PUBLIC
THE COMPETITION TRIBUNAL
IN THE MATTER OF the Competition Act, R.S.C. 1985, c. C-34;
CT-2024-010
AND IN THE MATTER OF certain conduct of Google Canada Corporation and Google LLC relating to the supply of online advertising technology services in Canada;
AND IN THE MATTER OF an application by the Commissioner of Competition for one or more orders pursuant to section 79 of the Competition Act.
BETWEEN:
COMMISSIONER OF COMPETITION – and –
Applicant
GOOGLE CANADA CORPORATION AND GOOGLE LLC Respondents
REPLY
2.
3.
4.
5.
PUBLIC
OVERVIEW The Commissioner of Competition (the “Commissioner”) repeats and relies upon the allegations in the Notice of Application and Concise Statement of Economic Theory (the “Application”) and, except as hereafter expressly admitted, denies each of the allegations in the Response (the “Response”) of Google Canada Corporation and Google LLC (collectively, “Google”). Capitalized terms used herein are as defined in the Application.
The Application correctly identifies three relevant product markets: Advertiser Ad Network services, Publisher Ad Server services, and Ad Exchange services. These markets each constitute functionally distinct services used by advertisers and publishers, respectively, to buy and sell open web display advertisements programmatically.
In contrast, Google’s Response proposes a single two-sided platform product market that ignores commercial reality, and is contradicted by Google’s own internal ordinary course documents. Google’s proposed market also serves to mask its dominant position, anti-competitive conduct, and the substantial lessening and prevention of competition (“SLPC”) resulting from that conduct.
Google’s product market definition suffers from at least two flaws, both independently fatal.
First, Google’s market definition erroneously seeks to fuse together separate markets for functionally distinct, non-substitutable ad tech services, purchased by different parties, into a unitary two-sided platform market. In its Response, Google never identifies precisely what service its unitary two-sided platform provides or to what group of purchasers it provides that service. Instead, Google describes the market in vague terms, saying that market participants view ad tech tools as a “suite of services” that connect or match advertisers and publishers.
2
6.
7.
8.
9.
10.
PUBLIC
Second, Google’s product market definition is underpinned by the erroneous proposition that all digital advertising, regardless of form (e.g., static banner ad, in-stream video, native) and place of publication (e.g., webpage, in-app, social media), is substitutable. Google asks this Tribunal to ignore the significant differences between various types of digital advertising.
Google’s error in defining the relevant market flows through its entire Response and undermines its positions regarding dominance, anti-competitive conduct, and competitive harm.
When Google’s dominance is assessed in the relevant Publisher Ad Server and Advertiser Ad Network markets, it is clear that Google has substantial market power in those markets, given its high and sustained market shares, barriers to entry and expansion, and its ability to degrade the quality of its product offerings with no material price impact or loss of revenue.
Google has maintained its dominance over time through a systematic campaign of interrelated anti-competitive actions, designed to entrench its substantial market power and harm the competitive process. Google has tied the use of certain of its key ad tech tools to the use of other, functionally distinct, Google tools. It built upon those anti-competitive ties to engage in further anti-competitive conduct.
Google claims that its conduct has not resulted in an SLPC in its unitary two-sided platform market. That claim is without merit as it is underpinned by Google’s flawed market definition. Google also says that there has been no SLPC in the Publisher Ad Server and Advertiser Ad Network markets identified by the Commissioner. However, rather than responding to the Commissioner’s SLPC allegations in a substantive way, Google simply reverts to its product market argument, stating the effects in those markets are “incapable of being properly identified or assessed” as the Commissioner’s product markets are incorrect.
3
11.
12.
13.
14.
15.
PUBLIC
A consistent theme of the Response is Google’s focus on various alternative means or workarounds available to advertisers and publishers to buy and sell advertising which, while theoretically available, do not represent commercially viable options.
The entirety of Google’s conduct should be assessed under the current abuse of dominance provisions of the Competition Act (the “Act”). However, the Application meets the test for liability under both the current and former provisions of ss. 79(1) of the Act.
Google’s limitation defence is unavailing. Though Google’s anti-competitive conduct has taken different forms and evolved over time, it represents a continuous practice by Google to exclude rivals from the relevant markets and to have an adverse effect on competition. In any event, Google’s dominant position, practice of anti-competitive acts, and effects over the three years preceding the Application warrant the remedies sought by the Commissioner.
The remedies the Commissioner seeks are appropriate and within the Tribunal’s jurisdiction. The divestitures sought are appropriate because they are required to address Google’s anti-competitive conduct and because that conduct and its effects have a real and substantial connection to Canada. The administrative monetary penalty (“AMP”) sought reflects the upper limit for AMPs set out in the Act. The decision to order an AMP and the amount of any AMP ordered are matters solely within the discretion of the Competition Tribunal (the “Tribunal”), which is to be exercised having regard to factors set out in ss. 79(3.2) of the Act and the requirement in ss. 79(3.3) that any AMP ordered be for the purpose of promoting compliance with ss. 79(1) of the Act, not to punish the person against whom the order is made.
Google’s Response is replete with groundless ad hominem attacks on the Commissioner and the Application, including that the Commissioner’s allegations are “tactical”, “concocted”, “unprincipled”, “inappropriate” and “contrived”. The Tribunal should ignore Google’s groundless allegations.
4
PUBLIC
II. THERE IS NO SINGLE TWO-SIDED MARKET FOR ALL DIGITAL ADVERTISING TECHNOLOGY SERVICES
16.
17.
18.
Google erroneously defines the relevant product market as a unitary two-sided market that includes all the functionally distinct tools that advertisers and publishers can use to buy and sell all forms of digital advertising. 1 Google does not identify what service its claimed unitary two-sided market provides. Instead, it describes an outcome – a match between advertisers and publishers - which it says is accomplished with “a suite of services” that connect those market participants. 2
Defining relevant markets involves determining what products or services are substitutes for a given service or product. Key to that determination are the views of users of the product. In this matter, advertisers use a number of different services to buy digital advertising inventory. Similarly, publishers use a number of different services to sell their inventory.
Google’s market definition is flawed for two reasons.
a. First, Google errs in attempting to fuse together a number of functionally distinct and non-substitutable services, purchased by different parties, into a single two-sided market. These services include not only publisher ad servers, advertiser ad networks, and ad exchanges used to buy and sell programmatic open web display advertising, but also virtually every other buy-side tool used by advertisers to buy digital advertising and sell-side tool used by publishers to sell digital advertising.
b. Second, Google’s market definition rests on the erroneous proposition that all digital advertising, regardless of its form and place of publication, is substitutable.
1 Response, paragraph 7. 2 Response, paragraphs 175, 176.
5
19.
20.
21.
PUBLIC
A. Functionally Different Ad Tech Tools used by different Parties cannot be Fused together into a Single Market
Publishers use “sell-side” tools, such as Google’s DoubleClick for Publishers ("DFP") publisher ad server, as part of the process for selling their digital ad space. Advertisers use “buy-side” tools, such as Google Ads, as part of the process for buying ad space from publishers.
Publishers do not use buy-side tools like Google Ads and advertisers do not use publisher sell-side tools, like DFP. Those tools are functionally distinct and do not provide publishers and advertisers, respectively, with functionality they require to sell and buy advertising space. 3
That these sell-side and buy-side tools are used by different parties along with other tools, including ad exchanges, to ultimately achieve an end – the purchase and sale of advertising space – does not make them substitutes for one another. The fact that a screwdriver can be used with pliers to fasten a nut to a bolt, does not make pliers and a screwdriver substitutes for one another. DFP and Google Ads are distinct sell-side and buy-side tools, which along with AdX, form part of Google’s ad tech ecosystem. Google has used its dominance and anti-competitive conduct to lock advertisers and publishers into that system.
B. The Commissioner has Correctly Defined the Relevant Product Markets
22.
There are three relevant product markets in this matter, made up of functionally distinct services used for the programmatic purchase and sale of open web advertising: the Publisher Ad Server market; the Advertiser Ad Network market and, the Ad Exchange market.
3 Application paragraphs 33-37 (publisher ad servers) and paragraphs 38 to 41 (advertiser ad networks).
6
4 5 6
23.
24.
25.
PUBLIC
a. Publisher Ad Server Market: Publishers use publisher ad servers to manage and track the sale of their ad inventory across different demand sources and transaction types. 4
b. Advertiser Ad Network Market: Advertisers use advertiser ad networks to programmatically buy inventory including web ad inventory from publishers. Advertisers can also use products known as demand side platforms (“DSPs”) to buy publisher inventory programmatically, but as described below, there are important differences between advertiser ad networks and DSPs. 5
c. Ad exchange Market: Ad exchanges are two-sided platform markets which stand between publisher ad servers and advertiser buying tools such as advertiser ad networks. They serve as the digital auction-driven marketplaces where advertisers can bid in real-time on ad inventory made available by publishers. Ad exchange services are purchased by publishers. 6
Not All Forms of Digital Advertising are Substitutes for One Another
Google’s Response assumes that all forms of digital advertising are substitutable with one another. Google’s position is both incorrect and unhelpful.
It is incorrect because it fails to take account of the differing uses for various forms of digital advertising.
It is unhelpful because the claim that all digital advertising is substitutable fails to address the question of substitutability from the perspective of both advertisers and publishers, who require the functionally distinct tools at issue in order to facilitate the programmatic purchase and sale of open web
Application, paragraphs 33 and 34. Application, paragraphs 38 to 41. Application, paragraphs 42 to 46.
7
26.
27.
28.
29.
PUBLIC
advertisements. For purposes of product market definition, it is important to ask the question: substitutable from whose perspective?
In the product market context, substitutable does not simply mean that it is theoretically possible to substitute one thing for another, for example it is correct that the developer of the Grand Theft Auto videogame could substitute a Globe and Mail banner ad for a TikTok ad. Rather, substitutable means that, in the given example, there is a likelihood the developers of the game would see a banner ad in the Globe and Mail as a viable substitute for a TikTok placement, which to be clear they would not.
From an advertiser’s perspective, not all forms of digital advertising are substitutable. Advertisers use different types of digital advertising (e.g., social media ads, web ads, audio ads, etc.) and formats (e.g., banner, in-stream video, native, etc.) for different purposes (e.g., raise brand awareness or as a “call to action” for a specific user), as well as to target different types of consumers who may be categorized by any number of attributes and qualities (age, gender, interest, etc.).
A given advertiser may use a combination of digital advertising forms and places of publication to serve different purposes and/or as complements to one another.
From a publisher’s perspective, various digital advertising forms are substitutable only if the publisher has the technical capability to display that type of ad. For example, for open web publishers, other forms of digital ads (e.g., in-app ads, social media, etc.) are not substitutes because different ad formats require different digital infrastructure for publication and require viewers to access the publisher’s content through their place of publication (e.g., a viewer needs to be within an app to see in-app ads).
Open Channel Digital Ads are not Substitutable for Closed Channel
8
30.
31.
PUBLIC
Google erroneously claims at paragraphs 49-51 of its Response that open and closed channels can be substituted for one another. Google provides various screenshots of advertisements sold through open and closed channels in support of that claim. However, Google ignores important differences in how open and closed channels operate, instead focusing on the appearance of the ad in the examples given.
The primary difference between open and closed channel advertisements is not how they appear, but rather who is selling the inventory and by what means they are sold. In open channels, advertisers and publishers use third party tools to buy and sell ad space, with no constraint on where the advertisement can appear, be it a particular website or platform. Conversely, publishers of closed channel websites, or “walled gardens”, use their own dedicated ad tech tools to facilitate the sale of digital advertising on their platforms. An advertiser who wishes to buy inventory on one of these closed channels platforms must use the tools made available by that specific publisher. Simply put, if you want to buy Facebook social media advertisements, you must use Meta’s tools.
E. Within open web display advertising, direct deals are not interchangeable with programmatic sales
32.
33.
34.
Direct deals involve direct agreements between advertisers and publishers for the purchase and sale of open web advertising. Whatever inventory publishers cannot sell through direct deals, referred to as remnants, they sell programmatically.
Direct deals and programmatic sales are not substitutes for publishers or advertisers.
Unlike programmatic sales, direct deals are resource intensive from both the advertiser and publisher perspective because they require, among other things, manual negotiation and relationship building between sales agents.
9
35.
36.
37.
38.
PUBLIC
From the publisher perspective, direct and programmatic sales are not substitutes. Publishers typically cannot sell all of their inventory by way of direct deals. Publishers’ unsold or remnant inventory is therefore sold programmatically. In that sense, programmatic sales are a complement, not a substitute for direct sales.
Advertisers also view programmatic and direct sales as complements, as they use these two modes of advertising for different purposes. Advertisers use programmatic sales to place ads using psychographic, behavioral, and retargeting data to focus on specific users or audiences. Programmatic access to the open web allows advertisers to target users across multiple websites, following them wherever they browse. Advertisers target users utilizing a number of factors, such as demographics and interests, which they often learn from users’ previous internet behavior. Ads are then served to the consumer across various sites.
Advertisers use direct sales to buy ad space on specific websites (e.g. Wall Street Journal). Any targeting that occurs through direct deals tends to focus on content that an advertiser believes will be of interest to users they wish to reach and placement within the publication, which is relevant to the advertiser’s product. In the Wall Street Journal, for example, an advertiser might place an ad for a mutual fund on the stock market insights page.
Conclusion: the Commissioner’s product markets stand
Having narrowed the focus of consideration in the Application to programmatic open web display advertising, the Commissioner’s three proposed relevant product markets stand up to scrutiny. They are three of the major functionally distinct tools used to purchase and sell advertisements programmatically in open web display advertising.
10
PUBLIC
III. GOOGLE SUBSTANTIALLY OR COMPLETELY CONTROLS THE PUBLISHER AD SERVER AND AD NETWORK MARKETS FOR THE SALE AND PURCHASE OF PROGRAMMATIC OPEN WEB DISPLAY ADVERTISING
39.
Dominance and Google’s Flawed Unitary Two-Sided Market
Google claims that it is not dominant in the unitary two-sided market it proposes. Google’s position regarding market power is inherently flawed, as it is underpinned by Google’s artificial unitary two-sided market formulation.
B. Google Substantially or Completely Controls the Relevant Publisher Ad Server Market and the Relevant Advertiser Ad Network Market
40.
41.
42.
43.
Google is dominant and has a substantial degree of market power in the Publisher Ad Server market for the sale of programmatic open web display advertising, and the Advertiser Ad Network market for the purchase of programmatic open web display advertising.
Google's pervasive and persistent market power in these two separate markets has enabled it to implement programs and policies that have distorted competition by, among other things, steering users to its own ad tech tools through improper ties; impeding its customers from working with rivals; and otherwise denying rivals the scale necessary to compete effectively.
Google's ability to exercise its market power has not been meaningfully constrained by existing or potential competition given, among other things, that barriers to entry and expansion are high in the Publisher Ad Server and Advertiser Ad Network markets, owing in large measure to Google’s anti-competitive conduct.
Google's dominance in each of these markets is exacerbated by the compounding effects of scale and network effects which Google enjoys as a result of, among other things, its very high market shares, not only in the referenced Publisher Ad Server and Advertiser Ad Network markets, but also
11
PUBLIC
in the Ad Exchange and DSP markets. In that regard, in respect of the markets for the services used for the programmatic purchase and sale of open web display advertising, Google’s products have the following market shares:
Table 1 2022 Google market shares for each of the relevant markets Service Market Purchased by Google Product
Publisher ad server
Ad Exchange
publishers
advertisers
publishers
DoubleClick for Publishers / DFP
Google Ads
AdX
Google Market Share
90% (web impressions transacted for Canadian publishers)
70% (web impressions shown to Canadian users)
50% 8 (web impressions shown to Canadian users)
C. Google has Substantial Market Power in the Publisher Ad Server Market
i. Direct Indicia of Google’s Substantial Market Power in the Publisher Ad Server Market
44.
In its Response, Google states that there is no direct evidence that it has
7 Google’s DSP, Display & Video 360, has an over 60% share of the DSP market. 8 Google’s next closest competitor in the ad exchange market has a 6% market share.
12
9
45.
46.
47.
48.
49.
PUBLIC
substantial market power in the publisher ad server market.
However, Google’s Response fails to acknowledge its own ordinary course recognition, internally, of its market power in the publisher ad server market, and its recognition that its publisher ad server, DFP is the de facto, preferred ad server of choice for 90% of publishers and a “must call” for publishers. 9
Google also takes issue with the Commissioner’s allegation that its market power allows it to dictate the terms on which it sells its DFP product to publishers. In particular, Google denies that its market power has allowed it to make changes to degrade DFP’s service quality by implementing its Unified Pricing Rules (“UPR”), despite the fact that those rules eliminated publishers’ ability to set different price floors on different ad exchanges or advertiser buying tools, a feature that Google expressly acknowledged that publishers valued.
Google states that following the introduction of UPR, publishers that have chosen to use DFP to sell their advertising inventory have retained the flexibility to use non-Google ad tech tools. It also says that publishers can make their advertising inventories available to advertisers by using, for example, AdSense or header bidding wrappers on their properties instead.
Google’s Response again misses the mark in that it fails to address the Commissioner’s claim that by implementing UPR, Google took away a feature of DFP that publishers valued, i.e., being able to set different price floors on third-party exchanges and AdX.
Google’s Response also ignores the fact that Google knows that publishers rarely would, if ever, exclude AdX, because it knows that publishers cannot forego access to Google Ads advertiser demand. Subject to certain very limited exceptions, that Google Ads unique demand was (and is) not available to publishers unless they include AdX among the eligible exchanges. So, while
Application, paragraph 119.
13
50.
PUBLIC
Google’s Response may be technically correct in stating that publishers have the ability to exclude AdX, it is misleading insofar as it infers that publishers have a choice that, for all practical purposes, they do not. The commercial reality is that website publishers have perishable display ad inventory to sell when a user visits their webpage. If that inventory goes unsold, the revenue that could have been earned from that inventory is forever lost. Consequently, most publishers are ultimately captive to AdX given the value of must-have Google Ads advertiser demand.
The fact that Google knew publishers valued a feature of DFP and would be unhappy if it was eliminated, but did it anyway, is direct evidence of Google’s market power in the publisher ad server market. The fact that Google did this and did not lose market share in that market only serves to underscore the point.
ii. Indirect Indicia of Google’s Substantial Market Power in the Publisher Ad Server Market
51.
The Application sets out a number of indirect indicia which substantiate the Commissioner’s allegation that Google has market power in the Publisher Ad Server Market, including:
a. DFP’s estimated 90%+ share in that market in Canada, as measured by impressions transacted, a share that Google has maintained since at least 2019;
b. the weak competition Google faces owing to high barriers to entry and expansion, high switching costs, and the prevalence of web display publishers who "single home" (i.e., use only one publisher ad server);
c. the ties Google has created and maintained between DFP, AdX and Google Ads, DFP being the only publisher ad server that provides unrestricted access to real-time bids from AdX, which in turn is the near-exclusive source of Google Ads' unique, must-have advertiser demand; 14
52.
53.
54.
55.
PUBLIC
d. the fact that DFP is the industry-leading "full stack" publisher ad server results in website publishers, with a range of digital ad format inventory for sale, preferring DFP; 10 and
e. the fact that publishers lack countervailing power against Google's market dominance, as no single publisher or group constitutes a significant portion of Google's revenues.
Google offers no response regarding the Commissioner’s “full stack” and “countervailing power” claims.
Google’s denial of the Commissioner’s allegation that Google has a 90%+ market share in the Publisher Ad Server market rests on its flawed unitary product market definition.
With respect to barriers to entry in the Publisher Ad Server market, Google again relies on its unitary two-sided platform market formulation, saying that “barriers to entry and expansion in the properly defined relevant market are modest” 11 Google then goes on to say that even in the publisher ad server market proposed by the Commissioner, barriers to entry and expansion are “not high”. By way of illustration, Google states that “Microsoft, Criteo and Equativ have all grown and attained sufficient state scale to compete successfully.” It also says that many publishers can and have in fact created in-house ad tech tools on a cost competitive basis.
In fact, Microsoft, and Equativ have not grown so as to have gained sufficient scale to compete successfully in the Publisher Ad Server market, as evidenced by, among other things, the 80% to 90% market share Google has been able to
10 Application para 125: “Google's market power in the Publisher Ad Server market is enhanced by the tie it has created [/] maintained between Google Ads, AdX and DFP. DFP is the only publisher ad server that provides unrestricted access to [RTBs] from AdX, which in tum is the near-exclusive source of Google Ads' unique, must-have advertiser demand. The Google Ads/AdX tie gives DFP a unique advantage over all other publisher ad servers, and (by necessity) makes it the default publisher ad server for publishers. DFP has a 100% adoption rate amongst large publishers in Canada. 11 Response, para. 238(a).
15
56.
57.
58.
PUBLIC
sustain over time. 12 Notably, Microsoft, a firm with substantial financial resources and technological “know how”, as well as strong brand equity in technology markets, has been unable to make significant inroads and gain scale in the publisher ad server market.
Google’s claim that “many publishers can and have in fact created in-house ad tech tools on a cost competitive basis” is so general as to be virtually meaningless. Google avoids saying that many publishers can and have in fact created a publisher ad server. Given the time, money and technical expertise required to develop and operate a publisher ad server, very few publishers have done so.
In respect of switching costs in the publisher ad server market, Google says “publishers can and do substitute publisher ad tech tools and services and can sell advertising inventory without using any publisher ad tech tool.” Again, Google’s Response avoids referencing publisher ad server tools. Moreover, Google says nothing about the ease of switching and provides no indication, even at a high level, of the factual basis on which it claims switching costs are minimal. Moreover, Google’s claim that open web publishers can programmatically sell ad inventory without using any ad tech tool, is incorrect.
Finally, Google claims that the tie between the unique advertiser demand through Google Ads and AdX does not give DFP a unique advantage over other publisher ad servers. Google claims that publishers have a wide and increasing ability to mix and match ad tech tools and that their use of DFP is not the only way they can access Google Ads demand. The ties between Google Ads and AdX and between DFP and AdX are addressed below beginning at paragraph 74. However, the fact is that without market power in the publisher ad server market, Google would be unable to tie DFP to AdX.
12 Criteo does not offer a publisher ad server
16
59.
60.
61.
62.
PUBLIC
D. Google has Substantial Market Power in the Advertiser Ad Network Market
i. Direct Indicia of Google’s Substantial Market Power in the Advertiser Ad Network Market
In its Response, Google does not respond to the Commissioner’s allegation that internal Google communications explicitly acknowledge that its advertiser demand is "captive" and state that Google has "virtually no margin or inventory sourcing constraints". 13
In the Application, the Commissioner alleges that in 2014, Google profitably imposed a fee increase of more than 5% on advertisers for its Google Ads service, without meaningful constraint from advertisers switching to alternative products and that on more than one occasion, Google simulated material fee increases (e.g., 15% to 20%) and concluded that raising its prices would increase profits. Google ultimately decided not to impose the increases, not because it was concerned about advertisers switching away from Google Ads, but rather due to reputational and legal concerns.
Google’s Response does not address the foregoing allegations directly. Instead, Google states that it does not charge supra-competitive prices for Google Ads and that Google’s average revenue share for display advertising purchased by advertisers through Google Ads in North America from 2015 to 2024 was approximately 14%.
In the Application, the Commissioner alleges a further direct indicator of Google's market power can be found in the fact that Google is able to offer advertisers an inferior quality advertiser ad network without losing a meaningful number of advertisers. By way of illustration, the Application
13 Application, paragraph 131.
17
63.
64.
65.
PUBLIC
alleges that Google has long restricted Google Ads' ability to bid into rival ad exchanges, at a cost to advertisers in the form of decreased access to inventory, but at a benefit to AdX (and Google).
Google denies the foregoing allegation and claims that advertisers are able to bid for advertising inventory available on non-Google ad exchanges by using Google’s AWBid, gBid and DV360 tools. These claims are without merit. AWBid is addressed below and gBid and DV360 are addressed in paragraphs 82 to 86.
AWBid is not a Viable Option for Most Google Ad Demand
While Google Ads advertisers can use AWBid to buy advertising on third-party exchanges, Google has placed significant limits on that use. When AWBid was launched in 2015, 14 it limited advertisers to bidding on a narrow category of advertising called “remarketing”, which consists of highly tailored ads that show an advertiser’s products to a customer who had previously expressed interest in them. Although Google has somewhat expanded the categories of advertising campaigns eligible for AWBid beyond remarketing, Google Ads customers are still limited to bidding on only a small subset of advertising inventory on rival ad exchanges.
In addition, Google has systematically disadvantaged any Google Ads bids submitted into rival exchanges by making Google Ads charge take rates between 32% to 50% on rival exchanges compared to 15% on AdX. Google charges this fee in addition to any fees charged by the competing ad exchange. These high Google fees mean that rival exchanges cannot expect to win a meaningful volume of bids under the AWBid program, because Google’s higher fees effectively reduce the bid price going to publishers through rival exchanges.
14 AWBid was launched on a pilot basis in 2013.
18
66.
PUBLIC
In view of the limits on the type of advertising that advertisers could purchase with AWBid and the additional fees imposed by Google which disadvantaged bids submitted on rival exchanges, the vast majority of Google Ads’ advertising demand has remained exclusively available to AdX. Between at least 2019 and 2022, “91 percent of [Google Ads] impressions were transacted through AdX.”.
iii. Indirect Indicia of Google’s Substantial Market Power in the Advertiser Ad Network Market
67.
68.
69.
70.
Google claims that there is no such thing as an advertiser ad network and that Google Ads’ market share would be lower than the 70% alleged by the Commissioner if all participants and transactions in the appropriate relevant market were considered.
There is an Advertiser Ad Network market, a fact that Google has itself recognized, as reflected by its own internal communications.
Having denied the existence of the Advertiser Ad Network market for purposes of rejecting the Commissioner’s market share figure, Google then claims at paragraph 245 of its Response that the “supposed advertiser ad network market” does not have high barriers to entry, high switching costs or a prevalence of advertisers who use a single advertiser ad tech tool. Google also claims that scale is not inaccessible to rivals who compete in or seek to compete in the Advertiser Ad Network market.
In terms of scale, Google claims that numerous ad tech providers have attained sufficient scale to compete successfully against Google, including, among others, Meta, TikTok, Amazon and Microsoft. In fact, Meta is a walled garden and TikTok is a social media platform. Neither of those firms compete with Google in providing tools that allow advertisers to access open web display advertising. Amazon offers tools primarily intended for use by sellers on the Amazon platform to advertise their products to consumers. Finally, notwithstanding Microsoft’s very considerable financial and technological
19
71.
PUBLIC
resources, as well as its brand equity in technology markets, Microsoft has not been able to achieve scale in the advertiser ad network market.
Regarding barriers to entry, Google claims there are no significant barriers to entry in ad tech markets, including the advertiser ad network market. The Commissioner disputes Google’s claims and reiterates that barriers to entry in the advertiser ad network market are high. There has been no meaningful entry in that market for many years. Moreover, it is simply untrue that numerous rivals to Google have successfully entered the advertiser ad network market.
IV. GOOGLE HAS ENGAGED IN AND CONTINUES TO ENGAGE IN A PRACTICE OF ANTI-COMPETITIVE ACTS
72.
73.
74.
To justify its abusive conduct, Google relies on a number of alleged justifications including alleged workarounds or pro-competitive rationales. As described below, none of Google’s justifications are supported or credible. Rather they attempt to deflect attention away from the fact that its conduct has the purpose and effect of excluding competitors from entering and/or expanding in the relevant markets as defined by the Commissioner in the Application. Google’s Response conceals that its conduct, individually and collectively, increased the use of its own ad tech tools through distorting the competitive dynamics and diverting customers away from rivals, rather than through improving its own products (AdX, Google Ads, and DFP).
Google’s justifications are insufficient to outweigh Google’s clear subjective intent to exclude competitors or the reasonably foreseeable or expected exclusionary effects of Google’s practice.
Google unlawfully ties its Ad Tech Tools together
Google responds to the Commissioner’s tying allegations in the first instance by saying that advertisers and publishers are not contractually bound to use AdX. However, the Commissioner’s tying allegation is based on the
20
75.
76.
77.
78.
PUBLIC
commercial exigencies resulting from the strictures that Google has imposed on advertisers and publishers, not on contractual obligations.
As described in paragraphs 149 and 150 of the Application, before its acquisition of DoubleClick in 2008, Google had a large captive base of advertiser demand through its position in search advertising. With its DoubleClick acquisition, Google added DFP, a publisher ad server which had a sizeable pool of publisher customers and AdX, a nascent ad exchange, to its ad tech products. Shortly after the acquisition, Google took steps to lock publishers and advertisers into its ad tech ecosystem.
Google configured Google Ads to severely limit advertisers’ ability to bid for publisher inventory on rival ad exchanges, even though it has the technology to enable advertisers using Google Ads to bid on those exchanges. The limits Google places on Google Ads’ capability mean that advertisers have to, except in limited circumstances, bid on open web publisher inventory through AdX. In effect, by forcing Google Ads advertisers to use AdX, Google made AdX an indispensable ad exchange as it was the only way for a publisher to access the Google Ads unique demand.
A significant portion of Google Ads demand comprises smaller advertisers wishing to purchase programmatic web display advertising and only using Google Ads to do so. Cumulatively these smaller advertisers encompass a large portion of the total spend on Google Ads. Aside from its massive data and scale advantages, Google Ads has search as well as display advertising capabilities, and many smaller advertisers come to Google Ads looking to advertise on Google’s widely known search engine. When these budget conscious advertisers want to expand into web display advertising, they are able to do so within Google Ads, and thus Google retains them within its own ecosystem. Using multiple buy-side tools leads to additional costs for smaller advertisers, both in terms of time and money.
On the publisher side, Google has configured DFP such that if publishers want
21
79.
80.
PUBLIC
to receive real-time bids from AdX, including from Google Ads’ advertisers, they have to use DFP.
Google claims that both advertisers and publishers have a number of other ad tech options available to them, such that they are not locked into Google’s ecosystem. In that regard, Google references in various places in its Response proposed options, including on the advertiser side, AWBid, gBid, DV360, while on the publisher side, Google points to AdSense, AdMob and AdX Direct.
It is technically true that advertisers and publishers can use these services. However, given their limited capability and the dynamic created by the cross ad-tech ties described above, the reality is that for most of the trade in open web display advertising, these tools do not represent a commercially realistic alternative to overcome the ties Google has put in place.
B. Google’s Alternative Options for Advertisers are not Viable Workarounds for Google Ads/AdX
81.
82.
83.
As set out above, AWBid has limited utility to advertisers given that Google has placed severe restrictions on the type of advertiser campaign goals (i.e., remarketing) that it can be used to bid on, and given that Google has handicapped AWBid by imposing higher fees on bids placed using AWBid on third-party exchanges.
gBid is a feature of Google Ads that allows advertisers to purchase in-app advertising sold by third party publishers outside of Google’s ad tech tools. Inapp advertising is not interchangeable with open web display advertising and therefore is not a substitute for advertisers who wish to purchase open web display advertising.
DV360 is a demand side platform, or DSP, and is a functionally distinct tool from an advertiser ad network, like Google Ads. DSPs, such as DV360, allow for more customizability, which comes with greater complexity and cost. They typically cater to larger advertisers and advertising agencies that have training 22
84.
PUBLIC
and industry knowledge, who want to exercise more control over their advertising campaigns. By contrast, Google Ads is a seamless, easy to use tool that prioritizes simplicity and, as such, it is the preferred choice among smaller advertisers, who lack the knowledge and resources to utilize a DSP.
Additional differences between Google Ads and DV360 include the following:
a) Advertisers can use Google Ads for both search and display campaigns, whereas advertisers cannot use DV360 to run search campaigns;
b) There is a minimum threshold spend required to use DV360, consistent with larger advertisers possessing higher campaign spend, whereas no such threshold is required to use Google Ads. This minimum threshold precludes some smaller, more budget constrained advertisers from using DV360;
c) DV360 offers a wider variety of reporting metrics than Google Ads, consistent with the needs of sophisticated advertisers to track their campaigns, report to clients or shape overarching business strategies;
d) DV360 charges advertisers a monthly fee (the standard rate is 15%) on the basis of a percentage of their total spend. This is consistent with a large advertiser that would have a large volume of spend to warrant such an approach. Google Ads advertisers generally only pay a publisher when a user takes an action after viewing an ad, such as clicking on the ad. This correlates to the lesser spend per advertiser exhibited by smaller advertisers; and
e) DV360 supports a number of specialized advertising inventory agreements and deal types between buyers and sellers. These include guaranteed deals (deals agreed upon between a buyer and seller up front), non-guaranteed fixed deals (an exclusive, advertiser to publisher relationship for programmatically purchasing inventory in brand-safe environments), and non-guaranteed auctions (where a seller invites 23
85.
86.
PUBLIC
multiple buyers to participate in an auction and make a portion of their non-guaranteed inventory available for purchase at negotiated prices for each buyer). Google Ads does not support any of these deal types.
In light of the functional differences between DV360 and Google Ads, as well as the costs differences, a significant proportion of Google Ad advertisers do not use DV360 and therefore, are unique to Google Ads.
Moreover, even if DV360 were a substitute for Google Ads, which for a significant proportion of Google Ads users it is not, DV360 is still a Google product and access to it does not alleviate the effect of Google’s conduct to lock advertisers and publishers into its ecosystem.
C. Google’s Alternative Tools for Publishers are not Viable Workarounds for Google DFP/AdX
i. Access to Real-Time Bidding on AdX is Essential for Publishers
87.
88.
89.
On the publisher side Google points to AdSense, AdMob and AdX Direct Tags as alternative means by which publishers can access Google Ads demand, without using DFP/AdX. For the reasons set out below, none of these are commercially reasonable alternatives to DFP/AdX.
An overarching deficiency associated with all of these alternatives flows from the fact that, while a third-party publisher ad server can technically accept bids from AdX, the critical real-time pricing functionality through AdX is only available through DFP.
Real-time bidding in auctions allows publishers to sell inventory at variable prices as individual impressions become available. Widespread adoption of real-time bidding offers benefits to advertisers and publishers. Advertisers have access to more inventory and are better able to use identifying target characteristics for each individual impression in real-time. In addition, the
24
90.
91.
92.
PUBLIC
highly interactive nature of real-time bidding helps to generate competitive pressure from bidders, driving up yield for publishers and enhancing efficiencies of the auction process.
The scheme put in place by Google resulted in approximately 99% of web display impressions sold on AdX coming from DFP publishers in 2022. The effects of the tie are multiplied given that most publishers single home their publisher ad server. Therefore, publishers who are, in effect, forced to use DFP to access real-time bids on AdX, also use DFP to manage their open web display advertising sales originating from third-party ad exchanges.
But for Google’s tie, DFP’s rivals would have been able to gain scale and compete against DFP on the merits. Furthermore, the tie gave DFP the degree of control over ad selection logic that it needed to operate its anti-competitive dynamic allocation tactic, as will be discussed further below.
AdSense, AdMob and AdX Direct – Additional Deficiencies
AdSense is not a viable substitute for DFP/AdX. It is a simplified market solution that has limited functionality relative to AdX or DFP. AdSense is designed for smaller publishers who are looking for a non-complex solution to sell their inventory. It lacks many of the features which publishers value in DFP. For example:
a) DFP/AdX offers real-time pricing capabilities, whereas AdSense does not.
b) DFP has web and app inventory, whereas AdSense can only offer web;
c) DFP allows publishers to carry out the contract terms of direct deals, whereas Ad Sense cannot; and
d) DFP offers consistent reporting across all ad serving activities, whereas Ad Sense does not.
25
93.
94.
95.
96.
97.
98.
99.
PUBLIC
Google asserts at paragraph 112 of its Response that most publishers transact through AdSense and not through DFP and AdX. In fact, the volume of sales, as measured in dollars, is far greater on DFP/AdX, as compared to AdSense.
Moreover, even if AdSense were a substitute for AdX and DFP, which it is not, like DV360 on the advertiser side, it is still a Google product and does not alleviate the effect of Google’s conduct to lock advertisers and publishers into its ecosystem.
AdMob can only be used to sell in-app advertising. It is not therefore an alternative for publishers to sell their open web display inventory.
AdX Direct does not provide a workaround to overcome the tie between AdX and DFP. AdX Direct is an outdated way for rival publisher ad servers to connect with AdX to receive bid information from AdX. However, AdX Direct tags do not provide real-time price information to publishers using rival publisher ad servers.
Google claims that it is in theory possible to use AdX without DFP and vice versa; however, as noted in paragraph 78, access to real-time bids from AdX is only available through DFP.
AdX and DFP are products for publishers and it would be highly beneficial to publishers if real-time bidding were opened up to all publisher ad servers. However in or around 2011, to prevent competitors from accessing this technology, Google purchased AdMeld, a competing yield manager with a core functionality of real-time bidding API technology that could be used across exchanges. Google used this purchase to close down this technology in or around 2013, ensuring rivals would never get to use it.
But for the tie between DFP and AdX, DFP’s rivals would have been able to gain scale and compete against DFP on the merits. This innovative real-time bidding function might have been a staple in open web display advertising. Instead DFP has been able to further entrench its substantial market power. 26
PUBLIC
Furthermore, this tie gave DFP the degree of control over ad selection logic that it needed to operate its anti-competitive dynamic allocation program, as will be discussed further below.
Dynamic Allocation was not pro-competitive
100. Google has in its Response weaved two separate but interrelated misleading narratives that disguise the exclusionary nature of dynamic allocation.
101. First at paragraphs 133-150 of its Response, Google paints the picture that dynamic allocation was a feature that was abused by publishers, while Google was merely a passive intermediary that itself was exploited by them.
102. Next at paragraphs 261-264 of its Response, Google claims that dynamic allocation was a pro-competitive initiative that worked to the benefit of publishers and innovation in the industry as a whole.
103. Both of these narratives are a pretext for the true purpose, or the reasonably foreseeable result, of dynamic allocation, which was to distort and pervert auction dynamics for the anti-competitive purpose of self-preferencing AdX and ensuring third party exchanges could not compete with it on a level playing field. This becomes apparent when Google’s development of dynamic allocation throughout the relevant timeframe is contextualized.
104. As detailed in the Commissioner’s Application at paragraphs 54-55, open web display auctions initially were conducted through a “waterfall” process, wherein various exchanges would be called to bid sequentially, until the first bidder that met the publisher’s stated criteria (i.e., floor price) was identified and selected. Priority in terms of the sequence in which the various exchanges were canvassed was based on the historical price a given exchange would pay for an impression, with exchanges historically offering higher prices being called first.
105. The first rendition of dynamic allocation, known throughout the industry as
27
PUBLIC
“first look”, was designed to use this sequential process to advantage Google. Under first look, AdX was always at the front of the line, got access to a greater volume of inventory and won more auctions. This allowed AdX to “cherry pick” higher quality, more competitive inventory, with higher cost per mille or CPM and thus higher revenues flowing to AdX. The exchanges which did not enjoy AdX’s first look advantage were left with less desirable inventory to select. Under first look, third-party exchanges were unable to compete on the merits for impressions and unable to gain scale. Consequently, they became less attractive to both advertisers and publishers, while AdX increasingly cemented its position as a “must have” source of demand.
106. Google claims at paragraph 261 of its Response that first look created value for publishers above what third party exchanges would pay. However, Google fails to mention that the price floor in AdX was based on the highest historical bids from third party exchanges and fails to acknowledge that a third-party exchange might bid higher than AdX for a particular impression (e.g., a rival exchange could outbid its historical average bid). For example, an exchange might have bid $1 historically for a given impression, thus AdX would set the floor at $1, and perhaps win at $2. But that exchange might have had an advertiser willing to pay $3 on this particular occasion, and could have won the impression if AdX did not receive the first look advantage.
107. Seeking to mitigate the exclusionary effects of first look and increase competition for the supply of advertising, header bidding was created by publishers in or around 2015. Google viewed header bidding as an enormous threat because it offered real-time bidding capabilities to other exchanges besides AdX, outside of DFP. Google described this as “giving dynamic allocation away for free”.
108. As a result of header bidding, AdX lost its first look advantage in that it had to compete with other real-time bids, as opposed to historical bids from third party exchanges. Google knew that publishers would benefit from having multiple exchanges competing in real-time but that this would also level the playing 28
PUBLIC
field. In response dynamic allocation evolved into the next rendition, known to the industry as “last look”.
109. In the header bidding process, rival exchanges submitted bids simultaneously not knowing what their rivals were bidding. With last look, DFP gave AdX the ability to see the highest bid from the header bidding auctions. AdX could then use the bid that won through header bidding as the floor in its own auction, and bid just enough to ensure that the impression was won on AdX instead of that rival exchange. Last look thus restored AdX’s real-time bidding advantage. It also created an informational asymmetry which benefitted AdX and enabled Google to again exclude rival ad exchanges by disadvantaging them in the bidding process.
110. Last look also harmed publishers. Had AdX not had an informational advantage it would have had to bid without knowing the highest bid from the header bidding auction. This would have resulted in generally higher bids and more publisher revenue.
111. At paragraph 139 of the Response, Google claims that it took no action to turn first look into last look following the advent of header bidding. Google’s assertion is of no consequence, as it deliberated internally about whether to maintain its “look” functionality and chose to maintain it as last look because it allowed Google to shield AdX from competition from rival ad exchanges. The anti-competitive effects of maintaining its anti-competitive practice were objectively foreseeable and foreseen by Google. In any event, Google’ assertion misapprehends the Commissioner’s allegations.
112. Both first look and last look involve the same self-preferencing prioritization operating in the same manner. Through DFP’s control over the ad selection logic it always ensures AdX is the first exchange that it calls for bids. Therefore, as soon as a publisher is using DFP, AdX has an advantage. The only material difference between first look and last look was the floor price that was entered into the AdX auction, which corresponded with what source of competition
29
PUBLIC
AdX was getting preference over. As mentioned, in first look that source was the highest historical bid from third party exchanges further down in the waterfall. In last look that source of competition was the highest bid, and therefore winner, of the header bidding auction. In both instances, DFP placed AdX in an advantageous position vis-a-vis its competitors by giving it information that those competitors did not have. This allowed AdX to bid just what it needed to take the transaction for itself.
113. In other words, Google’s entire auction set-up allowed it to consistently self-preference AdX over rivals.
114. With respect to sell-side dynamic revenue share, detailed at paragraphs 145-147 of Google’s Response, the combination of sell-side dynamic revenue share with last look, as implemented by Google, heavily exacerbated dynamic allocation’s distortive effects on the auction process. Through these two working in tandem, AdX gets to see all of the other bids and only afterwards adjust its own price to ensure that it wins a competitive impression it might have otherwise lost, after which competitors are shut out. This feature may, as Google states, have increased the match rate, but in so doing, it funnelled transactions toward AdX. If not for the combination of sell side dynamic revenue share and last look, other exchanges would have had a greater opportunity to compete and win these transactions.
115. At paragraph 146 of its Response, Google suggests that sell-side dynamic revenue share and last look had no connection; however, Google knew or should have reasonably foreseen how sell-side dynamic revenue share would interact with last look to its advantage when AdX and only AdX was able to see all other competing bids.
116. At paragraphs 137 of the Response, Google wrongly states that Publishers had the choice whether to enable dynamic allocation. Publishers cannot practically disable or opt out of dynamic allocation within DFP.
30
PUBLIC
117. It may be theoretically possible for a publisher to work around dynamic allocation by using an AdX Direct tag to allow third party publisher ad servers to receive bids from AdX, as this would eliminate DFP – dynamic allocation’s source. However, in reality this is not feasible. These tags do not provide real-time price information and, as detailed above, that leaves publishers no better off.
118. Further, Google states at paragraph 134 of its Response that it simply inherited dynamic allocation as a result of the acquisition of Double Click in 2008, a claim which it utilizes to distract attention away from its anti-competitive intent. In effect, Google implies that it did not develop and modify dynamic allocation throughout the relevant time period by, inter alia, introducing enhanced dynamic allocation, which is untrue.
119. In any event, the fact that dynamic allocation was initially developed by Double Click and purchased by Google does not absolve Google for using it unlawfully.
E. 2016 Position Statement re Dynamic Allocation does not assist Google’s defence to this Application
120. The conclusions reached by the Commissioner in the 2016 Position Statement were based on information obtained more than ten years ago. The Bureau has since obtained additional facts, including more than four million internal records from Google that have cast new light. More specifically, the Bureau has since learned that: (a) publishers cannot practically opt-out of Dynamic Allocation within DFP (contrary to Google's claim), (b) Google has removed the ability for publishers to set different price floors for different ad exchanges; and (c) despite the use of header bidding wrappers, Google maintained its last look advantage until in or around 2019 when it transitioned to the Unified First Price Auction, at which time it continued its anticompetitive conduct, as set out at paragraphs 178 and 186-193 of the Application.
31
PUBLIC
Project Bernanke was anti-competitive
121. Contrary to Google’s assertions at paragraphs 153 and 154 of its Response, Project Bernanke was not pro-competitive or legally innocuous. Rather it sought to and did preserve and enhance Google’s market power in the advertiser ad network and ad exchange markets, as set out in paragraphs 179-185 of the Application.
122. Bernanke was inherently anti-competitive, as Google Ads charged negative take rates on competitive auctions. These negative take rates allowed Google Ads advertisers to win a greater number of these competitive auctions than they otherwise would have, against other buy-side tools that had buyers with a higher willingness to pay than Google. Over time, this allowed Google Ads to siphon this desirable high quality inventory away from competitors, thus strengthening Google Ads’ competitive position.
123. Project Bernanke did not, as Google suggests at paragraph 265 of its Response, reduce “the price of a service in an effort to compete more effectively”, which in this case refers to the take rates charged by Google Ads. As Google itself points out at paragraphs 153-154 of the Response, Google increased its take rates in non-competitive auctions, and used the additional revenue to subsidize its negative take rates in competitive auctions, with Google Ads’ average take rate remaining the same. This scheme allowed Google to win competitive auctions at the expense of its rivals, without harming Google’s overall revenues.
The Unified Pricing Rules were not pro-competitive
124. Google mischaracterizes the Unified Pricing Rules (“UPR”) as pro-competitive to distract from its true purpose which is to ensure that publishers do not set floors that disadvantage AdX in auctions involving multiple exchanges. Google claims at paragraphs 160-161 of its Response that UPR was implemented to make auctions simpler and more fair. However, Google adopted UPR knowing that it would not be well received by publishers, as the implementation of UPR
32
PUBLIC
led to a loss of control over their own inventory and pricing.
125. As detailed by the Commissioner at paragraph 187 of the Application there are a variety of reasons why publishers set different price floors for different exchanges. Some of the primary reasons are as follows:
a. Ad quality: certain exchanges were prone to produce more lower quality ads than others. A publisher could set higher price floors for exchanges associated with producing lower quality ads to increase the likelihood that an exchange with higher quality would win the auction. Notably, around the time UPR was implemented, AdX was seeing an increasing number of lower quality ads;
b. Revenue diversity: publishers may want to use different price floors to ensure that the same exchange, often AdX, wasn’t always winning auctions for their inventory. To that end, they may set lower floors for alternative exchanges sacrificing a small amount of revenue on an individual transaction in order to generate revenue from multiple sources. This ensured greater competition in the industry and more revenue security for publishers overall through having their demand coming from multiple sources;
c. Increased fill rate: certain exchanges might have a better fill rate, meaning that because they have better targeting mechanisms and different policies they are able to fill more impressions for the publisher. A publisher may be willing to lower the floors to give these exchanges a volume discount in order to ensure that more of their inventory is filled; and
d. To increase yield: publishers could set lower price floors for exchanges known to provide better yield.
126. In removing the ability to set differential price floors, Google presupposes that publishers do not value these options. 33
PUBLIC
127. Many publishers showing a preference to set different price floors based on the first three criteria outlined above (ad quality, revenue diversity, and fill rate) did so in an attempt to select exchanges besides AdX. By removing their ability to select based on these criteria, it is plain that Google’s intent was to prevent publishers from weeding out AdX specifically.
128. Google’s claim at paragraph 162 of its Response that the need for publishers to set variable price floors across differing auctions was no longer necessary in a first price auction is incorrect. The first three justifications above remain valid in a first price or real-time bidding auction format. Publishers still place importance on them after Google’s implementation of the Unified First Price Auction in or around 2019.
129. In sum, DFP is a product specifically designed for publishers, and removing the publisher’s ability to set different price floors for different exchanges removed flexibility and control of a publisher over their own inventory and pricing, thus degrading the quality of the product offered to them. Google’s intent in doing so was to remove publishers’ ability to preference rival exchanges, ensuring instead that AdX continued to be at an advantage.
SUBSTANTIAL LESSENING OR PREVENTION OF COMPETITION
130. Google's practice of anti-competitive acts, as described in the Application, has had, is having and is likely to have the effect of preventing and lessening competition substantially in the Publisher Ad Server, Ad Exchange and Advertiser Ad Network markets in Canada. But for Google's practice of anti-competitive acts, barriers to entry in the Publisher Ad Server, Ad Exchange and Advertiser Ad Network markets would be substantially lower. In addition, those markets would be substantially more competitive, including by way of lower prices, enhanced innovation, and higher service quality.
131. In its Response, Google claims that its conduct has not resulted in an SLPC in its proposed unitary two-sided platform market, which it describes as the
34
PUBLIC
relevant market. Google claims there have been a “number of significant innovations, advancements, and new entrants” in the relevant market, which has “evolved rapidly over the years, including recently, and is characterized by increasingly intense competition…”. 15 Google claims that there are “dozens of competitors in the Ad Tech industry today”, all of whom are participants in the relevant market. 16 Finally, Google claims that “vigorous healthy and effective competition exists in the relevant market.”
132. All of Google’s claims regarding the absence of an SLPC are underpinned by Google’s flawed product market definition and are therefore without merit.
133. Google also claims that there has been no SLPC in the relevant product markets identified by the Commissioner. However, Google does not respond substantively to the Commissioner’s SLPC allegations in respect of those markets. Instead, Google reverts to its unitary two-sided platform market argument, stating that the effects in the Commissioner’s proposed markets are “incapable of being properly identified or assessed” as those product markets are incorrect.
134. The effects of Google’s conduct in the Publisher Ad Server, Ad Exchange, and Advertiser Ad Network markets are capable of being properly identified or assessed. That assessment reveals:
a. In the absence of Google's practice, significant new entry and expansion into the Publisher Ad Server, Ad Exchange and Advertiser Ad Network markets would likely have occurred or would likely occur in the future;
b. Google's conduct has raised barriers to entry and expansion, deprived rivals of scale and insulated it from competition, enabling it to exercise a materially greater degree of market power, through materially higher
15 Response paragraph 270 16 Response paragraph 271
35
PUBLIC
prices and materially lower levels of service quality, than would otherwise prevail in the absence of Google's practice;
c. Enhanced rivalry from new entry and/or expansion would likely have resulted and would in the future result in substantially more competitive Publisher Ad Server, Ad Exchange and Advertiser Ad Network markets in Canada;
d. Google's conduct has undermined the ability for publishers and advertisers to make informed choices and exacerbates problems of adverse selection in the auction process, leading to lower quality matches;
e. But for Google's practice, publishers and advertisers would benefit from greater choice, more innovation and lower prices. Google's conduct restricts the ability of publishers and advertisers to effectively multi-home, limiting their ability to take advantage of higher-quality and/or lower-priced services;
f. As a result of Google’s conduct, advertisers pay more for advertising and publishers receive less;
g. Google's practice has intentionally stalled, stifled and deprecated innovative technologies that stood to benefit publishers and advertisers to preserve and enhance its market power.
VI. THE APPLICATION IS NOT TIME BARRED AS CONTENDED BY GOOGLE
135. Google has pursued a systematic campaign of interrelated and interdependent anti-competitive actions to create, enhance and preserve market power in the relevant markets. Google’s anti-competitive actions are all part of a common “practice” that continues to the present day and is not a series of discrete and unrelated activities, as Google suggests in its Response.
36
PUBLIC
136. In any event, Google’s dominant position, practice of anti-competitive acts, and effects over the three years preceding the Application warrant the remedies sought by the Commissioner.
VII. THE ENTIRETY OF GOOGLE’S ABUSE OF DOMINANCE MUST BE ASSESSED UNDER THE ACT AS CURRENTLY IN FORCE
137. The current Act applies both as it relates to the legal test that must be applied by the Tribunal in assessing Google’s conduct and as it relates to the remedies that the Tribunal will be required to dispense, if any. However, the Application meets the test for liability under both the current and former provisions of ss. 79(1) of the Act.
138. The fact that some of the anti-competitive actions and anti-competitive programs that formed part of the single “practice” occurred before section 79 was amended is without legal consequence. Google has over the course of a decade, through the adoption of a series of anti-competitive actions and anti-competitive programs, tied together its different ad tech products, hamstrung rivals’ ability to compete on the merits, and purposely restrained and deprecated innovative technologies that stood to threaten its market power.
139. Google’s same and continuous single practice adapted and progressed as circumstances evolved. What is material for the purpose of assessing the application of the Act is that Google’s “practice” continues through to today and engages section 79 of the Act, as it exists in its current form, including all remedies provided for under the Act.
140. With a finding that a person has engaged in or is engaging in a practice of anti-competitive acts that amounts to conduct that has had or is having the effect of preventing or lessening competition substantially in a market in which the person has a plausible competitive interest, the remedies provided for under the Act, including the AMP provisions found in ss. 79(3.1), are available to the Tribunal. The fact that some of the anti-competitive actions and anti-
37
PUBLIC
competitive programs that formed part of Google’s “practice” first occurred prior to the 2022 legislative amendments does not in any way restrict the Tribunal in the remedies that it may decide to dispense.
141. For example, as it relates to ss. 79(3.1) and the levying of an AMP, Parliament has made it clear that the purpose of an AMP is to promote a practice that is in conformity with the purposes of section 79. The impugned “practice”, both as it existed a decade ago and as it exists today, is not in conformity with the Act and an AMP under ss. 79(3.1) is not only possible but warranted.
VIII. GOOGLE MISCHARACTERIZES THE ADMINISTRATIVE MONETARY PENALTY REMEDY SOUGHT BY THE COMMISSIONER
142. Google misreads the Act and the relief sought by the Commissioner in the Application, all with a view to lending support to an ill-conceived constitutional challenge to the administrative monetary penalty or AMP provisions of the Act.
143. The Act allows the Tribunal to assess Google’s conduct and award an appropriate AMP amount to promote conformity with the Act, after having assessed several factors. An AMP is ultimately at the discretion of the Tribunal, although the Act circumscribes that discretion by providing general parameters on the AMP amounts that can be awarded.
144. Subsection 79 (3.1) of the Act provides that the Tribunal may make an order requiring a person to pay an AMP amount that it determines to be appropriate, not exceeding the greater of: (a) $25,000,000 and, for each subsequent order under either of those subsections, an amount not exceeding $35,000,000; and (b) three times the value of the benefit derived from the anti-competitive practice, or, if that amount cannot be reasonably determined, 3% of the person’s annual worldwide gross revenues. The continuous duration of the “practice”, which stems back to about 2008, is a relevant consideration in determining an appropriate AMP under the Act.
38
PUBLIC
145. The AMP provisions make it clear that the maximum AMP amount of 3% of a person’s annual worldwide gross revenues is only possible where: (a) three times the value of the benefit derived from the anti-competitive practice cannot be reasonably determined; and (b) after various factors have been considered.
146. Subsection 79(3.1) of the Act provides the Tribunal with discretion to order the payment of an AMP and its amount. In deciding the amount of any AMP, the Tribunal is required to take account of the factors set out in ss. 79(3.2) of the Act. Moreover, the Act expressly provides that the purpose of any AMP ordered against a person “is to promote practices by that person that are in conformity with [ss. 79(1)] and not to punish that person.” The Commissioner acknowledges that though the Application has requested an AMP in the maximum amount provided for in the Act, as a matter of law, any AMP order issued by the Tribunal must conform to the foregoing limits. As such, any AMP ordered by the Tribunal cannot be punitive or constitute a true penal consequence.
IX. THE TRIBUNAL MAY ORDER THE DIVESTITURES SOUGHT BY THE COMMISSIONER
147. The divestitures sought by the Commissioner are reasonable and necessary to overcome the effects of Google’s practice and the Tribunal has jurisdiction under the Act to issue the appropriate remedial order, even though assets of Google necessary for an effective remedy may be located in the United States of America.
148. Commercial activity is increasingly transnational and Canadian Courts have shown a willingness to assert jurisdiction beyond the borders of Canada, provided that the impugned conduct or its effects demonstrate a real and substantial connection to Canada. This is clearly the case here.
149. The Act circumscribes the territorial application of certain provisions to only Canada in some instances but not in others. This is deliberate and in recognition
39
PUBLIC
of the fact that commercial activity is transnational. The operative portion of section 79 of the Act speaks to a practice of anti-competitive acts and remains silent on its territorial application.
150. Google’s acts or conduct and/or the effects of the conduct have a real and substantial connection to Canada for which this Tribunal is entitled to assert jurisdiction. The various Google ad tech tools operate in respect of digital advertising that occurs in Canada and the effects of the conduct have a real and substantial connection to Canada.
151. Where, as here a remedy implicating assets outside Canada is necessary to address the underlying conduct and/or effects of the conduct, the Tribunal has the remedial jurisdiction under the Act to dispense an appropriate remedy to secure compliance. To the extent that a divestiture of Google assets is necessary to overcome the effects of its anti-competitive practice in Canada, the Tribunal is entitled to issue a divestiture order that affects Google and Google assets in the United States of America.
152. A remedial order of the Tribunal can also affect the rights and interests of third parties. Courts in Canada make orders on a regular basis that affect the rights and interests of third parties that have not been impleaded. The intervention rules and the requirements to serve notice on third parties in the different rule sets across the country speak to the fact that the rights and interests of third parties can be affected. The impact on third parties is a matter to be considered by the Tribunal in deciding what remedy is appropriate.
X. CUSMA DOES NOT PROSCRIBE THE JURISDICTION OF THIS TRIBUNAL TO MAKE THE APPROPRIATE ORDER REMEDYING GOOGLE’S ABUSES OF DOMINANCE
153. Contrary to the allegations of Google at paragraphs 35, 279-281, and 291 of its Response, the Agreement between the Government of the United States of America, the Government of the United Mexican States, and the Government
40
PUBLIC
of Canada, dated November 30, 2018, as implemented under the Canada– United States–Mexico Agreement Implementation Act, S.C. 2020, c. 1 (collectively “CUSMA”), does not narrow the remedies available for contraventions of section 79 of the Act. Further and in any event, the administrative monetary penalty and divestiture order sought by the Commissioner are in harmony with CUSMA.
154. CUSMA recognizes that Canada (a) maintains national competition laws that proscribe anti-competitive business conduct to promote competition in order to increase economic efficiency and consumer welfare, and take appropriate action with respect to that conduct; and (b) endeavors to apply its national competition laws to all commercial activities in its territory.
155. CUSMA expressly recognizes Canada’s ability to apply its national competition laws to commercial activities outside its borders, that have an appropriate nexus to its jurisdiction. CUSMA permits this Tribunal to apply the Act to Google’s commercial activities as identified in the Application, these activities having had, having and being likely to have the effect of preventing or lessening competition substantially in geographic markets that include Canada. Google’s abuse of dominance, as identified in this Application, has harmed and threatened to harm commerce in Canada, substantially reduced choice, reduced innovation, and increased prices for products in geographic markets in Canada or that include Canada.
156. Contrary to Google’s allegations at paragraph B of its Response to Demand for Particulars dated February 25, 2025, the financial penalty sought by the Commissioner does not conflict with CUSMA. The criteria the Tribunal may apply in determining administrative monetary penalties are transparently set out in section 79 of the Act. In exercising its discretion, the Tribunal must consider the revenue/profit Google derived from activities affecting Canada, and issue a penalty no greater than this amount, if that amount can reasonably be determined. The remedies sought by the Commissioner have an appropriate nexus to Canada and are necessary to restore competition in Canada, as 41
PUBLIC
Google’s anti-competitive conduct has harmed, is harming, and is likely to harm commerce in Canada.
157. CUSMA does not immunize Google from the divestiture sought by the Commissioner. The Commissioner does not, through this Application, seek to require Google to transfer any source code of software, or any algorithm expressed in that source code, as a condition for the import, distribution, sale or use of that software, or of products containing that software, in Canada. Any relief granted in this proceeding under the Act will be in respect of the contravention of Canada’s competition laws, as maintained in harmony with CUSMA.
CONCLUSION
158. The Commissioner maintains his position in the Application and respectfully requests that the Tribunal grant the relief sought in paragraph 217 of the Application.
DATED AT Gatineau, Quebec, this 28
th
day of March 2025.
ATTORNEY GENERAL OF CANADA Department of Justice Canada Competition Bureau Legal Services Place du Portage, Phase I 50 Victoria Street, 22nd Floor Tel: (613) 296-4470 Fax: (819) 953-9267
Attention: Alexander M. Gay Email: alexander.gay@cb-bc.gc.ca
Donald Houston Email: donald.houston@cb-bc.gc.ca
John Syme Email: john.syme@cb-bc.gc.ca
Derek Leschinsky
42
TO:
PUBLIC
Email: derek.leschinsky@cb-bg.ca Katherine Rydel Email: katherine.rydel@cb-bc.gc.ca
Sanjay Kumbhare Email: sanjay.kumbhare@cb-bc.gc.ca
Lawyers for the Commissioner of Competition
DAVIES WARD PHILLIPS & VINEBERG LLP 155 Wellington Street West Toronto, ON, M5V 3J7 Tel: (416) 863-0900 Fax: (416) 863-0871
Attention:
Kent E. Thomson Email: kentthomson@dwpv.com
Elisa K. Kearney Email: ekearney@dwpv.com
Chantelle T.M. Cseh Email: ccseh@dwpv.com
Chanakya A. Sethi Email: csethi@dwpv.com
Chenyang Li Email: cli@dwpv.com
Lawyers for the Respondents, Google Canada Corporation and Google LLC
43