Home Ad Exchange News The Summer Scoop On PMax; Risky Influence

The Summer Scoop On PMax; Risky Influence

SHARE:

Here’s today’s AdExchanger.com news round-up… Want it by email? Sign up here.

The AdExchanger team is celebrating our independence with a long weekend off! The daily news roundup will return on Wednesday, July 5.

Maxxed

Mike Ryan, head of ecommerce insights at Smarter Ecommerce, posted a Twitter thread about the relative performance of Performance Max, Google’s world-eating, AI-brained ad product, based on 3,000 PMax ecommerce campaigns compared to Google Standard Shopping.

TL;DR: PMax does work on its self-proclaimed merits of effectively reaching a target RoAS. It outperformed humans in four out of five campaigns.

But PMax flat failed for marketers with low-volume sales (fewer than 30 per month).

And consider this: Counterintuitively, overachieving RoAS isn’t necessarily a good thing. By targeting people at or near a conversion, which Google knows based on search history or product page visits, PMax can report great efficiency yet undercut revenue. (Many would have purchased or converted anyway, so why even buy the ads or offer a discount?)

PMax achieves its RoAS score without buying more media, because PMax’s impression share is steady, according to Smarter Ecommerce data and an assortment of proxy metrics.

In Ryan’s words: “In the pessimistic case (right), which I endorse, efficiency is being over delivered. Which potentially means that revenue is being underdelivered. F*ck.”

Under The Influence

Influencer marketing has become a lot more professional over the past five years or so. But it’s still new and comes with real risks.

Bud Light is still reeling from an influencer deal, a Pride Week tie-in that included Dylan Mulvaney, who’s known for documenting her gender transition.

Three months after Mulvaney posted the video that sparked outrage (the content itself was innocuous; the reaction was primarily that Bud Light was endorsing a transgender influencer at all), beer stores in conservative areas are practically giving away Bud Light for free, but still can find no takers, The New York Times reports.

Meanwhile, skincare brand Bioré and an influencer got in hot water when she brought up anxiety attacks following a shooting at Michigan State University, where she went to school. (It was a mental health awareness month campaign.)

The Bioré influencer’s post would probably have created no drama if it wasn’t sponsored.

And, sometimes, influencers boil their own hot hater.

A group of influencers are facing scrutiny for a sponsored trip to a Shein factory in China, part of a Potemkin village-type effort to nix Western reports of Shein’s appalling labor standards.

AI Opener

Meta is offering a peek into the machine learning algorithms that power content distribution on Facebook and Instagram, The Verge reports.

Meta’s automated recommendation system collects public user content, including photos and Reels (its short-form video snippets), and looks at how users have interacted with similar interests or content.

Facebook and Instagram are also providing more tools for users to shape their personal content recommendation output and inputs. Users can take actions such as saving content, labeling things as “not interested” or choosing “Show more” or “Show less.”

And Meta will soon roll out its “Why Am I Seeing This?” feature to Facebook Reels, Instagram Reels and Instagram’s Explore tab.

A blog post by Nick Clegg, Meta’s president of global affairs, frames the info Meta shared as “part of a wider ethos of openness, transparency and accountability.” With regulators and the general public alike uneasy about how AI tech stockpiles, slices and dices user data, Meta’s move could be considered an olive branch or preemptive self-defense.

But Wait, There’s More!

Major news publishers want to team up to address the impacts of AI search and content generation technologies. [WSJ]

The FTC has updated its Endorsement Guides, which provide advice to businesses and others to ensure they use reviews and endorsements truthfully in ads. [release]

Google will remove Canadian news from search in response to a newly passed law that requires digital platforms to pay local news publishers. [Bloomberg]

Programmatic ads pose new brand risks amid the generative AI boom. [Digiday]

New investors bring Moloco’s valuation to more than $2 billion. [release]

You’re Hired!

Gannett names Chris Cho as its president of digital marketing solutions. [release]

Must Read

How Advertisers Can – And Cannot – Get In Front Of Chatbot Shoppers

Brands have plenty of ways to boost search visibility—paid, organic, and earned. But if a CEO demands presence in customer journey recommendation engines and is ready to pay, what can a marketer do?

Northbeam Adds The Third Leg Of The Attribution Stool With Incrementality Testing

There’s MMM and MTA, but no single ad measurement works for brands with multiple points of sale. On Tuesday, Northbeam launched an incrementality tool to complete what it calls “the trifecta of digital attribution.”

Comic: The Great Online Privacy Battle

What Regulators Talk About When They Talk About Ad Tech

If you want to know what privacy regulators think about online advertising, it’s not a mystery. Just listen to what they’re saying.

Privacy! Commerce! Connected TV! Read all about it. Subscribe to AdExchanger Newsletters

Keyword Blocking Demonetized More Than Half Of Reuters’ Brand-Safe Stories

The effect wasn’t just limited to news content. The Reuters.com/lifestyle vertical also had some of its brand-suitable pages blocked.

The Agentic Marketplace Is Here. Where Does That Leave DSPs and SSPs?

Swivel and Olyzon’s new partnership brings buy-side and sell-side agents together as early examples of an agentic marketplace.

Comic: Causal Meets Casual

Jones Road Beauty Is Using A New Type Of MMM To Reset Its Media Measurement

Inside how Jones Road Beauty is trying to turn messy, conflicting measurement signals into a single testing roadmap for its media mix.