Google Leveraged Its Scale to 'Kill' Rivals, According to Architect of Digital Advertising

Days 6-7 of the landmark antitrust trial against Google's alleged Ad Tech monopoly focused on Google's anti-competitive conduct, including a history of problematic acquisitions.

We heard a lot about Google’s acquisitions and other anti-competitive conduct during Days 6-7 of US v Google (the Ad Tech case). We’ll get to the testimony in today’s headline down below, which comes from an early architect of online digital advertising, Brian O’Kelley, who described that when Google saw threats to its intermediation of digital advertising, it “took hits” in order to “kill” its rivals.

But first, if you’ll indulge me…

There’s an old saying in the law. It goes: “When the law is on your side, pound the law. When the facts are on your side, pound the facts. When neither is on your side, pound the table.”

In this case, we’ve heard the government pounding the facts. Those facts are piling up, they are often coming straight from Google executives themselves, and Google consequently is having a hard time disputing them. Google, for its part, has pounded the law, focusing on pro-competitive rationales for conduct that they aren’t really disputing.

Google may also be trying to unscrew the legs from the table. The problem for Google is that the government’s table has many legs. The evidence of anti-competitive conduct against the company is now cumulative in its scale and often quite damaging in its scope.

Nonetheless, as poor as its deck of cards has looked, Google’s lawyering, lead by white-shoe firm Paul Weiss, has been top rate (no pun intended). I’m sure when this is through, Judge Brinkema will commend both parties for their excellent work, as did Judge Mehta in the Google Search case.

Since the very beginning of this trial, from opening statements, Google has been more surgical in its probing—at nearly every turn aiming to hit the key legal elements at play—trying to elicit testimony that plays up the level of competition Google faced, taking advantage of rivals’ statements to regulators describing the market as competitive, waging pitched and well-learned battles on objections, and opening the door to counterpoints to be used later on.

The government, for its part, has not gone into the nitty gritty of the law as much as Google, at least not as sharply and conspicuously. On the other hand, the government need not wield such precision tools in this case—its arsenal of evidence is overwhelming in its own right.

This is a case that will most certainly be appealed, no matter the outcome. Facing what many onlookers anticipate will be a loss, Google may be tailoring its strategy to leave behind legal issues to raise with a higher court.

With that in mind, it makes sense Google is fighting battles where it can, and drawing out evidence wherever possible with respect to competing notions of market definition, alternative explanations for undisputed conduct, and generally leaving some hints about what we’re going to hear when its case in chief starts next week.

I expect Google will continue to argue that “walled gardens” (like Facebook and TikTok) and in-app advertising are the future of the industry, that fights over open web display ads stemming from 2009 are in the past, and that this is moot. Recall in her opening statement on behalf of Google, Karen Dunn compared the Justice Department’s case to a dispute over a Blackberry.

More cynically, Google may be looking forward to a change in administration with the election two months away to see if the government continues to sail in this direction, or if Google will get off for time served - a la US v. Microsoft in 2001. (Bush drops plan to break up Microsoft, The Guardian)

This is not a jury trial (Google’s money made sure of that) and with the facts largely on the government’s side, it makes sense Google wants to fight a narrow, technical, legal battle.

For everyone else in attendance for this trial however, this sets up some repetitive testimony, particularly during patches of overwrought and dense cross-examination.

We still expect the government to rest its affirmative case by the end of this week and for Google to begin presenting its case next week. Below is some of the key testimony and key witnesses of the past two days.

“we should discuss this offline.”

Key Takeaways

1. Emails show Google considered Yield Manager technology like AdMeld to be “irrelevant” to Google, and that it acquired AdMeld with the intent to “park it somewhere.”

2. Google purchased potential rival AdMeld for $448.5 million, more than $100 million than its high-end valuation at the time of acquisition.

3. Google leveraged its scale advantage to "take hits" in order to "kill" rivals, according to early architect of digital advertising.

4. Google reiterated its privacy and security justifications for acquiring what the Justice Department considers potential competitors.

Witness: Neil Mohan, CEO of YouTube

Day 6 of the adtech antitrust case against Google kicked off with the testimony of Neal Mohan. Mr. Mohan is the Current Chief Product Officer of YouTube (an Alphabet/Google property) and the Former Senior VP of Display & Video Ads for Google.

Mr. Mohan joined Google from DoubleClick as a result of the 2008 acquisition of the ad tech firm by Google. By way of review, DoubleClick was an early technology for websites (publishers) to make their ad inventory (like banner ads) available for advertisements. In digital ad tech speak, it’s a “publisher ad server.”

Google’s acquisition of DoubleClick was a big deal, and today Mohan and the Justice Department put it into some context. Prior to Google’s acquisition of DoubleClick, Google saw weakness in its publisher ad server tech as an existential threat to its entire digital ad business. That motivated Google to acquire DoubleClick, and Google today has a 91% market share of the publisher ad server market for advertisements across the open web.

The Court was shown a Google slide deck covering the acquisition, referencing the ad exchange as an “operating system” with “high switching costs.”

A 2010 document which Mr. Mohan said he believed contained Google meeting notes, described the strategy to acquire DoubleClick as not being done for “revenue,” but to enable the broader ad exchange and “backfill strategy.”

A subsequent exhibit showed a 2009 email, which interestingly described Google’s perception of strategic challenges with respect to Yahoo, Microsoft, AOL, and others that had better “publisher platforms.” It may seem quaint now, but in the very early days of YouTube, when websites like MSN, AOL, Yahoo, and others were still successful digital publishers, it is interesting to see Google’s perception of the marketplace. Of course, in the case of MSN, Yahoo, and AOL, all of these businesses today are shell of their former selves.

This Google email showed executives describing this challenge as “the most strategic battle,” in terms of having a strong publisher platform, and in turn, content to monetize.

Mr. Mohan was asked if customers being free to choose among competitors wouldn’t make Google enough money, to which he said he disagrees with “that characterization,” a common demurral we’ve heard from Google executives.

Subsequent email exhibits touted the importance of the GCN/AdX bundle, in other words, what the government alleges to be illegal tying.

Mr. Mohan said he disagreed that “bundling AdX w/ buy side team” would be a “HUGE conflict perception in the market,” as was alluded to in an email of his, written at the time. This continued a trend of Google witnesses being impeached with their own emails and prior statements.

Some of these emails contained a trifecta of Google characters familiar to us by now: Brad Bender, Jonathan Bellack, and Mr. Mohan himself.

A quick note: I do wonder about the timeframe of the exhibits we’ve seen. I’ve kept a rough tally in my notes as the trend has become apparent of the Court being shown such a volume of exhibits from 2009, with a number from 2008, 2010, 2011, and 2018-19, with what seems to be a handful thrown in from the interstitial years.

While evidence is evidence, technology is a fast-evolving business. We’ve heard arguments that the future of digital advertising is in walled gardens, not open web display ads, and as the conduct at issue drifts further into the past and comprises a smaller percentage of how Google makes its money, with each new 2009 exhibit, the trend starts to stand out a little.

Subsequent emails showed a hesitancy on Google’s part to allow non-AdX exchanges to use APIs, and worries about yield managers.

Next the government got into the Admeld acquisiton. Admeld offered website publishers yield management software. Yield managers better predicted the prices publishers could expect to receive from advertiser ad networks and other demand sources.

While Google (and Mr. Mohan) wanted to frame Google’s acquisition of Admeld as merely complementary to Google’s ad products, and not an effort to subsume a rival, emails show Google considered Yield Manager technology to be “irrelevant,” and that Google merely wanted to take over Admeld to “park it somewhere.” Bellack adds this would let them “solve the problems from a position of strength.”  

The Admeld acquisition was a big deal for Google. We saw evidence that its valuation at the time of acquisition was $182M - $355M. Google ended up paying $448.5M.

Then came Bill Isaacson, a Google attorney, for cross examination.

This would be a good place to discuss Google’s strategy. It might sound surprising, but this trial has flown by. With the government’s case entering the home stretch, it would be a good idea to take stock of where Google stands in all of this. While many expect Google to lose this case, and indeed, the evidence against it is staggering, Google will have its say too, and make, quite literally, the best arguments money can buy.

On cross, Mr. Isaacson worked in notions that Google was worried about publisher quality standards (i.e., an alternative explanation for Google conduct) and mentioned privacy and spam concerns.

This has been an interesting thread woven throughout this trial—this specter of “privacy” or “spam” or “security” concerns among publishers and advertisers. This is not to say this concern was not legitimate. My issue is that it comes in typically only as an afterthought—a sort of talisman to ward off accusations of anti-competitive conduct the way other tech companies plead “cybersecurity” to try and cast off antitrust arguments against them—re-framing a competition problem into a security problem.

Google presented an email describing AdMeld and Criteo as being “plagued” with ongoing issues, apparently to undermine the narrative of vibrant competitors being done in by Google’s monopoly machine.

On redirect, Mr. Mohan, sticking to the script as ever, would not acknowledge Admeld as a competitor—bizarrely referring to them as being “parallel” with Google, whatever that means. (Does Google have a “Communicate with Care” training for how to speak in federal court?)

Mr. Mohan was similarly vague as to whether Facebook was a competitor, despite the fact Google has spent the entire trial working hard to label Facebook as squarely that.

Witness: Brian O’Kelley, Co-Founder of AppNexus, early architect of digital advertising

Later, Brian O’Kelley, Co-Founder and former CEO of AppNexus, gave some of the most damning testimony yet, via video deposition.

He discussed how in an auction where you can’t see other bids, you’re more likely to give close to your best bid whereas with Last Look, Google can see the highest bid and merely bid a penny more to lock up transactions and snuff out rivals for literal pennies on the dollar.

Mr. O’Kelley discussed how threatened Google felt by header bidding, how it blocked OpenRTB (Open Real-Time Bidding) from becoming adopted as a standard through leveraging its financial contributions and control over the IAB (International Advertising Bureau), and that “Google felt very threatened by real competition.”

Some of the best testimony came when Mr. O’Kelley expressed that Google could - and would - lower its own rates and revenue share and “take a hit,” exploiting its massive scale in order to “kill” rivals, upon which time it could return to economics more favorable to it—classic monopolist behavior.

When (who I imagine to be Google’s lawyer, it’s difficult to tell because the deposition video was full of jump cuts) took his turn, he referenced AppNexus’ S1 filing with the SEC in which it described the market it was participating in as “intensely competitive.”

This has been an issue in this trial multiple times, with multiple companies, where Google’s lawyers jump at the opportunity to pull out regular SEC filings (or “statements to regulators made under the penalties of perjury,” as Google would frame it) and pound language where the alleged Google rivals describe their markets as competitive as evidence that the markets are in fact competitive, and not dominated by Google.

This tactic has had mixed success. If anything, the language in these filings is pro forma boilerplate from corporate lawyers. But it certainly doesn’t help for a purported Google market victim alleging to the Justice Department they’ve been squashed by Google to tell their investors (and the SEC) how fiercely competitive the market is on the other hand.

Witness: Rosa Abrantes-Metz

Ms. Abrantes-Metz, a PhD economist at the Berkeley Research Group, delivered some standout testimony. She attested that by “restricting customer choice, Google was able to impair rivals from competing” ensuring Google’s market power, “leading to higher prices.”

The government preempted a regular nitpick of Google’s counsel on cross examination, the extent to which expert witnesses did or did not do quantitative analysis on the issues they are testifying to, by asking Ms. Abrantes-Metz if she has done quantitative analysis in this case.

She responded that she did not even need to due to how apparent Google’s monopoly power is, and went on to discuss Google conditioning (or tying) of its ad products, saying that “there’s a wide amount of evidence related to publishers not liking this tie” and that “rivals cannot compete because of [it].”

A government exhibit, PTX 758, was an email between executives at Adzerk (now Kevel) describing Google’s bundling, or tying scheme as being “impossible to break.”

To hear more about the testimony of Rosa Abrantes-Metz, and Gabriel Weintraub, Professor at Stanford , check out usvgoogleads.com.

---

Witness: Nirmal Jayaran

Picking up on Day 7, the day began with direct examination of Nirmal Jayaran, a Senior Director of Engineering at Google.

An exhibit showed a report prepared by Mr. Jayaran forecasting what would happen if Google allowed its advertisers on third party exchanges in a variety of revenue sharing scenarios.

Indeed, Mr. Jayaran prepared many reports and conducted numerous experiments testing Google’s revenue under a variety of competitive scenarios. These exhibits comprised a significant amount of his testimony, which dominated much of the day.

These exhibits presented Mr. Jayaran’s findings, that Google would face some measure of a loss in revenue when publishers could choose to make their inventory available to exchanges other than Google.

Mr. Jayaran was another in a series of Google witnesses who frequently claimed to be unable to remember things when they were problematic for the company, even to the point of absurd results.

For example, Mr. Jayaran attested that he did not agree with conclusions reached in some of his own reports. In another instance, when presented with a report stating that Google’s DV360 should directly participate in header bidding and “pass on the gains to the advertiser,” Mr. Jayaran said that he does not “completely agree.” The government would then routinely layer in to their questioning that Mr. Jayaran’s name appeared, usually as the author, of the documents in question to which he was distancing himself.

This pattern dominated much of the morning.

PTX 363, which has not yet been publicly posted by the Department of Justice as of the time of this writing, in my view is one of the most damning exhibits yet for Google.

It was a slide titled “Header Bidding Benefits Pub[lisher]s and Advertisers, Challenges DRX (Google DV360),” and came from a Google slide deck.

In the slide, Google listed as “pros” of header bidding for publishers, “universal competition” and the lack of “Google policies and rev share”.

It listed as a pro for buyers “rapid ungoverned innovation” and listed as challenges for Google that header bidding creates a “new auction outside our decision logic, policies, or fees.”

On the bottom, the slide said that header bidding could lead to a “competing ad platform.”

In one slide, Google’s internal deliberative process was encapsulated—expressing concern for “rapid ungoverned innovation,” “auction[s] outside [Google’s] decision logic, policies, or fees” and the threat of a “competing ad platform.”

In subsequent exhibits, Google discussed putting “pressure on companies providing header bidding” and considering “levers” that could “be pulled.”

All of these exhibits paint the picture of a dominant firm with engineers able to, quite literally, pull levers to affect the market and suppress rivals, nearly on command.

Mr. Jayaran continued to “disagree” with his own prior statements in exhibits frequently. His lack of memory with respect to events the government sought to question him about, while simultaneously having excellent recall on cross examination when Google was doing the questioning, persisted. Of course, Mr. Jayaran would either have to say he agrees with anticompetitive conduct depicted in his own reports and Google exhibits—which could then be imputed to Google itself—or instead contradict his past statements in an effort to put distance between them and Google. He did the latter.

I wrote in my notes a question of why the government did not try to pull on this quite obviously coached witness strategy or highlight the clear contradictions. From listening intently to Mr. Jayaran, I felt that if he were challenged on his contradictions, he was the type of witness who would probably struggle to explain himself when he’s forced to think on his feet and deviate from Google’s script—indeed he admitted to having prepared with Google counsel about six times for his testimony.

When Mr. Jayaran responded to DOJ questioning time and again with “I would not characterize it that way,” I can’t imagine he’d be able to keep this act up for too long if asked “Well how would you characterize it…”

Later Ms. Wood, the lead government lawyer, did some of this, asking Mr. Jayaran if he felt at the time of his statements that his views were accurate.

Ultimately, the trend of Google witnesses having faulty memories under government questioning is quite conspicuous and speaks for itself. Judge Brinkema is a seasoned jurist responsible for evaluating the credibility of witnesses in this bench trial, and has no doubt encountered all manner of witness performances before, and knows what’s going on. Pressing further on it may not be worth the government’s time.

The government then returned to a familiar problem for Google—the abuse of privilege. Mr. Jayaran was questioned about documents where he wrote “PRIVILEGED AND CONFIDENTIAL” on the top of them, and admitted under questioning that no legal advice was in fact being sought, and no lawyers were copied on the email. This strategy was intended to get documents discussing “sensitive issues” particularly issues of competition, put on a privilege log and excluded from discovery, even when those documents were not truly subject to privilege and should not be withheld.

Another striking exhibit from Google expressed an internal view that two of the most common reasons to discontinue Last Look is that it is “perceived as unfair in the market” and the acknowledgement that “we have to give this up at some point.”

Some of the most eyebrow raising testimony involved an experiment Mr. Jayaran conducted, where Google, in October of 2016, switched off its flow of ads to domains identified as using header bidding, in order to collect data on what would happen from a financial standpoint if Google were to turn its levers at a much broader scale. This experiment resulted in about $20,000 of lost revenue. Google apparently felt no reservations about conducting costly experiments on its clients. These exhibits continue to reinforce the image of a company so powerful that it is able to control market forces so readily.

On cross examination, Google hit its usual notes—putting a smokescreen over market definition by layering in analysis of Facebook, Yahoo, AOL, Twitter, and MSN, and mentioning brand safety, fraud and spam. Fraud and spam have become this sort of free space on Google’s cross examination bingo card—they check it off with every witness by eliciting testimony that concerns for this issue were part of Google’s decision-making process, even though we have not once during this trial been told what types of fraud or spam are at issue here.

Witness: Tim Cadogan, CEO of OpenX

Next, we heard testimony from Tim Cadogan, CEO of OpenX. Direct examination revealed much of what other Google competitors (or would be competitors) have attested to—the inability to overcome Google’s massive inventory and scale, its lack of responsivness, being unable to grow because Google’s structural advantages denied OpenX the ability to generate the revenue it needed to retain its employees or respond to what customers needed.

There was not much new here—evidence of Google’s conduct has become cumulative.

Cross examination however, was electric. Mr. Cadogan stood his ground when Google’s counsel peppered him with what are by now familiar lines of attack—pulling out old pitch decks and trying to hammer the witness by papering over the Google conduct elicited on direct examination by trying to impeach the witness with what seem to be minor inconsistencies—old SEC filings describing a competitive market, pitch decks painting rosy pictures to clients, etc. Mr. Cadogan was having none of this, politely firing back at Google’s counsel that people “pound their chest” in pitch decks and that he had “found a needle in the haystack” in drawing out seemingly inconsequential lines from mountains of discovery that could be used to try to hit him with.

I winced a few times when the witness became indignant with Google’s counsel, expecting Judge Brinkema to rebuke Mr. Cadogan for his lack of decorum (in this case, simply answering yes or no on cross and not becoming headstrong with the attorney) but she largely did not, an apparent indication she didn’t see a problem with Mr. Cadogan roughing up Google’s lawyer, albeit in a largely good-natured way.

We finished the day with the remaining 9 minutes of former AppNexus CEO Brian O’Kelley’s deposition video, as well as a read-in from the deposition of Jeremy Helfand, former EVP of Ads and Data Platforms at Disney, and a read in from Susan Schiekofer, Managing Director of Digital Investment, GroupM. There was not much here. If anything, Mr. Helfand’s deposition read-in was interesting because it gave the rare testimony of a company that given its standing as a media heavyweight with the money to compete with Google (albeit in a different way, Disney is not in the business of selling open web display ads) was able to find success.

That’s all for now.