AdCP: Saviour or Just Another Way to Fleece Publishers?

My initial take on the Ad Context Protocol and why it might be the best hope yet for a sustainable open ad ecosystem

Disclaimer: This is my initial view, subject to change as I learn more and it goes out of theory into reality.

How I Made My Mind Up

Look, I'm increasingly using AI to get my head around complex industry shifts. For this one, I loaded the YouTube video on AdCP, the best articles about the protocol, plus the best stuff on the underlying problems (including my own writing) into AI (Notebook LLM, Grok for real time and Claude). Then I had it help me explore how AdCP could actually help, what the pitfalls were, and where the gaps might be. Shout out to Erez Levin and Gabriel DeWitt as well as last week Fern Potter , Gareth Glaser and Scott Messer

Bloody good way to get a proper view quickly. Here's what I reckon.

What Is AdCP Then?

The Ad Context Protocol (AdCP) is an open-source communication standard that launched on 15 October 2025, backed by a coalition of over 20 organisations including Yahoo, PubMatic, Magnite, Scope3, and others. At its core, it's meant to standardise how AI agents plan, execute, and optimise advertising campaigns across different platforms.

Think of it as the protocol for AI agentic media buying.

Built on Anthropic's Model Context Protocol, AdCP creates a unified interface that lets AI agents—whether built by advertisers, publishers, or ad tech companies—talk to each other using a common language. Instead of mucking about with proprietary APIs, custom integrations, and platform-specific workflows, advertisers can just describe what they want in plain English: "Reach eco-conscious car buyers on connected TV platforms in the United States this week with a budget of $50,000."

The agent sorts it from there.

Why I'm Cautiously Optimistic

Here's my view: AdCP is the best hope yet for publishers and the rest of us to have a sustainable open ad ecosystem. I'd liken it to agentic curation pushing private deals at scale.

1. Finally, Actual Transparency on Supply Sources

AdCP solves one of the web's biggest problems: you can actually see who you're buying media from. If there are breaches in future—fraud, quality issues, whatever—you can block those vendors at an ad server level. This closed loop between publisher and advertiser solves so many problems that have been festering for years.

Selling Rolexes next to fakes is a way to remove money from the market. Being able to select only quality publishers gives you confidence to pay the appropriate price. When you can't tell the difference, you pay the lowest common denominator price—or worse, you stop buying altogether.

Finding out who was at fault for issues used to be a complete nightmare. I've written about this loads. The opacity of programmatic's intermediary layers meant that when something went wrong, accountability just evaporated into a black box of exchanges, SSPs, and DSPs.

AdCP's design includes verified authorisation through functions like list_authorized_properties, which lets buyers confirm publisher authorisation before buying inventory. The protocol flips the traditional auction flow so advertiser agents talk directly with publisher agents, cutting out the intermediary bid streams that hide where the money's actually going.

2. Solving the Tragedy of the Commons

This transparency tackles what economists call the "tragedy of the commons." In programmatic today, bad actors pollute the pool and get away with it because they operate under cover. When transactions are opaque, fraud thrives.

AdCP's machine transparency and auditability mean every transaction can be verified by both parties. Systems like Swivel's Seller Agent document every single conversation with buyer agents—every query, every offer, every object created in the publisher's ad server. You get an immutable log, which leaves no room for the discrepancies and mysteries that currently plague the supply chain.

3. Publishers Actually Own Their Audiences

This is massive for both privacy and publisher remuneration. AdCP supports first-party data signals and contextual targeting, aligning with where the industry needs to go on privacy. Publishers can expose their inventory and audience data directly to agents whilst keeping control.

Optable, one of AdCP's founding members, specialises in privacy-first data collaboration. The protocol's designed to handle audit trails, data provenance, and identity interoperability as core requirements—crucial in a post-cookie world where transparency around data usage actually matters.

4. Competition and Innovation Can Emerge Again

By creating an open standard, AdCP lets fraud detection vendors, quality measurement tools, and attention metrics providers actually scale. The protocol's metadata scaffolding is built to facilitate composable sustainability—you can inject emissions or attention metrics as first-class filters in discovery.

Scope3, a founding member focused on carbon emissions measurement, demonstrated this at launch. This means competition can thrive again, and barriers to entry come down.

5. Fixing Private Marketplaces

Most private marketplaces don't work well. They're manual, operationally heavy, and limited in scale. AdCP reduces this cognitive load massively. Publishers can offer any advertising product requested in a brief—no matter how niche—without being constrained by manual packaging or operational faff. Integration timelines drop from months to days.

This is essentially agentic curation at scale, making premium inventory accessible without the usual friction.

The Critical Watch-Outs

No technology is a silver bullet, and AdCP has some important limitations you need to know about:

1. Retargeting Isn't Really a Thing

Because audiences are defined supply-side, any retargeting will need to be clean room style. This is actually good from a privacy perspective, but it's critical that buyers understand this for like-for-like campaign comparisons. Your traditional pixel-based retargeting strategies won't translate directly.

2. Measurement Gaps

Measurement isn't directly in the scope of this protocol. Your existing ad server should somewhat work—but only for environments where cookies still exist. For Safari, and as the cookie crumbles elsewhere (if Google ever actually removes it), we'll need better solutions. The walled gardens still have a huge advantage in measurement.

For AdCP to truly scale and take over, we really need either:

An open web conversion API that scales, or A wholesale move (walled gardens and all) to a cohort privacy-first measurement paradigm

Here's the reality: the DR budget for unmeasurable media is zero. Without solving measurement, we're only tackling half the equation. The upcoming Curation Protocol (planned for Q2 2025) will help with brand safety and context-based inventory curation, but performance measurement remains a proper challenge.

3. The Governance Question

AdCP is forming a non-profit governing entity to ensure neutrality—so no single company controls how the protocol evolves. This is critical. The history of ad tech is littered with well-intentioned standards that got dominated by a handful of powerful players.

Waiting for the regulator to fix things is a long wait for a train that won't come. The industry needs to sort this itself, and open governance is the way. The protocol is open-source (MIT licensing), which is promising. But will the governance structure genuinely represent publishers, advertisers, agencies, and ad tech platforms equally? That remains to be seen.

4. Adoption Uncertainty

Despite 23 launch organisations and strong initial momentum, network effects still favour dominant platforms with loads of existing integrations. Success depends on:

Enough publishers implementing the protocol to offer valuable inventory access Enough buyers adopting tools that use this standardised interface The shift to agentic AI actually materialising at scale

If any of these fall short, AdCP becomes another interesting experiment rather than a transformative standard.

This Is NOT "PMAX for the Open Web"

Let me be crystal clear about one comparison I've seen: someone called this "PMAX for the open web." Absolute nonsense.

Performance Max is a black box that strips control from advertisers in the name of automation. AdCP is the opposite—it's built on transparency and customisability. It can be a source for good, not the greatest enemy to effective advertising the world has seen.

AdCP maintains human-in-the-loop approvals, allows for custom integrations, provides audit trails, and operates through open governance. It's designed to complement existing systems, not replace them with an opaque algorithmic black box.

The Bottom Line

AdCP represents a genuine attempt to solve programmatic's original sins: opacity, lack of control, and a supply chain riddled with fraud and waste. By standardising agent-to-agent communication, enabling direct publisher-advertiser relationships, and building in transparency from the ground up, it offers a proper path forward.

As Mark Twain said, "The secret of getting ahead is getting started." Break down your complex task into a series of steps and start there. This is a great first step but needs great follow-up. The protocol addresses many of the foundational issues, but measurement, adoption, and governance will determine whether this becomes transformative or just another well-intentioned standard that fades away.

Is it perfect? No. Will it succeed? That depends on industry adoption, whether governance is executed well, and if measurement challenges can be solved in parallel.

But for the first time in ages, I'm seeing an industry initiative that could genuinely benefit the open web rather than just creating another layer of complexity. Publishers get more control and better monetisation. Advertisers get transparency and access to premium inventory. The ecosystem gets accountability.

That's worth paying attention to—and worth supporting, provided we stay vigilant about governance and ensure the protocol serves the many, not the few.

Congratulations Too

Brian O'Kelley and everyone else who put the effort into driving this protocol. I should list more but its clear Brian has been a driving force here. Perhaps he can pass on my congratulations appropriately. We had a lot of hope with programmatic, I hope we can learn the lessons of the past and do better this time!

What's your take? Are you optimistic about AdCP, or do you see it as another ad tech false dawn? Let me know in the comments.

#AdTech #Programmatic #AdCP #DigitalAdvertising #AIinAdvertising #OpenWeb #AdTransparency