This week, AdExchanger hosted Programmatic IO in Las Vegas, and we recorded The Big Story podcast live on stage.
It’s not easy to distil two-and-a-half days of content down into a takeaway, but if there was a theme of the conference, it would have to be the “‘i’ word: “incrementality.” It came up again and again as a pressing concern for advertisers.
Incrementality testing is a useful tool for many traditional CPG and grocery brands. The vast expansion of retail media platforms means that there are now far more mini walled gardens, all of which have strong purchase data – and all of which are using programmatic campaigns to gain attribution credit for the same sale.
‘Incrementality’ is a buzzword on social platform advertising, too. Meta, for example, has adopted incrementality testing and uses markets as holdout groups to try and isolate its own campaign ROAS.
But what else was top of mind at Prog IO?
We’d be remiss not to mention generative AI technology.
Needham analyst Laura Martin, speaking on-stage Wednesday, predicts that the eventual winner among the major LLM operators – Microsoft/OpenAI, Google and Amazon’s roster of platforms – will eventually be twice as big as Apple is now.
Chew on that.
But the AdExchanger team chewed on something else: the fact that no one seems to be acknowledging that generative AI tech is being fed by thievery.
“We’re all kidding ourselves if we think AI is anything other than wholesale theft of content, and complete plagiarism at scale,” said Anthony Vargas, our senior editor, to applause from the audience. “That’s what it is.”
Then we dropped the mic and everyone went for a drink at the Prog IO happy hour.
AdExchanger Daily
Get our editors’ roundup delivered to your inbox every weekday.
Daily Roundup
Also this episode: Our other “a-ha” moments from the conference, including observations on CTV, the state of commerce media, programmatic transparency – and the lack thereof. (Oh, and as an aside, somebody at the show told me they’d fed my AdExchanger articles into an LLM that could practically reproduce my work in a snap. Which is cool. Not.)
