Thank you for inviting me to give evidence at last week’s meeting of the House of Lords Communications and Digital Committee. This letter responds to your suggestion that I further elaborate on some of the views I expressed during my testimony.
I would like to address three issues in particular:
Although the Competition and Markets Authority (“CMA”) takes a cautious approach and use prudent terms, the findings it makes in its Final Report are rather devastating for Google with respect to its practices in the so-called “ad tech” sector, i.e. the various intermediation tools that are used by publishers and advertisers to respectively sell and buy inventory.
Google’s position (as illustrated in a recent expert report they submitted to the Australian Competition and Consumer Commission or “ACCC”, which I will discuss later), is that the ad tech sector is a crowded and competitive marketplace with the presence of a large number of players that constrain Google’s behaviour. Google also presents itself as a great innovator acting in the best interest of publishers and advertisers, which has never engaged in any exclusionary behaviour to the detriment of rival ad tech intermediaries as a “competitive and open ad tech marketplace helps publishers optimize yield from ad sales”, which in turn “helps web publishers fund the creation of free and original online content, which is good for the open Internet and thus search engines.”
The problem for Google is that the CMA Final Report gives a different picture. It shows that online ad intermediation comprises a series of vertically related markets (e.g., ad serving for publishers, ad exchanges/Supply Side Platforms or “SSPs”, Demand Side Platforms or “DSPs”, ad serving for advertisers) in all of which Google has the strongest position.
The CMA notes that, while vertical integration can give rise to some efficiency benefits (e.g., when it comes to cookie matching), it may also be a source of conflicts of interests. The extent to which such conflicts harm customers depends, among others, on the market power of the intermediary facing the conflict and the degree of information asymmetry between it and its customers. Of all the forms of vertical integration, the integration of a DSP and of a publisher ad server within a single provider (as is the case of Google) was held to be the most problematic.
The CMA examined several ways in which Google has engaged in leveraging practices to build its position in open display. In particular, the CMA found that:
Google can leverage the importance of YouTube as a source of inventory to increase its market power in the DSP market (since 2016 only Google’s DSPs may buy YouTube programmatically). The CMA dismissed Google’s privacy arguments put forward to justify cutting third-party access to YouTube.
Google can leverage its search advertiser customer base through Google Ads (e.g., through the use of default settings) to strengthen its position in open display. Google has made it harder for non-Google publisher ad servers to access Google demand (inter alia by not participating in header bidding), reducing competition in ad serving. The CMA dismissed various arguments Google put forward to justify its decision not to participate in header bidding.
Google historically had a “last look” advantage over header bidders. While this “last look” no longer exists as part of Google’s switch to a single unified first-price auction in 2019, Google has rolled out a series of changes that may still enable it to tilt the balance to its favour, to the effect that publishers have to trust Google.
The CMA noted that Open Bidding was designed in a way to disadvantage third-party SSPs and Unified Pricing Rules were rolled out to increase Google demand’s win rate.
These findings, which largely vindicate my research papers on ad tech, thus show that Google has engaged in vertical leveraging to exclude rival ad tech intermediaries, which is a well-known theory of harm under EU and UK competition law, and for which Google has already been condemned in the past.
If Google claims that, as indicated above, a “competitive and open ad tech marketplace helps publishers optimize yield from ad sales”, which in turn “helps web publishers fund the creation of free and original online content” then, conversely, its anticompetitive behaviour in the ad tech sector must have harmed publisher by preventing them to optimize their yield from ad sales.
This is why, as I indicated during my testimony, I believe that, even if the setting up of a Digital Market Unit and the adoption of codes of conduct are desirable objectives, that should not detract the CMA from launching proceedings against Google on the basis of the evidence of wrongdoing they have identified in the context of their market study. A finding of infringement would in turn allow the victims of Google’s behaviour to obtain redress in court for the damages they have suffered.
In this respect, it is important to note that in its Final Report the CMA observes that:
“In relation to enforcement action, the CMA is currently actively considering possible cases in the digital sector, drawing on the work of the market study. More broadly, the CMA will also continue to consider any examples of digital platforms exploiting their market power or otherwise engaging in anticompetitive conduct, and will stand ready to take enforcement action where it identifies evidence of anti-competitive conduct.”
I would strongly support such possible enforcement actions.
In response to questions 5 and 6 that you addressed to Google following their appearance before the Communications and Digital Committee, Google refers to an expert report prepared by Messrs. Bitton and Lewis in which they claim to clear certain “misconceptions” about Google’s ad tech business and in particular several mistaken propositions that Dimitrios Katsifis and I would have made in our research papers. In our response to the Bitton and Lewis report, we show that it does not contain a single valid criticism of our analysis.
One of the most fascinating aspects of Google’s answer to question 5 is that, when responding to our criticism that its dominant presence across the ad tech ecosystem has created conflicts of interest, Google quotes the Bitton & Lewis Report, according to which “as a search engine and vertically integrated ad tech provider, Google also has incentives to balance the interests of all ecosystem participants—users, publishers, and advertisers—and promote the long-term viability of the open Internet.”
Besides the fact that this claim is in tension with the findings of the CMA that Google’s vertical integration may be a source of conflict of interest, in our submission to the ACCC we noted that one could wonder whether it is possible – let alone efficient – for Google to balance the interests of competing stakeholders. Publishers wish to sell their inventory at the highest possible price, while advertisers want to buy inventory at the lowest possible price, so their interests are inherently in tension. In this respect, in our response to the Bitton & Lewis Report, my co-author and I observed that:
“When it sells its ad serving software to publishers, Google is a service provider, acting like a publisher’s real estate agent. A real estate agent you hire for selling your house should aim to sell it at the maximum price, not to try at the same time to optimize for the buyer and help her buy at the lowest price. Suggesting that DFP has to take into consideration the interests of buyers, e.g., Google Ads, would be peculiar. Conversely, the DSP’s function is to buy inventory for advertisers at the lowest possible price, not to balance the interests of the advertiser with those of the publisher. To continue the analogy mentioned above, the DSP is like a real estate agent of the buyer, whose aim is to buy a house at the lowest price, not to try to optimize for the seller and help her sell at the highest price.”
We further observed that
“the concept of Google acting as a form of neutral referee that stands sober above the petty fights of various myopic stakeholders in order to ensure the long-term viability of the ecosystem is hardly credible. The reason is that, unlike Uber, Google is not only the platform where different third-parties transact; at the same time it is one of the transacting parties (e.g., in the form of Google Ads and DV360). Suggesting that Google, in its capacity as an ad exchange, balances the interests of sellers with those of buyers – including its own interests as an ad buyer – amounts to an admission of a conflict of interests on Google’s behalf.”
It is for that reason that in its Final Report the CMA observes that:
“there is a strong case for the power to introduce separation and access interventions in the open display market. The strong position of Google’s publisher ad server, SSP and DSP, and its unique access to Google’s ad inventory, means that each of these businesses potentially faces a conflict of interest, potentially acting on the advertiser side, on the publisher side and on Google’s own account.
Further, Google can increase its market power by tying access to YouTube to use of its DSP services. Based on our analysis, we consider that there is a case for two broad forms of intervention to address these concerns in open display advertising:
• Separation of the function of ad serving from the advertising advisory function (DSP), where the ad server has market power; and
• The prohibition of a DSP restricting access to its inventory, where that inventory is sufficiently important to generate market power for the DSP.”
Finally, I should say that I am not impressed by Google’s attempts to justify its introduction of uniform price floors as in response to question 6 for the reasons my co-author and I explained in our response to the Bitton & Lewis Report.
The advent of the Internet and the fast development of digital platforms operating a two-sided business model, such as Google and Facebook, triggered significant changes to the news publishing industry. While digital platforms have allowed news publishers to reach new audiences, they have also had a negative impact on their ability to monetize their content. As a result, the number of newspapers has shrunk, and newsrooms have thinned. In the absence of sustainable sources of income, the future of the news publishing industry is bleak.
As I discuss in one of my research papers, news content is undoubtedly an input for digital platforms. Google, for instance, uses news in multiple products, such as (i) its Search engine (which return responses to news-related queries on its Search Engine Results Page (“SERP”)), (ii) the Google News tab (a news aggregator, which presents a continuous flow of news articles); (iii) the Google news carousel on mobile devices (comprises news articles and appears at the top of Google’s SERP in response to news-related search queries), and (iv) Google Discover on mobile devices, which is a personalized content feed that serves content (news, articles, video etc.) to a user before she even enters a query. Google does not produce any journalistic content, but rather curates and arranges content produced by news publishers based on computer algorithms.
While news content is an input used by digital platforms to satisfy the demands of their users, digital platforms offer newspapers referral services for their content without monetary compensation. Thus, thanks to Google or Facebook, newspapers like The Guardian or The New York Times may be able reach audiences they might not have been able to reach. Traffic referral from digital platforms is important for news publishers as it can be monetized in two ways: (a) it creates advertising opportunities and (b) may also generate subscriptions as Internet users repeatedly hitting a paywall may decide to take a subscription to have access to a news publishers’ content. Moreover, traffic is usually an important factor in determining the ranking of a news publisher’s content by the digital platform’s algorithm.
Given their complementary nature, news publishers and digital platforms should be able to enjoy a mutually profitable relationship. This is unfortunately not the case for several reasons, all linked to the power imbalance between news publishers and digital platforms.
In the first place, to the extent that digital platforms use news content as an input, they do not pay anything for it, whereas news content is critically important to their success. Imagine, for instance, a scenario where Facebook posts or Twitter feeds could no longer contain hyperlinks to a much-discussed news story. Would they retain the same attractiveness to users, in particular those using these platforms as a source of news? That very point was eloquently made in a Google-sponsored study submitted to the ACCC:
“Aggregator platforms, including search engines, are, however, also complements to the underlying content, just as a distributor is a complement to a manufacturer of automobiles. Without content, irrespective of who generated that content - platform users, third parties, or the platforms themselves - the aggregator platforms would have no reason to exist. Were the supply of content to dry up, there would be no Google News; were there no content worth embedding, there would be far less value to the owners of Facebook, Twitter and LinkedIn. That is, the viability of the platforms depends on the viability of the suppliers of content, including both news content and user-generated content.”
Yet, the value extracted from the traffic flowing through the platform thanks to news publishers is kept entirely by the former. In order to understand this free-riding problem, it is helpful to examine in greater detail how platforms monetize directly and indirectly such traffic.
Platforms directly monetize news-related traffic when they insert an ad between two posts or tweets that refer to the last important news story. For instance, Google Discover directly monetizes news content through the display of ads between the various news articles in the Discover feed. So-called “Discovery ads” are visually appealing, native ad formats targeted to users according to their interests, not their search intent, much like native ads offered by Facebook. News content is also a key input for social networks, such as Facebook, which for some users is their main source of news.
Platforms also monetize news-related traffic indirectly, by keeping users in Google’s ecosystem and collecting data that can be used to refine the platform’s algorithm and to provide unparalleled targeted advertising services. Moreover, when Internet users rely on a platform as a source of news, they are encouraged to use the platform as its default gateway for other uses, unrelated to news consumption (e.g., for online shopping), which will offer the possibility of direct monetization (e.g., in the form of search advertising in the case of Google). As noted in the Cairncross Review, for platforms “the content provided by newspapers is a reason for people to visit their sites and then, potentially to look at or use other content or services they offer.” The possibility of such spill-over effects seems to have been acknowledged by Google on multiple occasions. For instance, in a testimony to the FTC, former CEO of Google Eric Schmidt observed:
“So if you, for example, are an academic researcher and you use Google 30 times for your academics, then perhaps you’ll want to buy a camera…So long as the product is very, very, very, very good, people will keep coming back…The general product that creates the brand, creates demand and so forth. Then occasionally, these ads get clicked on.”
The value Google places on traffic is further demonstrated by its willingness to “pay for traffic”, i.e. pay Original Equipment Manufacturers, such as Apple, significant amounts to have Google Search installed as the default search engine in their devices.
However, all the value generated by the traffic flowing through the platform – be it advertising revenues, unrivalled access to data or spill-over benefits – is not shared with the publishers. In other words, the news publisher that produced – sometimes at great expense – the story from which the platform benefits will not be compensated for its creative effort.
Of course, platforms could argue that such direct and indirect monetization of the traffic flowing through is a form of consideration for the referral services rendered to the news publisher. However, reality is more complex. Platforms engage in free-riding when the following conditions are present: (i) they benefit from the traffic flowing and (ii) news publishers, i.e. the “destination” and ultimate cause of the traffic, are either entirely prevented from or significantly hindered in monetizing the traffic. News publishers may never get the chance to monetize traffic in the first place, because the user is encouraged to stay on the platform. As described in a recent US paper, formats that are integrated by platforms, such as the AMP format for Google, described below in greater details, result in lesser monetization for publishers.
In the second place, to the extent that digital platforms distribute content, they may do it in ways that are not necessarily advantageous to newspapers.
First, while digital platforms generate valuable traffic for news publishers, they also divert traffic away from the publishers’ own delivery means. For instance, while Google Search results generate traffic for news publishers when Internet users click on the hyperlinks to their stories, there will be instances where the users will be satisfied with the amount of information provided by Google in its top stories that appear on its SERP. Some Internet users may also be distracted from visiting news publishers’ websites or their mobile apps, preferring instead to obtain their news from Google search, including the news snippets it features. For instance, the Cairncross Review refers to a study revealing that “42% of adults admit that they often skim headlines on social media without clicking through the relevant story.” In all these instances, while news content is used by digital platforms, the producers of such content are deprived of traffic and the inherent value it represents (in terms of advertising opportunities, converting users to subscribers, producing higher rankings).
Second, platforms may leverage their dominance in their core market to effectively coerce their customers (the publishers) to adhere to a policy that benefits the platform but is fundamentally at odds with their own interests. Google required publishers to produce articles that are consistent with its Accelerated Mobile Pages (“AMP”) standard, a new “stripped-down” format aimed at achieving higher speed for mobile pages, on pain of having their content excluded from Google’s “news carousel” that appears on top of the screen of mobile devices in response to news-related search queries and demoted on Google’s SERP. Because unlike traditional mobile pages AMP pages are loaded on and served from Google servers, Google can maintain readers in its environment, as well as collect all the data generated on such pages. These examples show that Google can leverage its market power in search to coerce news publishers to adopt policies that harm them.
Third, digital platforms commoditize news content. Search engines, news aggregators and social networks disintegrate online newspapers into a collection of single item stories, which is detrimental to news brands. This is the case because the strength and attractiveness of a newspaper is not only determined by the quality of its individual stories, but also by the strength, diversity and organisation of its content. As pointed out by Susan Athey and her co-authors:
“We also see that Google News is a substitute for articles accessed through direct navigation – highlighting that large outlets lose some of their curation role. If part of the long term incentive for news outlets to maintain their brand comes from the way they curate news, through the selection of articles to highlight prominently on their landing pages, then the fact that Google News is in effect selecting what articles from each outlet to highlight on the Google News home page may decrease the incentive of publishers to invest in the quality of their curation and thus their brand. This is an example of a broader concern publishers articulate surrounding aggregators and social media: they worry that they are being “disintermediated” and “commoditized,” consistent with a decreased ability to differentiate their products in the eyes of consumers, as their content is accessed instead through an intermediary.”
As the ACCC observed in its Final Report on its Digital Platforms Inquiry, “news accessed through digital platforms is ‘atomised’ – presented as individual pieces of content alongside other news and non-news material.” This atomisation “directly affects the consumption of journalism on digital platforms by weakening the association between news and its source.”
Fourth, digital platforms may allocate traffic in a way that distorts competition to the detriment of quality newspapers. For instance, Google does not rank news stories on its SERPs based on the originality or quality of these stories. Thus, a story that took six months to an investigative journalist of The Times of London to produce may be ranked below a rehash of that story posted minutes later by a rival publication. That distorts competition between news producers and devalues journalistic investment. While in late 2019 Google announced it would adjust its search algorithm to boost what it considers “significant original reporting”, it still remains to be seen whether this will make any practical difference.
This is not to say that all news publishers have failed since the advent of digital platforms. The new paradigm created by these platforms has forced news publishers to be more creative and, in some cases, re-think their business models, but the above discussion shows that to the extent newspapers and digital platforms are in a vertical relationship, this relationship is not healthy. Digital platforms free ride on the newspapers’ creative content (as they do not pay for the significant value they draw from it), disintegrate it into disparate news items, commoditize it, and ultimately devalue it. That is not surprising considering that digital platforms are attention brokers for which displaying content that informs readers is not an end in itself, but merely a way to draw attention that can be resold to advertisers. Whether the content displayed is of superb or moderate quality, original or borrowed does not ultimately matter as long as it attracts eyeballs.
The problem is that publishers lack the means to resist the platforms’ exploitative conduct as their reliance on platforms as a source of traffic leaves them in a weak bargaining position. For instance, a publisher could in theory have stopped Google’s freeriding by using the appropriate technology tools to prevent Google from crawling and indexing its content. However, that approach would be ultimately self-defeating as Google’s referred traffic would go to other publishers without having any disciplinary effect on Google’s conduct. While digital platforms are usually dominant on their core market, news publishers face numerous competitors and have a collective action problem.
Three conclusions can be derived from the discussion above:
First, Google and Facebook’s claim that they draw little financial benefit from news content is simplistic because it only takes into account the direct benefits they draw from news content through advertising. This claim, however, ignores the vast indirect benefits that they draw from news content as illustrated above. That explains why, for instance, Google offers so many news products on its platforms.
Second, Google and Facebook’s claim that they compensate news publishers for the use of their content by referring traffic is not as simple as it may look. For instance, while digital platforms generate valuable traffic for news publishers, they also divert traffic away from the publishers’ own delivery means. Their practices may also lead to the commoditization of news content.
Finally, although news content is very valuable for the platforms, news publishers are unable to monetize it because they have no bargaining power. For instance, while Google does not strictly need The Guardian’s content (as there is a range of reasonable alternatives to that content), The Guardian cannot do away with Google as there is no alternative to Google.
This is why I strongly believe that a mandatory framework for negotiations is needed to ensure that news publishers are properly rewarded for the use of their content by digital platforms. This framework should be mandatory because the Australian experience suggests that otherwise Google and Facebook will simply refuse to participate or engage in delaying tactics. After months of unproductive efforts on behalf of the ACCC to get Google and Facebook to the negotiating table, the Australian government eventually directed the ACCC to develop a mandatory code of conduct to address bargaining power imbalances between digital platforms and media companies.
While I understand that the DMU will have among its duties to negotiate one or several codes of conduct which would apply to Google and Facebook, I believe that in light of the significant financial problems experienced by the UK press negotiations between the platforms and UK news publishers cannot wait. To the extent that a mandatory framework for negotiation should be put in place, it should therefore be done sooner rather than later.
15 July 2020
 As I indicated during my testimony, I advise several news publishers. The views expressed in my oral testimony as well as in the present written submission are strictly personal.
 Daniel Bitton and Stephen Lewis, “Clearing Up Misconceptions about Google's Ad Tech business”, 5 May 2020, available at: https://www.accc.gov.au/system/files/Google%20-%20Report%20from%20Daniel%20Bitton%20and%20Stephen%20Lewis%20%285%20May%202020%29.pdf (the “Bitton & Lewis Report”).
 Id. pages 35 et seq.
 Id. page 10.
 Competition and Markets Authority, Online Platforms and Digital Advertising, Market Study final report, 1 July 2020, available at: https://assets.publishing.service.gov.uk/media/5efc57ed3a6f4023d242ed56/Final_report_1_July_2020_.pdf, page 266 et seq.
 Id. page 20.
 Id. page 28.
 Id. page 281.
 Id., pages 285 et seq.
 Id. pages 286-287.
 Case AT.39740, Google Search, 27 June 2017, available at https://ec.europa.eu/competition/antitrust/cases/dec_docs/39740/39740_14996_3.pdf; Case AT.40099, Google Android, 18 July 2018, available at https://ec.europa.eu/competition/antitrust/cases/dec_docs/40099/40099_9993_3.pdf.
 CMA Final Report, page 436.
 Damien Geradin and Dimitrios Katsifis, “Competition in ad tech: A response to Google”, Submission to the ACCC, 3 June 2020, available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3617839.
 Bitton & Lewis Report, page 3.
 Damien Geradin and Dimitrios Katsifis, “Competition in ad tech: A response to Google”, supra note 13, page 18.
 CMA Final Report, pages 27-28.
 Damien Geradin and Dimitrios Katsifis, “Competition in ad tech: A response to Google”, supra note 13, pages 78 et seq.
 Damien Geradin, “Complements and/or Substitutes? The Competitive Dynamics between News Publishers and Digital Platforms and What It Means for Competition Policy”, available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3338941.
 H. Ergas et al., Impact of news aggregators on public interest journalism in Australia, May 2018, page 37.
 Ginny Marvin, “New native Discovery ad campaigns from Google monetize Discover feed for first time”, Search Engine Land, 14 May 2019, available at https://searchengineland.com/google- discovery-campaigns-discover-feed-316920.
 It is also claimed that digital platforms may use news content in order to develop their artificial capabilities.
 The Cainrcross Review, A Sustainable Future For Journalism, 12 February 2019, available at https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/779882/02 1919_DCMS_Cairncross_Review_.pdf, page 37.
 See FTC Staff leaked memo, available at http://graphics.wsj.com/google-ftc-report/.
 For instance, Google is reported to pay Apple more than $ 7 billion annually to be the default search engine on iOS. See John Koetsier, “Apple Could Cost Google $15 Billion By Buying DuckDuckGo, Analyst Says”, Forbes, 8 June 2020, available at https://www.forbes.com/sites/johnkoetsier/2020/06/08/apple-could-cost- google-15-billion-by-buying-duckduckgo-analyst-says/.
 See http://www.newsmediaalliance.org/wp-content/uploads/2020/06/Final-Alliance-White-Paper-June-18- 2020.pdf
 The Cairncross Review, supra note 23, page 32.
 See Kristopher Jones, “Google AMP Reviewed: How Has It Changed SEO?”, Search Engine Journal, 21 August 2017, available at https://www.searchenginejournal.com/google-amp-review-seo/209324/#close; “A letter about Google AMP”, available at http://ampletter.org. Most recently, in May 2020, Google announced it will remove the AMP requirement from Top Stories eligibility, as part of a broader Search ranking change that will incorporate “page experience metrics” into Google’s ranking criteria. See “Understanding page experience in Google Search results”, Google Search, available at https://developers.google.com/search/docs/guides/page-experience. According to Google, any page that passes the eligibility test could be listed at the Top Stories carousel. This ambiguous post however does not specify in concrete terms which criteria non-AMP pages will have to meet to be included (with certainty) in the Top Stories carousel.
 S. Athey, M. Mobius & J. Pal, “The Impact of Aggregators on Internet News Consumption”, 11 January 2017, available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2897960, pages 15-16.
 Australian Competition and Consumer Commission, Digital Platforms Inquiry, Final Report, June 2019, available at https://www.accc.gov.au/system/files/Digital%20platforms%20inquiry%20-%20final%20report.pdf, page 350.
 Patricia Nilsson, “Google adjusts search algorithm to boost original journalism”, Financial Times, 12 September 2019, available at https://www.ft.com/content/e27ca6c6-d57c-11e9-a0bd-ab8ec6435630.