To divest, or not to divest.
During day two of the remedies phase of the Department of Justice (DOJ)’s ad tech antitrust trial, both sides underwent a grueling back-and-forth about whether a divestiture of Google’s publisher ad server and ad exchange products is necessary – or impossible.
Judge Leonie Brinkema, the US District Court Judge presiding over the case, stepped in to take a firmer hand in steering the conversation.
“I’m going to give you a hint,” Judge Brinkema told Google during its counsels’ cross-examination of The Trade Desk CRO Jed Dederick.
Rather than the parade of ad tech execs who have taken the stand so far, she said she’s waiting to see what Googlers and other first-hand authorities think is required for an AdX or DFP divestiture.
“This is to some degree window dressing,” she said of rehashing the ad tech debates. “Let’s get to the mannequins.”
A short time later, she halted a question and response in regard to a line of questions about header bidding.
“I remember that from last time,” she said. “Everything was all excited about it, and then … ” before trailing off to silence, indicating Google’s efforts to step in and squash the tech.
Google’s disputes
Google’s defense seems to rest on a couple of basic arguments: that the open web display ad market, where it maintains an illegal monopoly in publisher ad tech, is going the way of the dodo; and that no other company is a feasible acquirer of AdX and/or DFP.
To prove that open web display ads are on the way out anyway, Google questioned Omnicom’s Luke Lambert, a leader of its Confluence Agency, which is a one-client shop set up for Amazon, about an article where he said that a product like AdBridge, which auto-generates creative and audio ad formats, was “the future.”
Jay Friedman, until recently CEO of Goodway Group, also testified from the agency perspective. He was questioned about his own byline – an article in which he’d described the growth of generative AI as erasing open web display ads (if, say, someone uses an AI agent to do their shopping, rather than browsing the web themselves).
“This is a thought experiment,” he said. Not a sign that banner ads are actually going to disappear in the next few years.
On day one, Google had similarly tried to ensnare Index Exchange CEO Andrew Casale, who said generative AI had not affected competition in ad tech markets, despite having boasted of AI’s growing importance in Index’s blog posts and marketing.
Kevel CEO James Avery testified that his company would be interested in acquiring the AdX and DFP package if it hit the market. But also that Kevel processes some 20,000-100,000 ad requests per second. Is Kevel going to take on an ad server that does some eight million ad requests per second?
Which is apparently the number of ads processed on average per second by DFP (now called the Google Ad Manager ad server, not DFP).
Google’s counsel also noted throughout that Google offers DFP for free to publishers. Would any potential acquirer be able to maintain that standard? Or do hundreds of thousands of small sites and publishers suddenly start paying for something that’s been free?
Friedman responded to one such question by citing a line used in the ad tech space, to the point that “when it’s free, you’re the product.”
Justice Brinkema liked that one. And when pressed by Google’s counsel he noted that Google does not make DFP free as a goodwill gesture. There must be some inherent value that is not clear from the outside, but which makes sense (and makes money) for Google.
Google’s counsel also insinuated throughout that the DOJ’s witnesses are pursuing their own wishes and their client’s wishes, rather than the truth.
Lambert, for instance, who was chief innovation officer of the Omnicom agency OMD when he first testified last year, now co-leads the Amazon Confluence shop.
“You spend every second of your professional life thinking of what’s best for Amazon, is that right?” Google’s counsel pressed on him.
“No,” Lambert said.
The DOJ angle
The government’s case is built on the supposition that Google must sell off its publisher ad tech, because any connection the company maintains between its buy- and sell-side products creates incentives for more nontransparent supply chains owned by Google.
Even if Google does bid into Prebid, for example, which is one concession on the table, there’s no guarantee Google would do so on equal footing. And it would still bid directly between buy-side products like AdWords and DFP.
Publishers still wouldn’t switch their tech in that case, all the experts agreed, because they would suspect Google was preferring its own pipeline and passing better bids through that channel.
Whether Google can be trusted to uphold the letter of the law when the wording is vague is an important question at hand.
In Google’s proposed remedies, for instance, the company puts all PMP deals, or any campaign involving a deal ID or prenegotiated terms, like programmatic direct, under the “direct advertising” umbrella, more akin to old-school IOs. Google’s counsel keeps trying to get the expert witnesses to draw a line between old-school direct deals and programmatic deals that involve direct negotiation between a publisher sales team and a brand.
The DOJ must also prove that the downturn in open web display ads in recent years isn’t fated to continue.
Could the DOJ’s remedy proposal reverse the “anemia” in open web display? That was the question put to Dederick, who had used “anemic” to describe the category’s lack of innovation in recent years.
The supply exists, he said. “So if advertisers regain trust (in open web display inventory), we think ad budgets would float into open web display.”
Well, how do banner ads regain trust when they are still just banner ads?
For one thing, he said that “the sense among buyers is that Google’s sell-side tools are manipulated to prefer Google,” Dederick said.
Google’s counsel tried to have that one struck from the record, but unsuccessfully.
Dederick wasn’t the only buyer speculating further about Google’s untrustworthiness.
“I don’t have evidence,” Friedman said, carefully noting that the word “evidence” comes with a high bar for provably demonstrating something. “But I have to believe [Google] is using signals from DFP to inform buying or other tools.”
Google tried to strike that comment, too, also unsuccessfully.
On the one hand, Google’s auctions are manipulated. The evidence poured out during discovery and the trial phase last September supporting that assertion. But even with the sunlight from that trial, there remain black-box channels within Google and wide distrust among advertisers, who also trust their gut feelings. For years, many publishers also had a bone-deep conviction that Google’s sell-side auctions were self-preferencing in some way that wasn’t clear, and weren’t being forthrightly presented by Google itself.
Those paranoid suspicions were completely vindicated.
A Google AdX and/or DFP divestiture wouldn’t be easy.
There would be immense “pain of divestiture,” Lambert said.
However, that pain mainly takes the form of lots of added work. The agency would need to contract new SSPs and inventory suppliers to fill the gap in demand and try to get campaign pacing back to the standard agencies have with AdX, with 90% of the web’s inventory.
But, for the sell side, and even for the agency to a degree, “pain is opportunity,” he added.
His description of the “pain of divestiture” is that the agency would urgently need other SSPs to step up and fill in. The agency would also work directly with more publishers to directly act as a stopgap and help meet the demand.
In other words, Lambert’s “pain of divestiture” sounds like a great thing for other SSPs and sell-side players, who would also urgently be stepping up to replace AdX supply.
Friedman surprised the counselors by saying it wouldn’t matter if AdX disappeared entirely.
People wouldn’t stop browsing the web, and publishers wouldn’t stop generating those ad units.
“If it’s gone, it’s gone,” he said. “Plenty of market participants would fill back in.”
