The real AI action is in media buying
Is the industry ready for even less control and transparency? And fewer people?
What happens when the 24-year-old media planner isn’t a real person?
This appears to be the question many in the ad industry will soon have to grapple with, if they aren’t already. While the excitement regarding artificial intelligence over the past year-plus has been focused on generative AI (machine generated ads! no more creative department!), media planning, buying and optimization are where the real action is - and where things promise to accelerate quickly.
It’s worth asking whether the advertising and media industries are ready for this. To review where we are now:
Google has a product called Performance Max, which, in simplistic terms, buys ad space for brands automatically, without them knowing exactly where their ads are running or why they ended up there. It’s already lead to some brand safety problems.
Meta has its equivalent product, Advantage+. Plus, the company has largely credited AI with its advertising turnaround following Apple’s crackdown on mobile ID-sharing.
There are surely lots more to come, as pretty much every company in ad tech, along with a slew of startups and giants like Amazon, are headed in this direction.
In many ways, the future is already here, and with it, many uncomfortable questions. For instance:
Are brands going to have to give up more and more control of their media budgets, and even their strategies?
How does that jive with the industry’s push toward a more privacy driven approach, and a need for transparency?
Are media buying agencies ready for all this? Do they have the technology and trained staff to adapt this change?
Same question for media sellers?
Look at the circus brought to light by the Adalytics bombshell report this week. Brands are ending up funding all sorts of BS made-for-advertising sites. How does AI media buying not make this much worse?
Will this sort of media buying translate beyond the big tech platforms? For example, can the TV media companies - many of which are under financial duress - be able to compete on this front?
What will regulators think of all this?
Will regulators even be equipped to monitor AI media buying tools? How will they know, for example, if a Performance Max is violating someone’s privacy?
“I think as most people would say it's very unclear what Performance Max does and how it does it,” said Jeremy Fain, CEO of the AI tech firm Cognitiv on my podcast this week.
Fain gave me a very helpful tutorial for AI dummies, talking about what the differences are between concepts like neural networks and large language models and the like (I highly recommend it). I’m still no expert, but what struck me was what the investment and expertise that is going to be required to operate in digital advertising in the coming years (you’re going to need Deep Learning, which is hard), and the knowledge/capabilities gap that is going to create.
“Machine learning has been a part of the media buying landscape in ad tech for a long time,” said Fain. “Deep learning is very new to it. And in fact, almost nobody's using it. Amazon’s DSP recently released some features that feed into their algorithm that are deep learning based. Nobody that we know of is using native deep learning algorithms to buy per client.”
Is this the kind of thing that only the Amazon’s and Meta’s the world are going to be able to compete on? Won’t that deepen the power of the duopoly/triopoly?
I spent last week at the Beet Retreat in Puerto Rico, during which AI was a hot topic. There was talk about the coming Arms Race, and the need for investment in tech and teams.
There was some optimism that big TV companies will be able to use AI to mine better information on what people are streaming, and when best to show them ads in real time. “ AI's role in programmatic has always been there - you cannot have true automation without a certain level of machine learning & AI tech behind it,” said Marika Roque, Chief Innovation Officer. “AI's progress will bring a new level of speed, intelligence, precision and transparency to the data that informs real-time, in-flight optimizations at a scale, unlocking accuracy we haven’t seen before because it literally wasn't humanly possible until now.”
However, this kind of tech is not necesssarily the sweet spot of content companies or agencies. Rogue noted that silos created between analytics, buying and creative, “will need to be bridged in a more effective way.”
I can see why some companies should be worried. As Fain put it, “I don't have to have data scientists creating an algorithm for every single advertiser.” said Fain. “I can automate that now.”
Ok, so of course people at AI tech companies are going to tout AI. But here’s what
S4 CEO Sir Martin Sorell told Beet back in January, “This is controversial…you won’t need ten to 15,000 people in an agency network to implement [media buys]…the platforms will have closer relationships with clients.”
Sounds like a win for the big tech guys - maybe not for transparency.