`
`
`UNITED STATES DISTRICT COURT
`FOR THE DISTRICT OF COLUMBIA
`
`
`
`Case No._______________________
`
`
`
`
`CLASS ACTION COMPLAINT
`
`DEMAND FOR JURY TRIAL
`
`CLIFFY CARE LANDSCAPING LLC
`15837 S. Mahaffie Street
`Olathe, KS 66062
`
`on behalf of itself and all others similarly
`situated,
` Plaintiff,
`
`v.
`
`FACEBOOK, INC.
`1601 Willow Road
`Menlo Park, CA 94025
`
`GOOGLE LLC
`1600 Amphitheatre Parkway
`Mountain View, CA 94043,
`
`and
`
`ALPHABET INC.
`1600 Amphitheatre Parkway
`Mountain View, CA 94043
`
`
`
` Defendants.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Plaintiff, individually and on behalf of a Class of all those similarly situated, brings this
`
`class action complaint for equitable relief and treble damages against Defendants, FACEBOOK,
`
`INC. (“Facebook”), GOOGLE LLC, and ALPHABET INC. (collectively, “Google”) for
`
`violation of the Sherman Antitrust Act, 15 U.S.C. § 1, and alleges as follows:
`
`
`
`
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 2 of 22
`
`
`I.
`
`NATURE OF THIS ACTION
`
`1.
`
`In 2019, spending in the United States on digital, online advertising reached
`
`$129.34 billion, exceeding for the first time the total spent in the U.S. on all forms of traditional
`
`print, radio, television and billboard advertising. In 2021, digital ad spending in the U.S. is
`
`expected to reach $198 billion, about a third of which, or $66.2 billion, will be spent on search
`
`advertising, in which advertisers target search engine users searching for a particular product or
`
`service and pay to have ads placed next to the search results. This audience is composed mostly
`
`of users of Google, which has an 88% share of search queries in the U.S. and a 92% share of
`
`search queries worldwide. The remaining U.S. digital ad spend, or about $132 billion, will be
`
`spent on display advertising, in which advertisers place images, banners, or videos on websites
`
`likely to be viewed by the advertiser’s target audience.
`
`2.
`
`About 86% of today’s online display advertising will be bought and sold
`
`electronically at high speeds through centralized trading venues known as “exchanges.”
`
`Beginning in 2005, the rise of electronic ad trading, known as “programmatic advertising,” has
`
`transformed advertising from a relationship business to a commodity business, with publishers1
`
`and advertisers transacting with each other in an electronic spot market. Google’s advertising
`
`exchange alone trades ad spaces targeted to billions of individual users and processes tens of
`
`billions of targeted ads daily.
`
`3.
`
`With enormous audiences, a huge inventory of content, and the advantages of
`
`network effects and technological prowess, Google, Facebook, and Amazon control about 79%
`
`of non-search digital advertising. Facebook’s 2.8 billion monthly users and Google’s 1.8 billion
`
`
`1 A publisher is the operator of a website or mobile app that makes space available for the
`display of advertisements, including images, banners, and videos.
`
`
`
`2
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 3 of 22
`
`
`Gmail account holders—together with their identities, search and browsing histories, spending
`
`habits, social connections, and locations—endow these firms with unprecedented capacity to
`
`reach and target consumers. In addition to selling display space on their “owned-and-operated”
`
`properties, Google, Facebook, and to a lesser extent, Amazon, also sell advertisers space on third
`
`party websites, acting as intermediaries in what has become known as the “open display”
`
`advertising market.
`
`4.
`
`A large proportion of open display advertising is sold directly to advertisers
`
`through the “ad networks” operated by each of these firms: the Google Display Network
`
`(“GDN”) (part of Google Ads), the Facebook Audience Network (“FAN”), and the Amazon
`
`Demand Side Platform (“Amazon DSP”). Defendants’ ad networks, GDN and FAN, accounts for
`
`about 67% of the total open display ad spend, while Amazon’s share and those of other networks
`
`are far lower.
`
`5.
`
`Defendants’ horizontally competing ad networks, GDN and FAN, should
`
`vigorously compete for advertisers and publishers, but they do not. Instead, as alleged in a
`
`Complaint filed against Google on December 16, 2020 by 10 state attorneys general2 and
`
`elaborated upon in an article in the New York Times on January 17, 2021,3 Defendants conspired
`
`to allocate to one another the advertisers and publishers affiliated with each network and to
`
`eliminate competition between them in the open display advertising market. In an agreement that
`
`Google insiders codenamed “Jedi Blue,” Facebook agreed to bid FAN’s demand through
`
`Google’s ad exchange, rather than directly through multiple exchanges using a competing
`
`technology called “header bidding.” In return for that agreement, FAN received (i) preferential
`
`
`2 Texas et al. v. Google LLC, Complaint (E.D.Tx. 4:20-cv-00957, filed Dec. 12, 2020).
`3 Daisuke Wakabayashi and Tiffany Hsu, “Behind a Secret Deal Between Google and
`Facebook,” New York Times, Jan. 17, 2021, available at https://nyti.ms/3imRzwt.
`
`
`
`3
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 4 of 22
`
`
`treatment over other bidders, including a guaranteed “win rate,”; (ii) superior information about
`
`the advertising opportunity, including the identity of the user most of the time; and (iii) increased
`
`“timeouts” for Facebook to bid before it was excluded from the auction, all of which allowed
`
`Facebook to bid and win more often relative to non-Facebook bidders.
`
`6.
`
`Defendants’ conspiracy to collude rather than compete violates Section 1 of the
`
`Sherman Act, 15 U.S.C. § 1, and has caused and will continue to cause injury and economic
`
`harm to Plaintiff and all other similarly situated advertisers that bid through Google’s Open
`
`Bidding program through any non-FAN ad network or demand-side platform, including Google
`
`Ads and GDN, for which Plaintiff seeks appropriate equitable relief and damages through this
`
`action.
`
`II.
`
`JURISDICTION AND VENUE
`
`7.
`
`This Court has original jurisdiction over Plaintiff’s federal antitrust claims, which
`
`arise under Sections 1 and 15 of the Sherman Act, 15 U.S.C. §§ 1, 15, and Sections 4 and 16 of
`
`the Clayton Act, 15 U.S.C. §§ 15 and 26. The Court also has diversity jurisdiction over this
`
`action under the Class Action Fairness Act of 2005, 28 U.S.C. § 1332(d), because at least one
`
`class member is of diverse citizenship from Defendants, there are more than 100 class members
`
`nationally, and the aggregate amount in controversy exceeds $5,000,000. This Court also has
`
`subject matter jurisdiction over Sherman Act claims pursuant to 28 U.S.C. §§ 1331 & 1337.
`
`8.
`
`This Court has personal jurisdiction over Defendants, which are found and
`
`transact business in this district.
`
`9.
`
`The Venue is proper in this District pursuant to 28 U.S.C. § 1391(b), (c), and (d)
`
`because during the Class Period (defined below), Defendants resided, transacted business, and
`
`had agents in this district.
`
`
`
`4
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 5 of 22
`
`
`10.
`
`Defendants’ acts were within the flow of, were intended to have, and did have a
`
`substantial effect on the interstate commerce of the United States.
`
`III.
`
`PARTIES
`
`A.
`
`Plaintiff
`
`11.
`
`Plaintiff Cliffy Care Landscaping is a limited liability company in good standing,
`
`registered in the state of Kansas, with a principal place of business at 15837 S. Mahaffie Street,
`
`Olathe, Kansas, 66062. Plaintiff purchased display advertising through Google Ads between
`
`September 2018 and the present.
`
`B.
`
`Defendants
`
`12.
`
`Defendant Facebook, Inc., is a publicly traded, for-profit company, organized
`
`under the laws of Delaware with its principal place of business at 1601 Willow Road, Menlo
`
`Park, California. Facebook’s principal business provides personal social network services to
`
`approximately 3 billion people throughout the world through control of its network, known
`
`internally as “Facebook Blue.”
`
`13.
`
`Defendant Google LLC, is a limited liability company organized under the laws
`
`of Delaware with its principal place of business at 1600 Amphitheatre Parkway, Mountain View,
`
`California. Google LLC is a wholly owned and controlled subsidiary of XXVI Holding Inc.,
`
`which is a subsidiary of Defendant Alphabet Inc.
`
`14.
`
`Defendant Alphabet Inc., is a publicly traded, for-profit company organized under
`
`the laws of Delaware with its principal place of business at 1600 Amphitheatre Parkway,
`
`Mountain View, California. Google LLC is a wholly owned subsidiary of Alphabet.
`
`15.
`
`Defendants Google LLC and Alphabet Inc., are collectively referred to herein as
`
`“Google.” Google is a technology company that provides internet-related services and products,
`
`
`
`5
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 6 of 22
`
`
`including online advertising technologies, the world’s most dominant search engine, and the
`
`world’s most visited website, YouTube.
`
`16.
`
`Defendants’ conduct was authorized, ordered, or performed by their directors,
`
`officers, managers, agents, employees, or representatives in the course of their employment and
`
`while actively engaged in the management of Defendants’ affairs.
`
`17.
`
`Defendants, through their subsidiaries, divisions, affiliates and agents, operated as
`
`a single unified entity with each acting as the alter ego, agent or joint-venturer of or for the other
`
`with respect to the acts, violations, and common course of conduct alleged herein and under the
`
`authority and apparent authority of parent entities, principals and controlling parties.
`
`IV. CLASS ALLEGATIONS
`
`18.
`
`Plaintiff brings this action on behalf of itself and as a class action under Rules
`
`23(a), (b)(2) and (b)(3) of the Federal Rules of Civil Procedure on behalf of the members of the
`
`following Class:
`
`All persons who purchased digital display advertising through
`Google Ads, Amazon DSP, or other non-Facebook demand-side
`platform to reach consumers in the United States between
`September 2018 and the present (“Class Period”).
`
`Specifically excluded from this Class are the Defendants; the officers, directors or
`
`employees of any Defendant; any entity in which any Defendant has a controlling interest; and
`
`any affiliate, legal representative, heir or assign of any Defendant. Also excluded from this
`
`Class are any federal, state or local governmental entities, any judicial officer presiding over
`
`this action and the members of his/her immediate family and judicial staff, and any juror
`
`assigned to this action.
`
`19.
`
`Class Identity/Ascertainability: The Class is readily identifiable and is one for
`
`which records should exist.
`
`
`
`6
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 7 of 22
`
`
`20.
`
`Numerosity: Due to the nature of the trade and commerce involved, Plaintiff
`
`believes there are thousands of Class members as above described, the exact number and their
`
`identities being known to Defendants and their Co-conspirators.
`
`21.
`
`Typicality: Plaintiff’s claims are typical of the claims of the members of the Class
`
`because Plaintiff, directly from one or more of the Defendants or their co-conspirators, and
`
`therefore Plaintiff’s claims arise from the same common course of conduct giving rise to the
`
`claims of the members of the Class and the relief sought is common to the Class.
`
`22.
`
`The facts of this case are the same for all members of the class in that Defendants’
`
`conduct in conspiring and agreeing to restrain competition was the same for all members of the
`
`class.
`
`23.
`
`The same legal standards govern resolution of each claim set forth below for all
`
`members of the class and across each of the class member’s individual claims. If Defendants are
`
`liable to one member of the class, Defendants are liable to all members of the class.
`
`24.
`
`Because the claims of each member of the Class have a common origin and share
`
`a common basis in terms of Defendant’s systematic misconduct, there are common questions of
`
`fact and law which exist and which are susceptible to common answers as to each Class member
`
`under Federal Rule of Civil Procedure 23(a)(2), and which predominate over any questions
`
`affecting only individual members under Federal Rule of Civil Procedure 23(b).
`
`25.
`
` Commonality: There are questions of law and fact common to the Class,
`
`including, but not limited to:
`
`a. Whether Defendants entered into a conspiracy and agreement to cooperate
`
`in the open display advertising market rather than compete;
`
`b. Whether Defendants’ agreement allocated customers between them;
`
`
`
`7
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 8 of 22
`
`c. Whether Defendants’ agreement allocated market segments in the ad tech
`
`stack between them;
`
`d. Whether Google entered into an agreement with Facebook for Facebook to
`
`win a fixed percentage of Google’s auctions;
`
`e. Whether Facebook was granted preferences not granted to other bidders in
`
`Google’s auctions;
`
`f. Whether Defendants’ agreement unreasonably restrained trade and thereby
`
`violated Section 1 of the Sherman Act, 15 U.S.C. § 1;
`
`g. Whether Defendants’ violation was a cause of injury-in-fact to Plaintiff
`
`and the class members and whether such injury constituted antitrust injury;
`
`h. The appropriate measure and quantum of aggregate damages suffered by
`
`the class caused by Defendants’ violation; and,
`
`i. Whether Defendants should be permanently enjoined from engaging in the
`
`same or similar concerted action or agreement in the future.
`
`26.
`
`Predominance: These and other common questions of law or fact predominate
`
`over any questions affecting only individual members of the Class.
`
`27.
`
`Adequacy: Plaintiff will fairly and adequately protect the interests of the Class in
`
`that Plaintiff’s interest are aligned with, and not antagonistic to, those of the other members of
`
`the Class and Plaintiff has retained counsel competent and experienced in the prosecution of the
`
`class actions and antitrust litigation to represent itself and the Class.
`
`28.
`
`Superiority and Manageability: A class action is superior to other available
`
`methods for the fair and efficient adjudication of this controversy since individual joinder of all
`
`damaged Class members is impractical. Prosecution as a class action will eliminate the
`
`
`
`8
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 9 of 22
`
`
`possibility of repetitious litigation. The damages suffered by individual Class members are
`
`relatively small, given the expense and burden of individual prosecution of the claims asserted in
`
`this litigation and Class members’ interests in individually controlling the prosecution of separate
`
`actions is de minimis and impracticable. Absent a class action, it would not be feasible for Class
`
`members to seek redress for the violations of law herein alleged. Further, individual litigation
`
`presents the potential for inconsistent or contradictory judgments and would greatly magnify the
`
`delay and expense to all parties and to the court system. The Class members’ interests in
`
`individually controlling the prosecution of separate actions is de minimis and impracticable and
`
`it is desirable to concentrate litigation of these claims this forum. There are no known or
`
`apparent likely difficulties in managing a class action. A class action will provide the benefits of
`
`unitary adjudication, economy of scale and comprehensive supervision by a single court.
`
`V.
`
`FACTUAL ALLEGATIONS
`
`29. When a user visits a website or uses a mobile application, the ad space on the
`
`page is instantly routed into one or more ad exchanges to be sold to the highest bidder in a real-
`
`time, electronic auction. At the conclusion of the auction, the winning advertiser’s ad displays to
`
`the user in time for the page to load and before the user has noticed anything has occurred. The
`
`user just sees the website page and the ad targeted to them. This process is known as
`
`“programmatic display advertising.” Although some advertising space is sold at fixed prices
`
`through direct deals between publishers and advertisers or media agencies, most display
`
`advertising is sold programmatically. More than 86% of online display advertising traded
`
`programmatically in 2020.
`
`30.
`
`The scale of Google’s audience and Facebook’s social network is large enough
`
`for Defendants to offer their own self-serve interfaces for programmatic trading. Advertisers
`
`
`
`9
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 10 of 22
`
`
`engage Google and Facebook to place ads on Google and Facebook owned-and-operated sites,
`
`such as YouTube (owned by Google) and Instagram (owned by Facebook). The placement of ads
`
`on Google and Facebook’s own sites are said to be placed within the firm’s “walled garden” and
`
`is sometimes known as “first party” display advertising. Over half of first party display
`
`advertising is generated by Facebook (for space on Facebook Blue and Instagram), with the
`
`second highest share belonging to Google (primarily for YouTube). Operators of walled gardens
`
`enjoy an intrinsic advantage over other sellers of programmatic display advertising because
`
`media companies with authenticated users achieve significantly better economics than media
`
`companies with anonymous traffic.
`
`31.
`
`For most other publishers of online content, however, programmatic trading is
`
`only possible through a range of intermediaries between advertisers and publishers known as the
`
`“ad tech stack.” Google and Facebook both leverage the information they know about users, their
`
`programmatic trading technologies, and their enormous customer base of advertisers and media
`
`agencies to sell impression-targeted display ad space on behalf of other market participants.
`
`Known as “open display advertising,” advertisers in this market buy inventory from many
`
`publishers with a smaller scale (for example, newspapers and app providers). Open display
`
`advertising comprises 30-40% of total programmatic display ad expenditures. Because they
`
`maintain much richer and higher quality user data and exhibit much greater scale and reach than
`
`their rivals, Google and Facebook enjoy a substantial competitive advantage over other
`
`intermediaries in the open display market. For example, Google can track users’ visits to at least
`
`70% of the top one million sites on the internet and has tags (including as a third party) tracking
`
`user behavior on over 80% of popular websites. Facebook has the second highest prevalence of
`
`tags, covering between 40-50% of the most popular websites. Amazon, which operates a similar
`
`
`
`10
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 11 of 22
`
`
`walled garden, is growing their open display intermediation business, but remains third after
`
`Google and Facebook, with less than 10% of the open display ad market. The non-walled garden
`
`share of programmatic open display advertising has declined by roughly $1 billion/year for the
`
`past several years.
`
`32.
`
`The ad tech stack that connects advertisers with publishers consists of the
`
`following types of participants, starting with the advertiser-facing (demand side) participants:
`
`a. Media agencies, used by large advertisers to plan and deliver advertising
`
`campaigns and provide technical expertise to execute programmatic
`
`buying;
`
`b. Advertiser ad servers, used by advertisers and media agencies to store ads
`
`and deliver them to publishers;
`
`c. Demand side platforms (“DSPs”), used by advertisers and media agencies
`
`to bid on and buy advertising space (inventory) from multiple sources
`
`according to the buyer’s objectives and data about the inventory and the
`
`final user;
`
`d. Supply side platforms (“SSPs”), used by publishers to collect bids from
`
`DSPs, automate the sale of inventory through real-time auctions or direct
`
`deals with advertisers, and perform the ad exchange function;
`
`e. Publisher ad servers (“PAS”), used by publishers to manage inventory and
`
`to determine which ads to serve based on the bids received from SSPs or
`
`agreed between the publisher and advertisers.
`
`33.
`
`On September 15, 2020, the Subcommittee on Antitrust, Competition Policy, and
`
`Consumer Rights of the Senate Judiciary Committee held a hearing on the subject of “Stacking
`
`
`
`11
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 12 of 22
`
`
`the Tech: Has Google Harmed Competition in Online Advertising?” The Subcommittee’s
`
`Majority Staff Report and Recommendations, released on October 6, 2020, reported that Google
`
`captures over 50% of the market across the ad tech stack. It runs the leading ad exchange
`
`(formerly known as AdX), while also running buy-side and sell-side intermediary platforms
`
`trading on this exchange. Google’s dominance is a result of a series of acquisitions completed
`
`since 2008, when Google purchased the leading ad server, DoubleClick, which provided the
`
`technology for Google’s current PAS. Two years later, Google acquired AdMob, the largest ad
`
`server for the mobile application market. In 2010, Google acquired Invite Media, which it re-
`
`launched in 2012 as DoubleClick Bid Manager and eventually converted into Google’s
`
`enterprise DSP, Display & Video 360 (“DV 360”). In 2011, Google purchased AdMeld, one of
`
`the largest display advertising SSPs, which it integrated into AdX, Google’s existing exchange.
`
`And in 2014, Google bought Adometry, an analytics and attribution provider it then integrated
`
`into Google Analytics. On October 20, 2020, the Department of Justice brought suit against
`
`Google for monopolization of the search and search advertising markets, followed by the suit
`
`brought by the Texas attorney general and nine other state attorneys general for monopolization
`
`and attempted monopolization of the PAS, ad exchange, and DSP display advertising markets. It
`
`is unusual, to say the least, for a single company to represent both sellers and buyers in the same
`
`market, and to set the rules for, and conduct, the auctions that determine the winners, losers, and
`
`prices in that market.
`
`34.
`
`After Google’s 2008 acquisition of DoubleClick, the leading intermediary on the
`
`publisher side, DoubleClick immediately began preferentially routing trading activity to AdX,
`
`Google’s ad exchange. By 2011, AdX was the leading exchange in the market with the highest
`
`trading volume. At the time, bidding for ad space was conducted sequentially, in a system called
`
`
`
`12
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 13 of 22
`
`
`the “waterfall.” In this system, a publisher ad server offers a particular impression to a particular
`
`exchange; if that exchange produced a bid above the reserve price, the ad was placed. If the
`
`exchange did not produce an acceptable bid, the ad server would offer the impression to the
`
`exchange next in line, proceeding down a “waterfall” until the impression was sold to a buyer.
`
`Typically, Google’s AdX was first in line, would assess whether the impression was a valuable
`
`user or was instead a low-value impression like a bot. If the former, then Google AdX would bid
`
`on the opportunity, if the latter, it would pass the opportunity to the next exchange in line. Aware
`
`it was in second place, the second exchange would bid low to account for the risk that the
`
`opportunity had low or zero value. The order of exchanges in the waterfall was determined by
`
`historical average bid levels.
`
`35.
`
`Google’s policy of refusing to let publishers on Google’s ad server route their ad
`
`spaces to more than one exchange at a time drew the ire of the industry. By 2015, publishers had
`
`developed a method known as “header bidding” in an attempt to avoid Google’s routing
`
`restrictions and to push the advertising market to function more like other electronic trading
`
`markets. In header bidding, a publisher inserts a piece of JavaScript code into the header section
`
`of their webpages. When a user visits a page, the JavaScript routes information about the
`
`website’s ad space to multiple ad exchanges, bypassing Google’s PAS. Publishers could route
`
`inventory to a dozen exchanges at once, instead of routing them to Google’s exchange first. This
`
`version of header bidding is known as “client-side” header bidding because the bid-request
`
`information is stored on the user’s device rather on a server. With client-side header bidding,
`
`websites use a “wrapper” that sets a universal timeout, a common time limit within which all
`
`exchanges must submit a bid. In header bidding, all the SSPs (exchanges) get the chance to view
`
`the impression at the same time and submit their respective bids before the ad server is called,
`
`
`
`13
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 14 of 22
`
`
`which is why header bidding is also referred to as pre-bidding. Header bidding solved the
`
`problem that arises when an ad exchange with an early place in the waterfall sequence produced
`
`a bid above the publisher’s minimum acceptable price and won the impression even if an
`
`exchange later in the waterfall sequence had elicited a substantially higher bid, a situation
`
`illustrated in the following figure:
`
`
`
`36. Moreover, by setting longer timeouts, publishers found they could neutralize any
`
`speed advantages that Google might have from the colocation of its DSPs (Google Ads and
`
`DV360) and its exchange, AdX. Publishers also found that header bidding not only offered
`
`increased yields but increased transparency, because publishers know what each advertiser is
`
`willing to bid before the publisher calls its ad server and can gauge the effect of adjustments in
`
`the timeout on yield. By 2016, about 70% of major publishers had adopted header bidding.
`
`37.
`
`Although header bidding offered publishers certain benefits, it also introduced
`
`some problems. Header bidding can be difficult to implement, requiring both advertising
`
`operations and development resources and it requires the addition of extra code on the webpage
`
`which can slow down the publisher’s website and detract from the site’s user experience.
`
`
`
`14
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 15 of 22
`
`
`Nonetheless, header bidding was widely viewed in the industry as a viable means of challenging
`
`Google’s hold on the exchange market.
`
`38.
`
`Google refused to participate in header bidding, which severely limited the
`
`participants in header bidding auctions because of the large advertiser demand it controls.
`
`Moreover, even though the introduction of header bidding created an environment where the bids
`
`submitted by SSPs participate in a final, first-price auction, Google’s refusal resulted in Google
`
`getting a “last look” at bid opportunities because the winning bids from the non-Google
`
`exchanges participating in header bidding were then sent to Google’s PAS. Google’s PAS then
`
`submitted it to AdX to see if the bid could be improved, essentially using the results of the
`
`header bidding auction as a floor price for AdX bidders. This set-up is inefficient, of course,
`
`because advertisers bidding on the other exchanges do not have the opportunity to outbid the
`
`winner of the AdX auction.
`
`39.
`
`Instead of adopting header bidding, Google responded by introducing two
`
`innovations, a development framework called Accelerated Mobile Pages (“AMP”), which it
`
`launched in early 2016, and its own version of header bidding, called Open Bidding (formerly
`
`known as Exchange Bidding), which was developed in 2016 but became available in April 2018.
`
`Both initiatives were ostensibly intended to address the latency and retooling problems with
`
`client-side header bidding, but according to the United Kingdom’s Competition & Markets
`
`Authority (“CMA”) (the U.K.’s antitrust enforcement agency), “a major reason for the
`
`introduction of Exchange Bidding appears to have been protecting Google’s revenues from the
`
`impact of header bidding … .”4 Moreover, the AMP solution restricted the use of JavaScript,
`
`
`4 Online Platforms and Digital Advertising, Market Study Final Report (July 2020),
`Appendix M, ¶ 36, available at:
`https://www.gov.uk/cma-cases/online-platforms-and-digital-advertising-market-study.
`
`
`
`15
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 16 of 22
`
`
`precisely the code that websites (in the case of AMP, mobile apps) needed to implement client-
`
`side header bidding.5
`
`40.
`
`Both innovations move control of the bidding process onto Google’s servers.
`
`Although both header bidding and Open Bidding are designed to allow simultaneous auctions at
`
`several exchanges, Open Bidding requires calls to ad exchanges to be made from Google’s ad
`
`server rather than from the user’s device. For this reason, Open Bidding is known as server-side
`
`header bidding. At the same time, Google imposed tighter timeout restrictions, refused to share
`
`user data (thereby requiring non-Google exchanges to engage in a time-consuming process of
`
`“cookie matching” to identify the user), and levied a 5-10% fee on ads from non-Google ad
`
`exchanges.
`
`41.
`
`Similarly, the AMP framework requires publishers to allow Google to host
`
`publishers’ content on its own servers to enable them to be served faster. Once a publisher loads
`
`content in AMP format, Google creates a cached version on Google’s servers. Each time a user
`
`navigates to the publisher’s AMP content from a Google property (e.g., Search or Google News),
`
`instead of directing the user to the publisher’s server, Google serves the AMP content from
`
`Google’s AMP cache server, a practice which prevents publishers from collecting their own data
`
`on its users.
`
`42.
`
`Google could profitably resist adopting header bidding and promote the AMP
`
`framework only because publishers cannot ignore its huge user base. But, Facebook controls
`
`even greater demand, accounting for over half of all display advertising revenue and considered
`
`
`5 Google recently loosened its restrictions on the use of JavaScript in its AMP solution, but it
`remains incompatible with client-side header bidding in its original form because it limits the
`number of vendors that publishers can route to five, allows only a single cookie-sync, and
`imposes strict trading timeouts.
`
`
`
`16
`
`
`
`Case 1:21-cv-00360 Document 1 Filed 02/09/21 Page 17 of 22
`
`
`a ‘must have’ platform by many advertisers. It has a significant user data advantage, which both
`
`increases the value of its advertising inventory and creates additional barriers for its competitors
`
`to overcome. Through FAN, the Facebook Audience Network, Facebook sells space outside its
`
`walled garden, including space on over half of all mobile apps available on the Google Play
`
`Store. More than a billion people per month see an advertisement in a website or mobile app
`
`through FAN. Moreover, because consumers now use their phones more than their computers
`
`and publishers receive approximately 60% of their traffic from mobile use, Facebook’s user base
`
`and mobile inventory have become even more important.
`
`43.
`
`In March 2017, Facebook announced that the company had chosen a group of
`
`technology partners through which it would implement a header bidding solution that opened
`
`FAN’s demand source directly to web and mobile app publishers, enabling publishers and
`
`advertisers to avoid the fees for transacting through Google’s ad server. The competitive threat
`
`posed by Facebook’s adoption of header bidding was not lost on Google. According to the Wall
`
`Street Journal, Google advertising executive Chris LaSala wrote in an internal document
`
`outlining the company’s 2017 priorities: “Need to fight off the existential threat posed by header
`
`bidding and FAN.”6 The Times reported a Google executive calling for “an all hands on deck
`
`approach.” 7
`
`44.
`
`The specter of Facebook, Amazon, and thousands of publishers lining up in 2017
`
`to resist Google’s dominance over the ad tech stack by promoting technology that would enable
`
`
`6 Ryan Tracy and Jeff Horwitz, “Inside the Google-Facebook Ad Deal at the Heart of a
`Price-Fixing Lawsuit,” Wall Street Journal, Dec. 29, 2020, available at
`https://www.wsj.com/articles/inside-the-google-facebook-ad-deal-at-the-heart-of-a-price-fixing-
`lawsuit-11609254758.
`7 Daisuke Wakabayashi and Tiffany Hsu, “Behind a Secret Deal Between Google and
`Facebook,” New York Times, Jan. 17, 2021, available at https://nyti.ms/3imRzwt.
`
`
`
`17
`
`
`
`Case 1:21-c