Google Performance Max (PMax) and the coterie of other AI-based ad platforms, including Meta Advantage+ Shopping Campaigns (ASC) and, more recently, Amazon Performance Plus, have taken hold in the past couple years.
In fact, the first edition of AdExchanger’s Commerce Media Newsletter, in 2023, was a report on the mechanics of PMax.
AI and machine learning services have come a long way since then, and these changes have introduced a whole new vocabulary for marketers trying to get a handle on how products like PMax work.
The stages of PMax
For many advertisers, their PMax adoption journey has mirrored the old adage about grief. The response model begins with denial – “That’s not my ad and I never wrote that copy” – and then moves through anger, bargaining and depression, before eventually reaching acceptance.
Several recent reports reinforce how advertisers are coming to terms with platform machine learning tech running their campaigns.
A week ago, The Wall Street Journal had a story featuring agency ad buyers pushing increasingly more of their overall marketing spend through PMax, ASC and their ilk. And just today, Eric Seufert of Mobile Dev Memo wrote about a concept called “satisficer’s regret,” which is the uncomfortable feeling that these products satisfy their stated goals, while only just sufficing. In truth, they could do better.
“Satisficing” is the unholy portmanteau of “satisfy” and “suffice” – it’s taken from economic and psychological theory – but it accurately describes the feeling many data-driven marketers have about platform ad products. Fundamental changes are happening in the background, and advertisers increasingly don’t understand how their own marketing works, or even what their marketing is.
For example, a PMax campaign might consist of copy generated by Google’s AI. It might target search keywords an advertiser wouldn’t think to use (or agree to use), and serve ad formats that advertiser wouldn’t select for themselves. PMax advertisers might serve thousands of YouTube ads, for instance, and never know which video creatives were used when and where.
That’s why the new nomenclature is that advertisers “steer” their PMax campaigns. Rather than operating the controls in flight, they feed the product with metadata and targeting parameters upfront and have to simply hope they hit the right audience.
Advertisers receive less and less “analytics” data, which is user-level info that can be incorporated into a their CRM system or CDP. Instead, they now get an “Insights” tab, which is full of useful tips but also means no data ever leaves the platform.
AdExchanger Daily
Get our editors’ roundup delivered to your inbox every weekday.
Daily Roundup
Why it matters
The ascension of AI-based ad products has an air of inevitability. There is no longer a sense that advertisers can hold the line. This is where online advertising is going.
But the reason for these vague yet pointed tweaks in terminology is that advertisers arguably still don’t really know what they’ve signed up for and are only just starting to glean how much of a change this new framework is from what came before. Advertisers are putting the controls in the “hands” of a system they can’t be certain is charged with doing the best it can on a brand’s behalf.
Take the notion of “satisficer’s regret,” as posited by Seufert.
PMax might meet your stated conversion goals and preset cost controls, but what if it could have done much, much better?
The fact is, PMax primarily serves the goals and purposes of Google, and its priority is to deliver adequately on an advertiser’s “stated budget goals” – which is also an important term.
One retail advertiser and an early PMax adopter recently told me that the clearest indicator that PMax, ASC and other AI ad platforms are built for the platforms, not just by them, is the fact that they always consume an advertiser’s full daily budget.
Compare that to what one DTC brand marketer who is a heavy PMax and ASC spender told me, which is that the campaigns run by their agency – which have similar KPIs and cost constraints as campaigns that run through the platforms – often don’t meet their daily budgets. These are tight budgets, and there simply isn’t an infinite supply of new customers.
Yet ASC and/or PMax fill that daily budget every time.
And not only do the machine learning products spend the whole budget on any given day, they actually often spend more.
In 2023, Meta had a policy that the ASC platform would be allowed to spend up to 125% of a brand’s daily budget cap, which resulted in painful losses when a glitch led to ASC overspending on some accounts by 125%. This was before advertisers knew to tightly control their PMax or ASC daily budgets, because back then campaigns rarely met their daily spend limit.
But Meta has only increased the limit over time, and ASC now spends up to 175% of daily budget caps.
Google, for its part, has invented yet more new terminology to deal with the confusion. Advertisers now set an “average daily budget” on PMax, which is not actually a spending limit. The “daily spending limit” is double the daily budget.
In reality, however, most advertisers don’t understand what’s going on under the hood of their own platform-based marketing anymore – and new terms to explicate the changes only add to the confusion.
But guess it doesn’t matter as long as advertisers are satisficed with the results.
