US Department of Justice sues Google, seeks to break up its ad tech business

On 24 January 2023, in what Assistant Attorney General Jonathan Kanter described as a “historic moment”, the US Department of Justice (“DoJ”) filed an antitrust lawsuit against Google in the US District Court for the Eastern District of Virginia over Google’s ad tech business (the text of the complaint is available here). The lawsuit, joined by US States such as the State of California and New York, seeks structural relief, that is a breakup Google’s ad tech suite, including a divestiture of Google Ad Manager, Google’s sell-side solution combining a publisher ad server (DoubleClick for Publishers or “DFP”) and an ad exchange (AdX).

The lawsuit comes after reports that the DoJ rebuffed an earlier proposal by Google to settle the probe by splitting part of its ad tech business and placing it under a separate entity under the Alphabet umbrella. It also comes nearly two years after a very similar lawsuit was filed by a coalition of US States led by Texas (the “State AGs complaint”; see our analysis here), which last summer made headlines for largely surviving Google’s motion to dismiss (with the exception of the Jedi Blue allegations, more on which below). And it comes among a flurry of ongoing and closed investigations into Google’s ad tech business, including in the EU, France, the UK, and Australia. To suggest that Google’s ad tech conduct is in the regulatory spotlight would be an understatement.

In this post I do not intend to summarize each and every aspect of the DoJ complaint; this is a lengthy and dense document, and doing so would require a long post that would exhaust the reader. Rather, I will provide some high-level thoughts, focusing on what is new in the DoJ complaint (compared to past findings of regulators, such as those of the French Autorité de la concurrence or “ADLC”) and discussing the case’s wider implications across the world. This post assumes some familiarity with the complex mechanics powering digital advertising.

High-level thoughts on the lawsuit

A disclaimer should be offered upfront, in that the trial judge may always take a different view than the DoJ on the facts or the law (and unsurprisingly, Google has already contested the complaint’s allegations). That said, the complaint looks strong on its face, and Google will have a hard time defending itself, particularly in light of the ADLC infringement decision (which Google accepted not to contest in return for a reduction in the fine imposed).

Having said that, here are my high-level thoughts on the lawsuit:

1/ The DoJ complaint paints the most complete picture of Google’s ad tech conduct to this date, presumably benefiting from extensive document requests, but also the findings of numerous regulators across the world. The lawsuit ties these findings together in a chronological manner to provide a compelling monopolization narrative, starting from the early days of Google Ads (then AdWords) selling search advertising, then moving to the “pivotal” moment of the DoubleClick acquisition in the late 2000’s, before documenting the numerous product changes introduced by Google during the last decade up until Google’s switch to a unified auction in 2019.

2/ To cut a long story short, Google’s overarching goal throughout the relevant period was (according to the complaint) to force as many transactions as possible to flow through its own ad tech products, with Google taking a cut of the ad spend at each step of the way (para. 266). The focal point of Google’s monetization strategy has been AdX, where it charges the highest revenue share, consistently around 20% for open auction transactions since 2009, whereas rivals charged only a fraction of that amount (para. 266).

3/ The lawsuit is largely aligned with the State AGs complaint. There are some differences in the complaints, although these are not necessarily significant:

  • The DoJ places greater emphasis on some auction programs which featured less prominently in the State AGs complaint – although this may not be a fair assessment of the State AGs complaint given the redactions. This is particularly the case for Project Poirot. Conversely, the DoJ places less emphasis on auction programs discussed at greater length in the State AGs complaint. This is the case for Reserve Price Optimization.
  • The DoJ discusses the Network Bidding Agreement of Google and Facebook (known as “Jedi Blue” from the State AGs complaint), but it does not analyse it as a form of collusion. Instead, it considers it part of Google’s scheme to thwart Header Bidding, an innovative technology that threatened to undermine Google’s market power. Reference is also made to Google’s unsuccessful attempt to convince Amazon to not invest in Header Bidding. Note that the EU opened a probe into Jedi Blue but eventually closed it earlier this year. I understand that the UK CMA’s probe is still ongoing. On the State AGs front, Judge Castel dismissed the complaint’s allegations on Jedi Blue.
  • The DoJ considers two of Google’s acquisitions as part of its anticompetitive conduct. This is the case for DoubleClick (acquired in 2008) and AdMeld (acquired in 2011). Both acquisitions had been cleared by regulators.
  • The DoJ is also seeking monetary relief (damages) as an advertiser using Google’s advertiser buying solutions.

What is new in the complaint

The bulk of the complaint is aligned with the findings of other regulators, including the uncontested findings of the ADLC. That is the case for, among others (1) Google’s policy of “tying” Google Ads demand to AdX, and in turn “tying” AdX to DFP, thus locking publishers into DFP and turning it to a quasi-monopoly; (2) Google using its control over publisher inventory (by virtue of owning DFP) to route transactions to AdX vs rival ad exchanges; and (3) Google engaging in a strategy to forestall the development of Header Bidding, which it viewed as an “existential threat”.

The complaint builds on the above by providing further details and citing extensively quotes from Google’s internal documents. My favourite quote (while being conscious it’s only an extract) is that of a Google employee remarking that the AdX Direct tag – an archaic method to call AdX while using a non-Google ad server – serves as a “concept for antitrust”, and thus was spared by Google despite initial plans to deprecate it (para. 160).

Now, in my view, the most important contribution or “novelty” in the DoJ complaint relates to Project Poirot (yes, Poirot, like Agatha Christie’s Hercules Poirot), an auction program Google allegedly launched in 2017 to “dry out” competitors engaged in Header Bidding. Project Poirot already featured in the State AGs complaint, but it is the DoJ complaint that discusses it in detail and places it in the context of Google’s strategy to thwart Header Bidding.

The value in the DoJ complaint’s discussion of Poirot is that it places into perspective Google’s recent shift to a unified auction, and the related introduction of uniform price floors (known as “Unified Pricing Rules” or UPR). While these have been examined by competition authorities, including the ADLC and the CMA, their analysis does not seem to have taken into account the existence of Poirot (which after all was secret). Seen in this light, UPR seems much more problematic.

In a nutshell, according to the lawsuit:

  • Google became increasingly concerned that spend from DV360 (its buy-side tool used by large advertisers) was driving Header Bidding’s growth and helping rival ad exchanges (para. 210). It came up with a strategy to direct DV360 ad spend to AdX in order to “dry out HB” ad exchanges (para. 213).
  • The initial proposed strategy was to block DV360 from buying on rival ad exchanges inventory offered by publishers Google believed were engaging in Header Bidding (para. 214). But experiments showed that this strategy would be harmful to Google, predicting a 30% loss of impressions and revenue for DV360 (para. 214).
  • Google came up with an alternative: bidding lower on inventory offered through Header Bidding, with a view to shifting transactions away from ad exchanges using Header Bidding and to AdX (para. 216). Launched in July 2017, Poirot worked by systematically lowering DV360 bids (by as much as 90% in some cases) to rival ad exchanges Poirot detected as using Header Bidding (hence the codename!) (para. 217). Poirot was quite effective, resulting in DV360 spending 7% more on AdX and reducing spend on most other ad exchanges (para. 224). Poirot shifted approximately $200 million of DV360 ad spend away from rival ad exchanges and towards AdX (para. 224).
  • Still, there was a big problem: publishers were using differential price floors to direct the flow of certain transactions to rival ad exchanges (paras. 232-236). Internal analyses suggested that this practice prevented AdX from winning transactions (and DV360 would purchase the impression on a rival exchange instead of AdX).
  • In response, Google decided to block publishers from setting differential price floors through its UPR policy. Google bundled the imposition of UPR with other changes to provide cover and make them more “stomachable”, according to a Google employee (para. 241). Internal documents identified AdX as the “winner” of the new rules (para. 242). Following UPR rollout in late 2019, AdX market share increased by 6%. Google’s “best guess” of the impact was an annual increase of $430 million in gross revenue for Google and $118 million in net revenue (para. 246).

Besides Poirot, the lawsuit is also very interesting for discussing how Google managed to replicate its “last look” advantage (removed in 2019 to cast Google in a pro-competitive light) by developing an algorithmic model to predict rival bids. According to the complaint, Google delayed giving up last look for months while it fine-tuned the “Smart Bidding” algorithm that would replace it. When launched, it fully offset the 30% drop in Google Ads’ revenue that Google expected from the loss of last look and turned an expected 10% drop in DV360 revenue into a revenue increase of 3% (para. 258). If true, this would be shocking, considering how at the time Google was already being probed by the ADLC and how it strived to portray its 2019 policy changes as “fair” (seems like regulators were right not to trust Google).

Remedies and implications

As mentioned at the outset, the DoJ is going for the maximum: structural relief, including a divestiture of Google Ad Manager. This makes me wonder whether the European Commission – traditionally wary of structural remedies in antitrust investigations, especially when foreign companies are involved – will feel emboldened and push for a similar remedy in its own investigation into Google (which is rumoured to be approaching the stage of a Statement of Objections). Guess we will have to wait and see. Other than that, the DoJ lawsuit will likely have wide-ranging implications. For one, it has added further credibility and legitimacy to the State AGs complaint, which until now Google had tried to discredit by baptizing it the “Texas” complaint or the “Paxton” complaint. Well, now Google is also facing the Feds, joined by States like California and New York. What is more, the scandalous quotes cited in the complaint seem to deconstruct the vision Google has put forward of operating ethical walls across its ad tech products and taking decisions to advance the interests of its customers. This narrative seems less convincing by the day.

Photo by Nik Shuliahin 💛💙 on Unsplash

Leave a Reply