Media Buying Marketplaces

Media buying is undergoing a profound transition today, as Agentic Operators are launching and presenting as viable challengers to incumbent Adtech, potentially upending the decade-long relationship between advertiser and established audience platforms entirely.

Today, there are three key audience markets affected by this change:

• Walled Gardens, principally dominated by performance marketing ad buys via Google, Meta, and Amazon etc;
• Open Web, where Display and Programmatic buying is dominant – led by The TradeDesk for size, but also bought by Walled Garden AdTech;
• Traditional Media / CTV, where media buying is dominated by large agency group media buying on established platforms – often with leverage and ownership of primetime slots as part of the agency proposition to large brand advertisers – integrated with Programmatic ad buys to place the same rich media ad on the same screen digitally.

Often, advertised media budgets and expectations of purpose and measurement remain divided by Brand – where scale, reach and audience demographics rule – and Performance, where ad dollars principally go to walled garden ad buys with ROAS and spend efficiency paramount.

These Ad markets remain largely siloed, despite attempts by attribution and data science practitioners offering MTA and MMM respectively to interconnect measurement and offer a unified view of ad effectiveness and reach.

And because of those failures, organisations tend also to remain siloed around Brand and Performance to this day – despite now 20 years of opportunity for digital to align with traditional.

So it is into this context we find Agentic AI offering to upend the norms and offer genuinely integrated media buying and ‘hands off’ optimisation from the realms of Performance allied to a strategy based in Brand.

If this proposition sounds familiar, that’s because it is – this is the same proposition offered by Programmatic itself when it broke onto the scene a decade ago looking to break the stranglehold of Search & Social Adtech – now our Walled Gardens – over digital ad dollars and championing technology as the solution to a ‘right ad, right place, right time’ proposition which proved initially very attractive to global brands and their agencies alike.

However, with the launch of a new protocol in AdCP, we have an attempt to break out from the confines of pure Display into all types of ad placement using the convenience of AI.

Automated Decision Making & Measurement Gaps

Programmatic ultimately failed to live up to its promise of placing the right ad in the right place at the right time due to poor measurement and management of fraud.

As a result, it often became a siloed outreach or retargeting platform for open web ads being measured on reach at worst, and Brand Video spend on CTV in a performance marketing environment at best. Both of these approaches ultimately have given up on the core promise of interconnected ad buying due to lack of unified audience signal.

Not exactly the data-led proposition we were promised.

Allied to that are the repeated challenges of Ad Fraud in Programmatic which continues to dog the market to the tune of ~$50bn annually: again due to measurement challenges. And ad slot margin skimming by agencies bumping up the price of prospecting to the point it becomes unviable. This ultimately culminated in the most public failure of Programmatic to succeed on its own terms when P&G cut its media spend by $140m in 2017 to … no effect at all.

What we learned as marketers over the last decade boils down to this: automation and technology-led buying is only as good as the quality of the data signals provided.

Walled Gardens have prospered as they offer ad placements to properties that guarantee human eyeballs – so at least the bare bones measurements of impressions are at least somewhat meaningful for a Brand marketer. Therefore, Instagram, YouTube, TikTok and the like can capture Ad dollars which may not be optimisable compared to spend in a different Adtech due to their siloed data sets, but which do at least perform for basic MMM uplift – meaning they are seen and therefore influence conversions later – just not with any detail to allow real optimisation of either the Ad, the Journey, or the Message as the future customer travels through each Walled Garden on their way to conversion.

Outside of Walled Gardens we have seen the opposite: a rush to the bottom for publisher ad prices, broken signals about if Ads were even seen by a human and plunging interest from major advertisers who would pay premiums for brand placements – which instead go to Walled Gardens or remain in Traditional Media buys and CTV.

So the question at large today is: can agentic AI bridge the measurement gap and open up efficient ad buying at scale?

Certainly, the market players you would expect to see believe so – in the same way they believed fervently in Programmatic all those years ago. But AI is a different technology – can it perhaps finally deliver?

Let’s look at the technology and what is actually being executed by Agentic AI.

Building the Future on AI: Risks & Opportunities

As illustrated in the launch walkthrough of AdCP – the proposition being offered by Agentic AI is a move to natural language ad buying, opening up the prospect of Media Buying sitting outside of the current owners – potentially even on the advertiser side.

To enable this of course, there are suites of new technology to be adopted – as with Programmatic, principally:

• MCP – for Agent to Agent communication via the most efficient way to share context to Agent following the Model Context Protocol standard. Essentially, making data available without requiring scraping.
• A2A – direct Agent to Agent communication to allow handshake on collaboration steps in a wide range of Agentic Environments.
• APIs – RESTful API usage will continue to support legacy technical integrations that do not move to MCP.

This failback approach to APIs means that a lot of AdTech will be supported out of the box.

However, one of the key challenges faced by Programmatic rollout the first time round was the buy costs – which became very significant, peaking in 2024 where 56.1% of the cost of an Ad was lost to the process of buying that placement.

The AdCP approach should not layer significant extra cost on the ad placement, and advocates would suggest that having a wider market availability would allow high cost routes to ad placement to be competed out of the market.

Lets see how that shakes out with early adopters!

And there will be lots of early adoptors. The upcoming ChiefMartech report will reveal that just over 90% of media buyers are today already using agentic AI to place media today. So, in a very real sense we are already in the Agentic future, the only part that remains a question is who will win as the market shakes out.

However, despite such widespread adoption, measurement remains the main risk when considering the future of agentic media buying. In my view, there are three major industry shifts make reliable attribution mission-critical:

• Tracking signals are disappearing

Privacy changes have erased much of the user-level data advertisers used to rely on.

• Media is more fragmented than ever

Across platforms, formats, and audiences, identifying what actually drives revenue is becoming harder.

• Costs and competition are rising

Brands need real-time, accurate insight into what’s working to keep customer acquisition costs under control.

The brands that win will be those that can measure correctly and adapt instantly.

The opportunity on the other hand is clear:

Agentic media buying is the next evolution beyond automation. AI is today acting like an autonomous media buyer.

It can:

• Move budgets between platforms in real time
• Test and refine messages
• Learn what works and scale it

But autonomy only works if the AI knows the true impact of each marketing action.

Here at Corvidae AI, we are well versed in the value of accurate measurement – with global patents granted on the use of AI to stich sessions together and bridge the measurement gap as people move across devices in a data-compliance-first process, we are building paths that are 50% longer than cookie based systems, and feeding that data back into systems which deliver real ROI return within the first weeks of bringing Corvidae Conversion actions into existing Google Ad campaigns.

Reach out to find out more about Corvidae today, and get a free walkthrough and value impact assessment so you can be ready for the agentic media buying future.