AdCP and AAMP: Reading a Standards Contest
Two rival agentic ad standards, who actually governs each, and why a DTC brand can track neither closely without missing anything that matters.
Chapter 1What Are AdCP and AAMP?
They are two rival answers to one question, and the bodies behind them are not collaborating. This is a standards contest, not a standard. The question both answer: when advertising is bought and sold by software agents rather than by people clicking in a UI, what language do those agents speak to each other?
AdCP, the Ad Context Protocol, is built on MCP and is usually described as OpenRTB for the AI era: agent-to-agent negotiation rather than impression-level auction. It launched in October 2025 with a founding consortium and is now published by an independent non-profit.
AAMP, the Agentic Advertising Management Protocols, is the IAB Tech Lab umbrella initiative. It was named on 26 February 2026 and published as a framework on 16 March 2026, reaching version 2.3 by mid 2026.
Chapter 2Who Actually Governs Each One?
This is the question most coverage gets wrong, and it is worth answering carefully because governance is what tells you whether a standard will still exist in two years.
The most common error is attributing AdCP to the IAB Tech Lab. It is not an IAB standard. AdCP is published by AgenticAdvertising.org, an independent non-profit, and the Tech Lab is behind the competing initiative. Getting that backwards credits a protocol to the body that built its rival.
AAO's governance, from its own page, 3 August 2026
Trade coverage says otherwise. It reports an interim board including additional names, and a membership count above 100. Neither appears on AAO's own governance page. Where coverage and the primary source disagree, the primary wins, and the coverage figure should be attributed as coverage or left out.
The Rothenberg detail is the one that tells you how serious the contest is. The rival camp recruited the former chief of the standards body it is competing with, as an interim advisor.
Chapter 3How Do They Differ Structurally?
AdCP is organised into four modules and AAMP into three pillars. The shapes are not equivalent, and the mismatch is itself informative: AdCP is organised around what an agent does, AAMP around what existing infrastructure has to absorb.
| AdCP | AAMP | |
|---|---|---|
| Published by | AgenticAdvertising.org, independent non-profit, in formation phase | IAB Tech Lab |
| Approach | New standards designed for agents from the start | Extends standards the sell side already implements |
| Built on | MCP | Existing IAB infrastructure, with an agentic real-time framework |
| Structure | Four modules: Signals Activation, Media Buy, Creative, Curation | Three pillars: Agentic Foundations, Agentic Protocols, Trust and Transparency |
| Identity layer | Not the organising principle | Agent Registry, providing verified agent identity and disclosure |
| Dated milestone | Launched October 2025 with a founding consortium | Named 26 Feb 2026, framework published 16 Mar 2026, version 2.3 by mid 2026 |
One production data point rather than a roadmap: FreeWheel and NBCUniversal ran the first MCP-based agent-to-agent media buy in January 2026. That is the kind of fact worth tracking, because a standards contest is settled by integrations that ship rather than by specification quality.
No AdCP version number appears in this guide
Chapter 4Which One Should a DTC Brand Track?
Neither, closely, and that is the honest answer rather than a hedge. Both are agent-to-agent protocols for the buying and selling of media between platforms. A DTC brand is not a party to that negotiation; its demand-side platform or its agency is. Neither standard changes what you do next week, and a guide that told you otherwise would be selling you urgency.
Ask your DSP or agency which one they are building against
This is the only question in this space with an operational answer today. If they have committed to one, that decision is being made on your behalf and you should know which way. If they say neither, that is also an answer and a reasonable one at this stage.
Watch integrations, not announcements
A standards contest is settled by what ships. A named production integration between two independent parties is worth more than any number of consortium members, and it is the signal that a protocol has left the specification stage.
Re-read the governance page, not the coverage
AAO leaving formation phase and seating its 40-director board would be a material change in how much weight AdCP deserves. That change will appear on its own governance page before it appears anywhere else, and coverage has already been wrong about this once.
Do not rebuild anything on either yet
The two camps do not regard each other as compatible, so building against the loser is a real cost. Nothing about your catalogue, your structured data or your agent readiness depends on which one wins, which is why that work is the safe place to spend the same hours.
Chapter 5What to Do While the Contest Runs
The work that pays regardless of which standard wins is the work that decides whether an agent can read and recommend you at all. None of it depends on a protocol shipping, and all of it is measurable this quarter.
That is not an abstract point. Six of the 72 DTC brands in our public registry answer every request with an error rather than a storefront, from a crawler user agent and a desktop browser user agent alike. No agentic advertising protocol resolves that. It is a rule at the edge, and it is the agent-commerce problem an operator can actually act on now. Registry figures read from our checked-in brand list on 3 August 2026; the live index carries current numbers.
Confirm an automated client can open your store
Request a product page with a non-browser user agent and confirm you get a 200 with real markup rather than a challenge page. A store that returns an error to every automated request is not losing the agent-commerce race; it is not in it, and nothing in its own analytics would say so.
Check the four OpenAI bots separately, not as a category
Blocking the training crawler is a defensible policy. Blocking the search crawler alongside it, which a blanket AI-bot rule does, removes you from ChatGPT recommendations entirely. The crawler guide covers the roster and how to verify a bot is genuine.
Make your product data machine-readable
Server-rendered Product schema with complete attributes, accurate pricing and current availability. Every protocol in this space transmits what your catalogue says; none of them improve it.
Measure whether assistants name you today
Unbranded, purchase-intent questions derived from your own catalogue, asked repeatedly, with failures recorded as failures rather than as absences. That number moves on work you control and does not wait for a standards body.
The two guides that cover those steps in detail are the OpenAI crawler reference and measuring AI visibility without fooling yourself.
The free growth audit checks the things that pay regardless of which standard wins: whether an automated client can open your store, what your robots.txt and CDN allow, and whether your product data is readable once a request gets through. Public surfaces only, no account connection.
Agent Commerce 101: How AI Agents Buy Products
What agent commerce is, how AI agents evaluate and recommend products, and what brands have to change to be picked in a channel with no second page.
Measuring AI Visibility Without Fooling Yourself
How to measure whether AI assistants name your brand, what a score of zero really means, and the five ways a visibility measurement goes wrong.