Jig’s up, guys. The new book “Subprime Attention Crisis,” by Tim Hwang, former Global Public Policy Lead for Artificial Intelligence and Machine Learning at Google, says the ad tech bubble is going to go pop just like the subprime mortgage market a dozen years ago. The chief problem—the central building block of digital advertising, aggressive microtargeting, isn’t nearly as effective as advertised.
We’ve been found out—empty the safe, ditch your phones, and drive south till you can’t drive no more!
By the way, do a search for “ad tech bubble” and you’ll find a ton of stories from 2019 and earlier claiming the ad tech bubble was going to burst—typically using that exact language. Copyranter from Digiday said an explosion was imminent in 2016—right when header bidding was breathing new life into SSPs everywhere.
But beyond the hype, and the kinda tortured comparison to the subprime mortgage crisis, Hwang does have some very good points—one summarized in Wired’s review: “despite the extent of surveillance tech, a lot of the data that fuels ad targeting is garbage.”
The “bubble” argument is strange, especially when noting how much Facebook and Google dominate digital advertising and often can offer the multi-touch attribution advertisers crave—as long as they stay within the walled gardens. Otherwise, the attribution situation is mediocre, and often leaves advertisers disappointed. Hwang also discusses the arbitrage and pricing games that are propping up a lot of ad tech players.
The “bubble” isn’t going to burst—advertisers aren’t going to stop advertising online en masse tomorrow. And for sure they’re not going to give up the walled gardens, particularly small and medium-sized advertisers that line the Duopoly’s pockets.
The thing with microtargeting is that it’s often not used well. You can only get so granular before the targeting is worthless (Single aunts who love scrapbooking and skateboard parks?). This is aggravated but the abundance of cheap third-party data, often scraped using third-party cookies. Part of the issue is that the advertisers demand targeting without really understanding whom they should be targeting and why. (Oh man, let’s not get into creative.)
But the overarching problem with targeted digital advertising is that advertisers have been sold a targeting playground that was never realistic—outside of single-party controlled platforms like Facebook or the Google search ecosystem. Yes, the advertisers dived headfirst into the hype about the open programmatic space, and it’s only been in the last few years that they’re pulling back and saying, “OK, what’s real here?”
While we’re sure it pops up in the book, it’s weird that there’s no mention in the Wired article about the depreciation of the third-party cookie. That tool really powered the growth of programmatic and arguably we’re in the current yucky situation because we relied too much on it. That’s precisely why the death of the third-party cookie is exciting—a lot of bad and cheap data will hopefully exit the market and advertisers will turn to more legit solutions like publisher first-party data. Hopefully, the prevalence of better data will make for better advertising.
The ad industry has built a solid backbone for delivering targeted advertising, but many of us are struggling how to effectively use it—and in particular, what data should fuel it. That’s why practitioners are less interested in hearing about bursting and bubbles and contemplating, “How do we get this to perform?” (Oh, and what does “performance” mean, anyway?)