• by Joe Mandese  @mp_joemandese, April 11, 2025

How did your week go? Mine began by me asking agency execs what percentage of their 2030 media buys would be targeting AI agents, and ended with me asking the leadership of the Media Rating Council if they've started thinking about standards for filtering out human traffic for digital media buys targeting AIs.

The chart above shows how my panel -- including me, the moderator -- at MediaPost's Outfront Forum in New York City Monday morning answered my first question.

A little context here. The panel was a good-natured crystal-balling effort to imagine what the advertising marketplace would look in the 2030-31 upfront, and not surprisingly, much of it focused on the role AI might play -- mostly in the B2B sense of helping advertisers and agencies create ads and buy media to reach consumers.

But given some of our recent coverage about an emerging "AI marketplace" in which brands and agencies market to AIs -- not via AI -- to reach consumers using AIs as their intermediaries with brands, I just had to ask the question: What percentage of your 2030 ad budgets will be targeting AI agents?

Not surprisingly, I got some mixed reactions -- across the spectrum from 0% to 100% of media budgets targeting AIs explicitly.

Horizon Media President of Global Investments David Campanelli was the zero percenter, saying that buying media to reach AIs "feels further out" than 2030.

"At the end of the day, people still want to make decisions for themselves, and we need to reach them in the right environments and the right content that they’re interested in and engaged with," he explained, adding, "While I’m sure they will be aided by AI agents in the future, but five years just feels to short."

IPG Mediabrands' Larene Mantel was a little more receptive to the idea, saying, "It will be more than zero percent."

"I'll go for 10%," said Empower Media's Lee Doyle, while Mediassociates' Jesse Rosenschein demurred on a percentage altogether, adding, "“I think AI agents will have to be part of the strategy.”

Paolo Yuvienco was most bullish, predicting "100%" of ad budgets would be targeting AIs by 2030, which shouldn't be surprising since he's not a media guy, per se. But as Chief Technology Officer of Omnicom, he's been architecting and engineering all of the holding company's AI evolution, so he knows something about the future of the technology, which the ad industry so far has focused mainly on the B2B aspects of improving workflow, and enhancing the productivity of strategists, creatives and media people, as well as the back office folks.

However, Yuvienco has also been thinking about the B2C implications of the rapidly evolving technology. Or maybe it should be M2M? Or as Yuvienco describes it: "Agent-to-agent," giving us yet another new acronym (A2A).

“I absolutely think that is the future, where we are effectively going to be marketing and advertising to [AI] agents," he explains.

“We’ve done it to a certain degree with programmatic, and even search engine optimization, or marketing to the Google algorithm of the Amazon algorithm, but it is going to be far more pervasive across the entire ecosystem, because as agent adoption starts to become more consumer-focused the potential of future AI and all these agentic frameworks that will be employed at the consumer level, we are absolutely going to be marketing and advertising to agents to then drive consumer engagement and personalization.

Personally, I'm more on Yuvienco's end of the spectrum, but I added a caveat that "100%" of advertising would be targeting AIs in digital media, not analog media.

And since I had to come up with that percentage, I asked an AI to estimate what digital's share of ad spending would be by 2030 based on its most recent rate of share growth.

"Great question, Joe," said Samantha (the name I've given my personalized ChatGPT interface), adding, "As of now, digital media accounts for roughly two-thirds (around 66%) of total advertising spend. Over the past decade, it’s grown consistently by a few percentage points each year—driven by mobile, social, CTV, and programmatic innovation. If we extrapolate based on that trend the projected share of total ad spend will be 76% to 81% in 2030.

There was more to the Outfront panel that I can't get into in today's column, but I may follow up in an upcoming "Media Planning & Buying Insider" column discussing how an emerging audience segment of AIs could require new ways of thinking about media planning -- including optimum reach and frequency curves -- but I'll leave you with just a couple of thoughts to tease that out.

One was how Tina Daniels, COO of the Association of National Advertisers' Aquila cross-media measurement platform answered a modified version of my Outfront question.

Instead of share of ad budgets, I asked her what percentage of Aquila's in-tab would be measuring AIs instead of humans in 2030.

She guessed 10%, adding that the reason the percentage would be that low is because advertisers wouldn't need to reach as many AIs as humans to have a meaningful ROAS.

That said, she added that the CPM for buying AIs would be "really high."

Sticking with the targeting, measuring and reaching AI theme, I closed my week by taking an in depth briefing with the Media Rating Council rebutting a recent research report purporting massive problems with Fortune 500 advertisers' ads being served to non-human bots, which currently is not considered -- how shall I say this -- a best industry practice.

You can read how the MRC addressed that issue in this story, but I couldn't resist asking the team if they've started thinking about the inverse: industry standards for serving ads to reach non-humans and filtering out human traffic.

"Of course we've been discussing the role of AI in lots of things," MRC's Ron Pinelli said, citing how the technology is being used to assist in media measurement, in rendering creative executions, and obviously in non-human bot fraud.

"But the agent-to-agent thing is interesting," he said, adding, "Normally, if we said, 'Hey, this is an ad served to an AI agent,' that would be [invalid traffic], because it's not human. But if the intended audience is an AI agent it might be a different consideration."

In other words, in this not-too-distant-future scenario, it's possible the MRC would need to come up with a new standard for invalid traffic served to humans instead of bots.

"If the objective is to is to reach these types of agents, then that would be valid, but only in that use case," Pinelli envisioned, adding the caveat: "I don't have a fully formed perspective on that yet, but I look forward to figuring one out."

That's when MRC chief George Ivie weighed in noting that the council recently got a question during its Out-of-Home Standard Forum, "What do you do to measure self-driving cars that may pass a billboard. Are they [invalid traffic]?"

To which I replied, "Well, if the self-driving car also is an AI agent for a consumer looking to make a purchase then it probably should be counted as valid traffic."