Special Report - The Future of Media Buying
All agents? All Meta? Can publishers and agencies compete?
This special report looks at the future of media buying, produced in partnership with Kochava.
As last week’s round of tech earnings made abundantly clear, artificial intelligence has infiltrated, if not already taken over, the practice of media buying (or really, media planning, buying and optimization).
Tech giants Meta and Google revealed staggering ad revenue growth, driven in large part by their uber-intelligent, somewhat black-box AI tools, which they plan to invest even more staggering amounts of cash.
That said, holding company Publicis provided a modicum of optimism that always-on-their-death-bed-agencies can maybe compete in this new world, with the right mix of data and technology. Yet there is plenty of skepticism that agencies will ever be able to compete with the engineering prowess and resources of Silicon Valley - along with a widespread belief that they’ll have to attempt to do so with far fewer people.
And then there’s the rest of the media world, from web publishers to mobile apps to traditional media firms, left to wonder: are we completely outgunned here? Or is there a way to counter this wave?
It leaves the future rather murky, if not ominous - with several (fairly extreme) scenarios potentially about to unfold:
The Zuckerberg, Just Give Us a Check Theory
The Maybe Big Media Needs to Band Together Theory
The Brian Lesser No More Media People Theory (or The Ad Tech Agents Rule Theory)
Whichever direction things head could determine how trillions are spent over the next decade, and how consumer brands are marketed, or what even constitutes a brand in the AI world.
Let’s try and examine each:
1)It’s Zuck (and Sundar and Jassy’s) World:
This would-be reality of course stems from Meta CEO Mark Zuckerberg’s infamous quote about brands just sending over goals and a budget, and he’ll take care of the rest. And while much of Madison Avenue has recoiled (at least privately) at this notion, the evidence points to many brands just throwing their hands up.
Over time, products such as Meta’s Advantage+, TikTok’s Symphony, Google’s Performance Max, Pinterest’s Performance+ and so on, may either become mostly the domain of small-to mid-sized performance brands (of which there are millions), or supplemental tools for big marketers - or their primary ad vehicle.
Yet many would argue that even as media buying becomes more performance-driven, brands still need to be the masters of their own customer relationship.
“To me, just writing a check to one of the big tech partners is a lazy way of doing things for marketers,” said Bob Lord, president of Horizon Media. “That disconnects you from your customers.”
Longtime programmatic ad veteran Matt Spiegel, now EVP, TruAudience Growth Strategy at TransUnion, told me on this week’s episode of Next in Media that going blindly all-in on AI optimization is “an overcorrection.”
“The best way to be really good at marketing in the future is to have a foundational data asset which is connected and scaled and then you can build from,” he said. “There will be instances, there will be executions where [automated AI tools] are great. But in order to plan my business, not just plan my next marketing campaign or not just get some transaction next month, but in order to understand ‘what products do I build?’, ‘what market segments do I take?’, ‘where’s the blue ocean?,’ I need to understand customers and consumers.”
There is also a huge question of whether AI models can understand, and optimize toward upper-funnel metrics like awareness, brand sentiment, etc. Even as dominant as outcomes have become, certain brands still need to tell stories and drum up interest in a product or category.
“The walled gardens are really good at demand capture, but brands need to move to demand generation,” said Jon Schulz, Chief Marketing Officer at Viant Technology. For example, “a channel like automotive is going to be tough [for a Meta or Google to deploy AI]. These companies need to think about, 'how do I stay in touch with a customer throughout that journey?’ Most major brands are not going to hand over the keys to a tech platform.”
2)The Publishers Strike Back Theory
This is an industry that seems to have tried umpteen consortiums aimed at blunting the power of the duopoly, to little avail. Still, this time may be different.
While it’s doubtful that a big group of media companies will band together and collectively share all their data with a large language model to build their own version of Performance Max, there may be an opening for the right third party to maybe help marry big media data with AI optimization.
For starters, Nielsen - which has been long known for playing TV ratings scorekeeper - is talking about taking on a much larger, more integral role in the media planning process.
CEO Karthik Rao told me recently that the researcher aspires to become more of an “enterprise” partner for brands and agencies - driven by AI and data.
“Our planning tools are embedded in all the agencies,” he said. What’s next may be Nielsen taking on more of “the role of activation. So we have a marketing cloud, we have other capabilities there that enable really using data to play a big role in the ad tech ecosystem for [the right] impressions to be reached.”
“Then you get into outcomes, the outcomes marketplace…For us, it’s that breadth of portfolio and how interconnected it is. And each of these have strengths in streaming now, each of these have strengths in broadcasting cable, and each of these have strengths in audio. So all these things sort of compound…That’s the power of the platform - is that it’ like it’s an enterprise use case.”
Kochava is another candidate. The measurement firm, which has its roots in tracking mobile attribution, has over the past year begun to help more media companies and digital publishers expand their attribution capabilities via its StationOne platform.
This week, at its annual summit in Sandpoint, Idaho, the company is unveiling Atlas Performance, which aspires to become a Supply Performance System, promising to help media buyers and brands unify all of their digital ad measurement and execution tools all in one interface.
The company is thinking big, envisioning helping ‘the rest of the web’ better demonstrate its performance value with the help of LLMs and a variety of data partners.
“When you look at Google, Meta, and now Amazon’s starting to catch up those guys just suck all the oxygen out of the room,” said Grant Cohen, Kochava’s GM, Measurable Extension. “Almost no other company does more than $10 billion in digital ad revenue, and that’s crazy when you think about the big publishers out there.”
The reasons for big tech’s advantages are obvious, Cohen explained - they have ubiquitous reach and massive data footprints. “We looked at that and said, ‘All right, we know what they’re doing. It’s a it’s a pretty clear playbook. It’s not easy to replicate. But, if we can we can focus on building out the tooling, with the help of advancements in technology and our position in the ecosystem. We believe we can develop something that helps make those incremental steps.”
Because Atlas will have visibility into how millions of ad dollars are deployed across the industry, plus a range of third-party-data sources, Cohen is hoping to build an attribution network infrastructure for publishers that can be competitive with big tech companies.
“Now we can start to build maybe not quite the data footprint that a Meta or Google have, but we’re enabling publishers to compete and make the most of their own data.”
This may not turn into an AI data collective, but Cohen said he could see non-competing media sellers, such as a Samsung and an Uber, sharing intelligence via the platform.
“If you’re a publisher and you could steal one, two, three % from those guys, that’s freaking monstrous,” he said. “It’s a massive change for the business.”
The Agencies are F-ed Theory, Thanks to Agents
Over the past few years, a growing number of industry execs have taken on the belief that media buying will become so machine and data-driven, that the role of the media buyer steering the ship will be as dated as Mad Men. Who needs junior media buyers and trafficking departments when you have AI agents?
Rajeev Goel, CEO of Pubmatic, described the Agentic Advertising vision on a recent Marketecture podcast
“Unequivocally we view it as really being the future,” he said. “I think this is much more than a technological shift. It’s going to be a massive value-chain shift. I think we can make advertising a lot more effective and probably actually compete ultimately with the walled gardens on performance. And I think that is really the big prize.”
“My prediction is that AI will upend and collapse the industry’s value chain and bring advertisers and publishers much closer together,” Goel added.
That sounds not great for many ad tech middlemen, and maybe agencies (though be clear, Goel did not say ‘agencies are dead’ or anything to that effect.
Of course, agencies have long been under fire. This time around, things feel - precarious (Axios: Robots Come for Advertising). In early 2025, WPP boss Brian Lesser said he believes that in five years, “there is no human that touches a media plan.”
Since then, pretty much every agency holding company and significant independent player has embrace AI, while touting building a proprietary platform.
How many of platforms are simply white-labelled versions of other company’s tech versus something proprietary is an open question.
MIQ has been one of the companies tasked with trying to help agencies, while maybe replacing some of their manpower, with its Sigma tool, which is aimed at helping agencies work across disparate platforms using AI tools and interfaces.
“Revenue per head - every agency we talk to is thinking about that,” said Moe Chughtai, Head of Advanced TV at MiQ. “That’s a real phenomenon.”
“The challenge is digital media is actually super inefficient. It requires tons of man power. So how do you change that?”
Initially, MIQ decided to focus on the bulk actions that agency ‘traders’ have had to execute in the past - including a unified UI for running multiple campaigns on several DSPs.
“We asked, much can the system do things? We are probably going to expand that later in the year.”
Of course, while advancements along those lines make things easier for agencies - they may also make it tempting for leadership to get rid of more people over time.
Yes but, historically, while ad tech innovations have made media buying faster and more accountable, they’ve also yielded new jobs and disciplines.
Could that happen again? Horizon’s Lord is betting on new types of talent coming to the agency world - and a much brighter future than some predict.
“I don’t believe that [Lesser quote],” said Lord.
“Now, I don’t think a big hold co can out innovate Oracle, IBM, Google, flat out. What you need to do is bring in these innovations, and be as open as possible.”
In fact, Horizon just required its employees to undergo 12 hours of training on new systems, including its blu platform, which allows employees and clients access to the same AI planning tools.
“We are seeing RFPs come through asking us to transform and meet the needs of our clients,” added Domenic Venuto, Horizon’s COO. “So we are redefining many of our roles.”
That’s hopeful, except for when procurement or the CFO get involved. Schulz recalled his time working at Ford in the mid 2000s, when he used to have a few hundred employees, while also farming out tons of tasks to contractors.
“That model is changing,” he said. “In the future, it’s not going to be as people-driven.
Ok, but there will still be some people, right? “I think we are a long ways away from ‘there’s no need for anyone to plan or do anything,’” said Cohen. “Maybe I’m old, but I don’t think you ever just push one button and it goes.”