What Does Agentifying AdCOM Really Mean?

In our blog post framing the Agentic Advertising Management Protocols (AAMP) Initiative, we outlined the pillar “Agentic Protocols.”

This is the management layer of agentic workflows. It defines how buyer and seller agents understand each other, discover, negotiate, transact orders and deals, exchange audience signals, and complete laborious and cumbersome setup tasks required before execution. This includes work like Agentic Direct, Agentic Audiences, Agentic Mobile, and Agentic Ad Objects (derived from the Advertising Common Object Model), along with buyer and seller reference agents.

We’re diving a layer deeper on the Agentic Ad Objects (derived from the Advertising Common Object Model) to think about the value of “agentifying AdCOM.”

What is AdCOM?

It’s the standard that many haven’t heard of. Everyone knows OpenRTB. Fewer are familiar with AdCOM.

Yes, AdCOM was part of OpenRTB 3.0
And, yes, OpenRTB 3.0 never surpassed the usage of OpenRTB 2.x.

But while the object model of AdCOM didn’t gain traction, the enumerations from the specification are highly extensible, extremely useful, and are leveraged in OpenRTB 2.x today. As such, they are the de facto common language for media, placement, and context definitions for billions of advertising opportunities.

Some of these definitions include critical components of media buys. An easy example would be the video placement type. Say a buyer wants to purchase only in-stream video inventory. They don’t want any interstitial video or standalone video ads. To go a little deeper, say they wanted to only buy the first slot in pods of in-stream video.

This requirement of a media buy would trigger some questions in an agentic workflow: What is considered in-stream inventory? Does every potential seller use the same definition? Does in-stream mean only specific commercial breaks in CTV inventory? Does it mean any video ads played alongside streaming video? How do you know an opportunity is for the first slot vs the last slot, vs any slot in the pod?

Don’t panic. These use cases have not only been raised but also maintained through industry learnings over ten years in Tech Lab Working Groups. It’s been discussed, peer-reviewed, and commented on. We have an agreed-upon standard for the industry definition; shouldn’t we expect agents to adhere to it for listing inventory and buying inventory?

Another feature of the AdCOM enumerations, describe support of various technical frameworks for an ad, for example, whether Open Measurement is available for impression and viewability measurement. We already know as an industry that if a buyer requires OM Support, they need only look for a signal from the seller that they support API Frameworks 7.

So why does AdCOM matter in Agentic Advertising?

Today, the AdCOM enumerations are used in OpenRTB transactions. As we test and learn in agentic buying and selling workflows, we want to leverage this common language for describing an ad opportunity when deciding on how to fill and configure a media plan.

Let’s think through an example:

I’m a buyer, and I want to fill the in-stream video portion of my media plan with 20% first slot in pod placements and 95% Open Measurement verified viewable impressions.

Great, an agentic workflow that already understands the AdCOM standard will know to use this definition of in-stream, target the enumerations for first in pod vs any or last, and know that OM SDK is the 3rd party verification standard available with the standard through the “API Frameworks” attribute and associated enumerations. Next the agent will know to include an <AdVerifications> element in the VAST tag, with the appropriate measurement partner’s <Verification> node.

Without AdCOM, what is the source of truth for what an “in-stream” video is?

How do you ensure you’re targeting an appropriate amount of ‘first in pod’ inventory?

And how do you prepare the correct assets for measurement requirements in the ad?

These are just three small use cases of a decade’s worth of learnings and definitions. These critical components that describe ad opportunities and buyer requirements for them, have already been discussed, debated and cemented in standard by IAB Tech Lab working groups.

AdCOM’s object model might not have achieved widespread market acceptance, but its enumerations serve as precisely the kind of extensible shared vocabulary needed for agentic workflows. In an agentic ecosystem, shared semantics are more important than perfect object model design because they provide machines with a consistent language for classification, constraints, interoperability, and trusted automation.

While we’re happy to continue to adjust the standards as we explore the new requirements of agentic media buying and selling, we believe in building on the solid foundation that has served the industry for over a decade.