Trial Update, September 16: Google’s power on full display as we learn IAB Tech Lab passed on Prebid to appease their largest funder

Week 2 began with long-time (understatement) Google employee Neal Mohan.

‍

Neal Mohan, CEO of YouTube

Previously, Neal was at NetGravity, which was acquired by DoubleClick. He then came over to Google with the DoubleClick acquisition, holding a number of product-side roles until he went over to YT in 2015.

We start with the DoubleClick acquisition, which emails reveal was bought for “enabling the exchange and backfill strategy,” as part of a 3 pillar strategy to own (1) inventory access via ad server, (2) aggregation via exchange, and (3) monetization via ad network (formerly GCN).

The “most strategic” battle, however, was seen as the ad server, as losing share would risk their disintermediation even if they had the best ad network in the world.

Neal was very clearly coached. He reiterated how much “choice” publishers have in the “two-sided” adtech market. But emails and docs told a different story. “I do think we need tight XFP (DFP) - AdX/GCN (exchange/network) bundles,” read an email.  In another, Neal talks about how AdX needs to be kept out of the buy-side teams b/c it would pose a “HUGE conflict perception.”

We get into the AdMeld acquisition next.  He insists it was complementary, but emails show Google considered Yield Manager (YM) tech to be “irrelevant” to them, and wanted to buy AdMeld to “park it somewhere,” per Mohan’s emails. Bellack adds this would let them “solve the problems from a position of strength.”  Valuation was $182M - $355M, yet all-in cost (est.) was $448.5M.

Cross-exam focused on landing what I can only assume were the coached buzzwords.  I lost count of references to innovation, 2-sided market, and inventory quality, and Mohan a few times insisted display is inclusive of mobile, video, etc.  The defense used many external-facing documents including one talking about a white paper distributed at a Google booth at an IAB annual conference.

Judge Brinkema rightly realized the paper focused on higher yield vs. non-dynamic allocation - NOT vs. allowing publishers to get real-time pricing from other exchanges in DFP.  DOJ on redirect also pointed out that the defense used external (i.e. sanitized) documents to make their points vs. internal strategy discussions that they referenced in their direct examination.

Mohan also explained how costly and time-consuming integration efforts were, and that dynamic allocation was something Google had to build - the implication being that to allow other exchanges to participate in RTB, Google would’ve needed to invest in building new capabilities.  Neal seemed to imply nobody was doing RTB at the time, but DOJ clarified this by asking if he is saying that Google invented the capability. Similarly, he said that Pubmatic and Rubicon didn’t run real-time auctions, playing a bit of a game with the timelines. When DOJ asked  him to confirm that either, at no time before he left for YouTube, ran real-time auctions, he said he didn’t know. Naturally, the DOJ also called back to the email we saw during Rajeev Goel’s testimony asking about Pubmatic access to the DFP API, which Neal forwarded internally saying publishers should just use AdX.
‍

Brian O’Kelley, CEO of Scope3, former CEO of AppNexus

We then get back into Brian O’Kelley’s video deposition. IAB makes its second appearance of the day as O’Kelley’s deposition tape continues. He explains that IAB Tech Lab rejected taking over prebid after Google “vehemently objected” to it at a board meeting, Google being the largest financial contributor to IAB at the time. An aside: surprising to nobody, IAB, IAB Tech Lab, and the ANA have all been markedly silent publicly on the ongoing trial, illustrating the breadth and depth of Google’s industry capture.

Apart from this, his testimony echoed what other market participants have stated about the impact of Google’s conduct. He explains how Header Bidding created competition and improved publisher yield, but was bad for Google. He spoke to how GAM didn’t allow for other exchanges to participate in dynamic allocation or real-time auctions at first. O’Kelley explains that AppNexus declined to participate in Open Bidding because Google set prescriptive rules like barring unique demand, i.e. rival DSPs, charged a 5% fee, and had full visibility into the auction data that they could use to court or poach customers. He figured that Google would just change the rules again anyway when it suits them, and AppNexus wanted no part of being beholden to Google’s unilateral whims.

His explanations were clear and concise, and if nothing else, reinforced the practical impact of Google’s conduct. For example, he explains what Last Look does to the auction dynamics by saying “wouldn’t it be nice” if you could know what everyone else bid, and just bid a tiny bit more. He explains how this also let Google play games with price (e.g. Sell-Side DRS), boost their win-rates, and “tell stories” to agencies and advertisers.

‍

Gabriel Weintraub, Professor at Stanford (Expert Witness)

Professor Weintraub is an economic expert that examined the impact of a number of Google’s conducts using Google’s own experimental results, monthly data on impression volume, and auction-level data from June ‘23.  The conducts he evaluated are (1) First Look, (2) Last Look, (3) Sell-Side Dynamic Revenue Sharing (as an enhancement of Last Look), (4) Unified Pricing Rules, (5) Poirot, and (6) Exclusivity.

I won’t walk through each one here, but for example, he explained that First Look decreased publisher revenue and decreased match quality for advertisers, he explains, alongside quantitative analysis that said it decreased rival share by 24.6% compared to a hypothetical “head-to-head” competition scenario. He similarly explains how SSDRS leveraged Last Look to change fees and increase AdX win rate, resulting in -$161.6M in ad spend annually for rival exchanges - roughly equivalent to the annual ad spend going through Equative in a given year.

Google’s cross-examination lasted a very long time. They worked to call Professor Weintraub’s methodology into question, and to make it seem like the slides accompanying his testimony omitted calculations less favorable to the government’s case. While much of this was clarified at re-direct, regarding methodology, Judge Brinkema did say at one point toward the end that by not relying on one specific Google experiment, but instead looking at median effects “lessens reliability” of that particular calculation.

Rosa Abrantes-Metz, Managing Director at Berkeley Research Group (Expert Witness)

The day ends with another economic expert. Dr. Metz is an expert in industrial organization and financial economics, that has specialized in platforms, monopolization, valuation, and several other areas. She was asked to determine if conducts were exclusionary, and if they harmed rivals, users, and market participants (i.e. advertisers and publishers). She concluded that Google’s conducts led to impairment and harm to competition in all relevant markets, while protecting Google’s own products from competition. She also concluded the conducts limited customer choices, lessening competition, enhancing Google’s market power, and leading to higher prices.

The conduct that she looked at is: (1) RTB exclusivity to AdX via DFP, (2) Google Ads exclusivity to AdX, (3) First Look, (4) Last Look, (5) AdMeld acquisition, (6) UPR.  She frames the issue a bit by saying that some conduct can harm rivals, but not competition - like lowering prices; however, in these examples, Google restricted choices of their own customers.

We began the walkthrough of each conduct today, with Google’s tying of AdX and DFP, as well as AdX to Google Ads. Her testimony will resume tomorrow, but not until after the 9AM slot where Google engineer Nirmal Jayaram is slated to testify.