from the pay-attention-to-this-one dept
There are lots of different antitrust actions currently ongoing against Facebook and Google, with varying degrees of quality. From the beginning, the strongest one has been the lawsuit a bunch of states — led by Texas — filed against Google. When it was filed, I noted that there were sections that were heavily redacted which had the potential to be pretty damning, but the redactions made it hard to tell. I also found some of the non-redacted bits questionable, as they suggested a complete misunderstanding of some aspects of the technology. Last Friday, Texas filed its third attempt at a complaint and it reveals a lot more about the stuff that was redacted in the earlier filings — and I’ll now say that this is the most serious, and the most damning, of all the antitrust lawsuits out there. How Google responds to the lawsuit will be extremely interesting and worth watching. Given the errors in the original filing, it’s possible that there are errors here too, but if what Texas is alleging in this latest version of the lawsuit are accurate, then Google should be in trouble.
The filing is 242 pages long, and much of it gets deep, deep into the weeds about the online advertising market. But there are two key parts that stand out as potentially very important and very damning for Google (and, partly, for Facebook as well). Most of the news coverage has focused on the claim that Sundar Pichai and Mark Zuckerberg signed off on a special deal to combat header bidding — a form of getting around Google’s position in the ad market. When the header bidding claims first came out, I found them questionable, as the early claims suggested that Google was abusing its position to kill off header bidding, which seemed ludicrous to me, because I’m contacted probably half a dozen times a day by ad companies offering their header bidding solutions for me to use. But, the details now revealed suggest Google really did work with Facebook to try to suffocate the header bidding market:
Facebook understood these stakes as well. Internal Facebook communications indicate
that Facebook?s March 2017 announcement was intended to signal Facebook?s willingness to
support header bidding. Facebook knew that Google would see its participation in header bidding
as a major threat. Evidently, Facebook was executing a planned long-term strategy??18 month
?header bidding? strategy to minimize ?[the Exchange Bidding] tax??by threatening to expose
the hidden costs Google charges publishers.
Google and Facebook entered into formal negotiations shortly thereafter. Both sides
recognized that Facebook?s leverage came from its critical role in supporting header bidding. As
a Facebook document from February 2, 2017 memorialized, ?What Google wants: To kill header
bidding (us baptizing their product will help significantly).? Elsewhere, Facebook employees
summarized an earlier meeting between the parties discussing header bidding and Exchange
Bidding (?EBDA?), ?We discussed the EBDA product they?re building. Both parties (FB and G)
were candid about why header bidding exists and that EBDA?s sole reason for existence is to kill
it.? In an October 30, 2017 email, senior Facebook executive [REDACTED] discussed the proposed
Google-Facebook agreement and explained to another Facebook executive, [REDACTED]
, ?they want this deal to kill header bidding.? Google put the
matter just as bluntly, explaining internally in 2017 that the goal of partnering with Facebook
would be to ?protect? Google?s ?leadership position in 3P [third party] ad buying/selling.? To that
end, the endgame with Facebook was to ?collaborate when necessary to maintain status quo.? The
?status quo,? in this case, was an unlawfully obtained ad server monopoly and an ad exchange
charging many multiples over the competition.
As negotiations proceeded, Google began to accept that Facebook?s price for
abandoning header bidding would require Google to share some of the auction advantages it had
previously taken for itself. In an August 9, 2018, internal Google presentation, one slide averred
that if Google could not ?avoid competing with FAN? in the trade for third-party inventory, then
it would instead collaborate with Facebook to ?build a moat.? Google thus preferred to share a
slice of its monopoly profits with a potential entrant rather than risk reducing its monopoly power.
The prospect of cooperating rather than competing with Google was enticing for
Facebook too. As internal Facebook documents reveal, Facebook believed that partnering with
Google was ?relatively cheap compared to build/buy and compete in zero-sum ad tech game.?
Facebook identified ?build/buy ad tech? as the company?s second-best option but noted that
entering the market would have required ?huge [engineering] and services investment, and
patience for sales cycle.? Compared to the time and expense of building a new technology and
competing on the merits, entering an unlawful deal with Google not to do those things was an
Facebook?s Chief Operating Officer [REDACTED] was explicit that ?[t]his is a big
deal strategically? in an email thread that included Facebook CEO [REDACTED] When the
economic terms had taken their form, the team sent an email addressed directly to CEO
[REDACTED]: ?We?re nearly ready to sign and need your approval to move forward.? Facebook
CEO [REDACTED] wanted to meet with COO [REDACTED] and his other executives before
making a decision.
The ultimate outcome of these negotiations was a September 2018 Google-Facebook
agreement signed by Philipp Schindler, Senior Vice-President and head of Google advertising
sales and operations, and Ms. [REDACTED] Facebook?s Chief Operating Officer and member of
Facebook?s Board of Directors, who herself was one-time head of Google advertising. Google
CEO Sundar Pichai also personally signed off on the terms of the deal.
Two quick asides before getting into the meat of this: First, I have no clue why the obvious names of Mark Zuckerberg and Sheryl Sandberg are redacted here. They are the CEO and COO of Facebook and there’s no mystery in that. It seems like a really bizarre choice to redact them when their titles are named, and even Sandberg’s former job is mentioned. Second, (and maybe this explains the redactions?), it sure feels like a similar suit about this very same deal may now be on the horizon for Facebook.
As for the substance, again it will be interesting to see Google’s response on this, but from this description, it sure looks like pretty garden variety anti-competitive behavior that involved two giants in the marketplace working together to limit the impact of a system that would chip away at the market they mostly controlled. If this is what happened and this is what the deal was about, then this lawsuit looks entirely legit.
The second really interesting bit in the case is the part I would argue is even more damning against Google, and if it’s shown to be true, I hope that Google is significantly fined and banned from this practice. Politico’s antitrust reporter Leah Nylen did a wonderful job breaking down the claims about how Google’s “Project Bernanke” supposedly worked and if this is what Google did, it sure sounds like straight up deceptive practices, bordering on fraud — and Google should probably need to pay out a lot of money to publishers who had Google ads on their site at the time.
At issue are the different types of auctions, and Google is accused of misrepresenting what kind of auction it was running. As the lawsuit explains:
By way of background, ?first-price? and ?second-price? auctions are common types of
auctions used in various industries and contexts. Generally speaking, in a first-price auction, the
buyer pays the amount of their own winning bid; and as the name implies, the buyer in a secondprice
auction pays the amount of the second-highest bid (sometimes with a negligible additional
amount, e.g., one penny). A ?third-price? auction, therefore, is one in which the buyer pays the
amount of the third-highest bid. As addressed below, Google?s secret Bernanke program
surreptitiously switched Google?s AdX exchange from a second-price auction to a third-price
auction on billions of impressions per month.
This may sound a bit confusing, but as Nylen points out, in a first price auction — the kind most people usually think of as an auction — the highest bidder (“the first price”) is what is paid. As Nylen says: “So, if Henry bids $19, Alice bids $18 and Scott bids $9 ? Henry wins and pays $19.” A second price-auction or a Vickrey auction means that in the above scenario the winner pays the second highest bid — i.e., Henry wins, but pays $18. This is the way that eBay auctions work, and some argue it provides a more efficient solution for getting bidders to bid the real value they put on the item up for bid.
From here, it’s not that difficult to figure out what a “third-price” auction means — in the above scenario it would mean that Henry wins, but only pays $9. It’s hard to see a situation in which that actually makes sense, but here the complaint accuses Google’s “Bernanke program” of effectively shifting between 2nd and 3rd party auctions within the same auctions and collecting all of the remaining cash. In the above example, the way this would work is that Henry bids $19, Alice bids $18, and Scott bids $9 in order to, let’s just say, put an ad on Techdirt. Again, according to the complaint, Bernanke would effectively ignore the 2nd bid on the publisher side, telling us that the second bid was actually Scott’s $9 bid, and therefore claiming to us (or any publisher) that the “clearing” price was $9 — and it would then take its 20% fee of $1.80. But for the advertiser, Google would still show them the actual top three bids, telling them that $18 was the clearing price. So now, rather than Techdirt (the publisher) get the $14.40 we should be getting, we’d be getting just the $7.20, and Google would walk away with $10.80 (rather than the $3.60 it claimed it would take).
That’s all a bit confusing, but this graphic in the complaint makes it clearer:
Now, those numbers are pretty stark, and the amounts would be quite different if you had 3 bids that were all pretty close, but the complaint notes that Google’s own internal data showed that Bernanke lowered publishers’ revenue by 40%.
Google examined some of the effects of its secret Bernanke program, finding that it
drops any given publisher?s revenue by upwards of 40 percent. Stating the obvious, one Google
employee observed: ?Bernanke is powerful.? Publishers had no idea Google was dropping second highest
bids and impacting their revenues in this way.
That certainly looks like a combination of deceptive practices and something close to fraudulently taking money that was supposed to go to publishers. At the very least, this revelation is absolutely going to lead to some publishers filing a separate civil suit against Google demanding their cut, and possibly having it turn into a class action lawsuit. Hell, I’d bet advertisers are going to file a similar suit noting that they overpaid for ads as well.
Given some of the other information in the complaint, there are some hints as to how Google got to this point. The description above sounds pretty damn evil, but there’s a kind of dumb and dangerous internal logic. Google realized that it was starting to lose ad auctions to 3rd party systems quite a lot, and the original point of Bernanke was apparently not to steal money from publishers, but to have Google start winning more auctions. So, the initial idea behind Bernanke seems to have been to inflate bids to try to win more auctions — and in those cases you can see a kind of logic to dropping certain bids from the auction as they were “artificially” inflated in order to win more auctions. Though… that’s still pretty sketchy.
Again assuming this is accurate, this is, to put it mildly, fucked up. It’s just out and out greed by the ad team, and it completely fucked the advertising market for tons of news organizations that relied on such things. I mean, as described in the complaint, this is so evil it’s almost stupid. Google would have been much better off simply admitting that it was taking a much larger cut of the ad revenue than cooking up this ridiculous scheme.
So, yeah, I’ll be reading the answer to this amended complaint carefully to see if there’s some alternative explanation, but based on what’s in this complaint, this is not just the strongest antitrust claim against Google, but one that Google should very much lose. There are lots of claims out there of bad behavior by various tech companies, and much of it is misrepresented or misleading. But if this is accurate, it’s incredibly damning and Google should be in serious trouble for it.
Filed Under: ads, antitrust, deceptive practices, price fixing, project bernanke, texas