets are reallocating toward systems where identity and measurement are natively integrated. Retail media is the clearest example: structurally first party with owned inventory, authenticated audiences and closed-loop measurement. Emarketer forecasts US advertisers will spend $69.33 billion on retail media in 2026, up from $58.79 billion in 2025, with $9.42 billion of the $10.53 billion in incremental spending accruing to Amazon Ads and Walmart Connect.
Premium publishers are adapting to the same reality. Instead of relying on third-party signals, they are investing in direct audience relationships and privacy-safe partnerships that allow them to sell based on outcomes, not just pageviews.
For The New York Times, digital ads using first-party data accounted for more than 20% of its core ad revenue in Q4 2020, up from 7% the year prior.
NYT has since built on its data advantage with the 2024 launch of its BrandMatch AI platform that matches advertisers to the publication’s logged-in users. After a year in market, the platform has improved both click-through rates and video completion rates by 30%.
The shift is happening in health care and pharma as well. In pharma, if an environment cannot support auditability, clear consent provenance and controlled data flows, the spend doesn’t clear or it clears under conservative constraints and discounted pricing. Emarketer forecasts 2025 retail media ad spend growth in US health care and pharma at 21.3%, versus 11.2% for search. In regulated categories, governance isn’t overhead; it’s performance infrastructure.
Across retail media, premium publishing and regulated industries, environments that can prove identity integrity and data lineage capture the budget.
From real-time bidding to agentic allocation
Premium advertising behaves like a capacity market, closer to hotel rooms or airline seats than financial securities. The real decision is allocation across time, constraints and uncertainty.
Brian O’Kelley captured the shift succinctly (attributing the line to Benjamin Masse): “OpenRTB is a protocol for day trading; AdCP is a protocol for investing.” Old ad tech optimized individual trades. AI systems behave more like portfolio managers, allocating spend across goals and time horizons.
Agents change the economics because they collapse the cost of complexity. They can interpret offerings, negotiate constraints and monitor performance without forcing every publisher into a single standardized auction. This is how an advertiser moves from buying a handful of platforms to a broader portfolio without tripling headcount.
But agentic allocation only scales if the market converges on interoperable standards for products, identity and permissions; so agents can transact across environments without rebuilding integrations publisher by publisher.
Allocation requires longitudinal learning and accountable feedback loops. If signals are anonymous, probabilistic and poorly governed, the agent cannot safely optimize toward outcomes. If signals are first-party, permissioned and auditable, the agent can allocate capital with confidence. In an agentic market, first-party identity is not “fuel.” It is the ledger that makes allocation possible.
Governance as competitive advantage
In an outcomes-obsessed, agentic future, governance becomes a performance requirement. The more autonomous the decisioning system, the more an enterprise must answer: What signals were used? Under what permissions? Where did the data flow?
First-party data offers the best path to identity integrity and minimal leakage because the relationship, consent and control sit in the first-party domain. It improves auditability (who collected the signal, under what permission, how it was used), making compliance enforceable rather than aspirational. Data integrity is becoming critical for AI models to function reliably.
When compliance becomes a capability, the advantage shifts to platforms that can deliver performance while proving accountability.
The future of digital advertising depends on who controls the intelligence layer: deterministic identity, governable signals, measurable outcomes and an allocation engine that can operate across partners without collapsing under complexity.
Real-time bidding is no longer the whole market. The market is expanding upward: from impression trading to outcome allocation, from snapshot targeting to longitudinal learning, from informal trust to enforceable governance.
AI has already decided which inputs it can trust. The market is now deciding who gets paid for providing them.
“The Sell Sider” is a column written by the sell side of the digital media community.
Follow Newsweek and AdExchanger on LinkedIn.
For more articles featuring Alvaro Palacios, click here.
