IAB: Fix Ad Data Standards Before AI Takes Over

As generative and agentic AI moves from experiment to infrastructure in advertising, the IAB is pressing a foundational argument: automation is only as good as the data and standards underneath it. In a recent interview, Jamie Finstein of the IAB contended that the industry must "speak the same language" — align on shared definitions, taxonomies, and data interoperability — before handing more decision-making over to AI systems.

The warning is less about AI capability than about the chaos AI can amplify. Machine-driven planning, buying, and optimization depend on structured, consistent inputs. When the same concepts — audience segments, content categories, placement types, outcome metrics — are defined differently across SSPs, DSPs, publishers, and measurement vendors, AI models inherit and scale that inconsistency rather than resolving it.

Why Standardization Precedes Automation

For programmatic and video ad serving, the point is concrete. OpenRTB bid requests, content signals, and supply-chain objects already carry standardized fields, but interpretation still varies widely across the ecosystem. A content category passed in one taxonomy may not map cleanly to another; audience and contextual signals are frequently encoded in proprietary ways. Humans have historically papered over these gaps with manual reconciliation and tribal knowledge. AI systems operating at machine speed do not.

Finstein's framing suggests that before the industry leans on AI to automate media planning, creative decisioning, or yield optimization, it needs a reliable common vocabulary. That means shared taxonomies for content and audience, consistent definitions of outcomes and quality, and interoperable data formats that let one system's output be another system's trusted input. Without that, AI risks producing confident outputs built on mismatched assumptions.

The Standards Body Role

This is familiar territory for the IAB and IAB Tech Lab, which have spent years building the connective tissue of programmatic advertising — from VAST and OpenRTB to ads.txt, sellers.json, the supply chain object, and content taxonomies. The AI moment reframes that work as newly urgent. Standards that were already valuable for transparency and fraud prevention become prerequisites for trustworthy automation.

For the video and CTV side specifically, the stakes are high. CTV inventory is fragmented across platforms, apps, and sales channels, each with its own metadata conventions. As buyers and sellers deploy AI to navigate that fragmentation — matching demand to pods, optimizing frequency across devices, and measuring outcomes — inconsistent signals directly degrade results. A shared language for content, context, and measurement is what lets AI reason across a fractured supply landscape instead of being defeated by it.

Technical Implications for Ad-Tech Teams

The practical takeaway for engineering and ad ops teams is that data hygiene and standards conformance are no longer back-office concerns. Clean, standardized signals in bid requests, consistent taxonomy mapping, and well-maintained ads.txt / app-ads.txt / sellers.json files become the substrate on which AI tooling is built. Teams investing in AI-driven yield or buying tools will find their returns capped by the quality and consistency of their underlying data.

There is also a defensive dimension. AI amplifies both signal and noise. In an environment already contending with invalid traffic, made-for-advertising inventory, and opaque supply paths, automated systems trained on poor data can scale bad outcomes — misdirected spend, mismeasured performance, and reinforced fraud patterns. Standards and verification frameworks act as guardrails, ensuring AI operates on validated inputs.

A Call to Coordinate

Finstein's message is ultimately a coordination argument. No single vendor can impose a common language on the industry; it requires collective adoption of shared standards. The IAB's position is that the window to establish that alignment is now, before AI-driven workflows become so entrenched that retrofitting interoperability becomes prohibitively difficult.

For publishers, SSPs, DSPs, and measurement providers, the implication is to treat standards participation and data interoperability as strategic investments rather than compliance checkboxes. The organizations that align their data to common frameworks will be best positioned to deploy AI effectively — and to trust what it tells them. The ones that don't may find their AI initiatives built on sand, accelerating decisions that the underlying data was never equipped to support.

Stay on top of video ad serving and programmatic. Follow Adelerate.