Regulatory Comments

ICLE Comments to CMA’s SMS Investigation into Google’s General Search and Search-Advertising Services

Introduction

We appreciate the opportunity to comment on the Competition and Markets Authority’s (“CMA”) investigation into Google’s general search and search-advertising services.[1] While we recognize the importance of ensuring competitive digital markets, we respectfully submit that several key market developments warrant careful consideration before any strategic market status (“SMS”) designation is applied.

As our comments explain, the digital-search landscape of 2025 differs markedly from the one analyzed in the CMA’s 2020 Online Platforms and Digital Advertising Market Study.[2] Some recent developments particularly merit the CMA’s attention. First, the emergence of artificial intelligence (“AI”) has fundamentally altered how users discover information online. Services like Perplexity AI attract users through novel approaches to information retrieval.[3] These services represent not merely incremental improvements to traditional search but fundamentally different paradigms for information discovery. Second, user behavior has fragmented across multiple specialized search channels. Third, the competitive dynamics of search advertising have evolved significantly. While the CMA notes that UK search-advertising spending grew to £15 billion in 2023,[4] this figure should be contextualized within the broader digital-advertising ecosystem. Major platforms like TikTok and Amazon have rapidly grown their advertising businesses,[5] offering advertisers increasingly sophisticated alternatives to traditional search advertising.

These developments raise important questions about whether Google’s current market position meets the statutory criteria for “substantial and entrenched” market power required for SMS designation.[6] While Google remains a significant player in search, the evidence suggests a more dynamic and contested market than historical market-share statistics alone might indicate.

Other factors also cut in favor of a cautious approach to this case. For a start, the CMA’s market study asserts that economies of scale, network effects, and data advantages prevent rival search engines from competing effectively with Google. This assumption, however, is rooted more in regulatory orthodoxy than in robust empirical evidence. While the study highlights factors such as the high fixed costs of web indexing, the self-reinforcing benefits of user data, and Google’s extensive query history as sources of competitive entrenchment, these claims lack rigorous economic validation. Indeed, as our comments explain, the notion that digital platforms enjoy perpetually increasing returns to scale is, at best, overstated. Contrary to the CMA’s premise, real-world financial data from major tech firms suggests that most costs—such as infrastructure, labor, and fulfillment—scale with output, rather than diminishing in proportion, undermining the idea of insurmountable cost advantages.

Furthermore, the oft-cited “data network effects” theory—whereby superior access to user data allegedly entrenches incumbency—remains largely speculative. Empirical research, including work by Catherine Tucker and others, finds little evidence that accumulating vast troves of data leads to runaway advantages. Studies show diminishing returns to additional data in improving algorithmic accuracy, with no clear link between data scale and competitive dominance. If data-driven economies of scale were as decisive as claimed, new entrants leveraging AI-driven search or alternative data sources would not pose a growing challenge to incumbents. Ultimately, the CMA’s conclusions rely on intuitive but unproven assumptions, rather than concrete evidence. This highlights the need for a more rigorous approach before endorsing interventions that could distort market dynamics.

Moreover, recent developments in AI services and applications pose a significant threat to Google’s dominance in the realm of general search services. The “AI revolution” will certainly offer new avenues for competition. AI assistants such as ChatGPT, Siri, Claude, and Grok—along with AI-powered search engines like ChatGPT Search, Perplexity, and Bing—are becoming significant rivals to Google. As AI becomes more accessible, smaller companies and startups could develop alternative search engines or interfaces that challenge Google’s leadership.

Finally, the CMA has much to learn from international jurisdictions, particularly the EU. The Digital Markets Act (“DMA”) and its regulations, including the self-preferencing ban and choice screens, have had unintended consequences for both businesses and consumers—reducing functionalities and diminishing visibility for smaller players like hotels. For instance, Google’s removal of certain features and alterations to search results, while purportedly aimed at promoting competition, have instead benefited intermediaries and created a less intuitive user experience. Similarly, laws requiring platforms like Meta and Google to pay media outlets to link to news stories have largely backfired, favoring large outlets at the expense of smaller, independent ones.

I. The Dynamic Nature of Search Competition

The rapid evolution of digital technologies has called into question traditional approaches to measuring market power in the search industry, particularly with regards to Google’s longstanding dominance. Relying solely on historical market share and legacy metrics fails to capture the emerging competitive landscape, where advances in AI and specialized search services are fundamentally reshaping user behavior and information discovery. Given this, the CMA should adopt a forward-looking and nuanced framework that reflects the dynamic realities of the digital economy.

A. Appropriate Metrics

The Invitation to Comment (“ITC”) emphasizes Google’s historic market share in general search, noting that it has persistently remained above 90% in the UK.[7] Traditional market-share metrics may, however, no longer capture the competitive dynamics of information discovery and digital advertising. Three key considerations challenge the notion that Google’s position constitutes “substantial and entrenched” market power under the SMS criteria.

First, the statutory requirement for SMS designation demands evidence of market power that is both substantial and entrenched.[8] This forward-looking test requires more than historical market-share analysis; it requires confidence in the durability of market power against emerging competitive threats. The rapid evolution of AI-powered search alternatives challenges this durability in fundamental ways.

Traditional search engines rely on indexing web pages and ranking them based on various signals. New AI-powered services, by contrast, can synthesize information directly, thereby bypassing traditional search-result pages entirely. This represents a structural shift in how users can access information, not merely an incremental improvement to existing search technology. The CMA’s investigation should carefully consider whether historical advantages in web indexing and click-through data remain as relevant to future competition as they were in the past.

Second, the boundaries between “general search” and other forms of information discovery are becoming increasingly fluid. The CMA’s proposed market definition focuses on “a service that searches the world wide web and returns information”.[9] But modern AI assistants and specialized search services often fulfill users’ information needs without necessarily searching the web in real time or returning traditional search results.

The bright line between “general search” and other information-discovery services has blurred significantly. Consider these scenarios:

• A user asks an AI assistant “What’s the best Italian restaurant near me?” The assistant draws on web data, user reviews, and real-time information to provide a synthesized answer.
• A user types the same query into a traditional search engine, receiving a mix of organic results, maps, and reviews.
• A user opens a specialized app like TripAdvisor or Instagram to find Italian restaurants through location-based search.

All three scenarios serve the same user needs, yet the CMA’s current market definition might only capture the second scenario. This raises a fundamental question: should the relevant market be defined by user needs (finding information) or by technical methodology (searching the web)?

Third, while the CMA correctly identifies certain historical barriers to entry in search markets,[10] technological developments are altering the fundamental nature of these barriers. The emergence of large language models (“LLMs”) and AI technologies means that new entrants can potentially offer compelling search alternatives without first replicating Google’s entire infrastructure. These market-definition challenges have direct implications for SMS designation:

• If the market is defined by user needs, rather than technology, the competitive landscape appears significantly more fragmented;
• If new business models break the connection between search and advertising, market power in one activity may not translate to the other; and
• If AI assistants constitute a new form of search, historical advantages in web indexing and click data may become less relevant to future competition.

The CMA’s guidance requires market definition to be “forward looking”.[11] Given the rapid evolution of AI technology, any market definition should anticipate further convergence between traditional search, AI assistants, and specialized discovery services.

B. AI Disruption

Modern LLMs can now generate high-quality responses to user queries without relying exclusively on historical click-and-query data.[12] This represents a fundamental shift in search technology. While search-interaction data remains valuable, it is no longer the only path to deliver relevant results to users. New entrants can leverage alternative training methods and data sources to achieve competitive quality in their search results.

Moreover, vertical-search providers often possess deeper, more relevant data in their specific domains than general search engines could obtain through click data alone. A travel platform’s direct booking data, for instance, provides richer signals about user preferences and outcomes than general-search queries about travel. This specialized data advantage challenges the notion that general-search data creates an insurmountable competitive barrier.

As users increasingly begin their queries on specialized platforms, the value of general-search click-and-query data may be declining in certain categories. When users bypass general search entirely for specific types of queries—such as product searches on Amazon or restaurant searches on specialized apps—the competitive advantage of historical search data diminishes accordingly.

Recent developments in AI services pose a significant threat to Google’s dominance in general-search services. The ITC implies that it considers such services to be an “avenue of investigation” in connection to Google’s potential SMS that AI assistants such as ChatGPT, Siri, Claude, and Grok—along with AI-powered search engines like ChatGPT Search, Perplexity, and Bing—now compete with Google Search, potentially undermining its market share and power.

Consider, for instance, the fact that the success of OpenAI’s ChatGPT in late 2022 led to a (apparently) hurried response from Google, which ultimately fell short of expectations and resulted in a 9% decline in parent company Alphabet’s stock price. As reported by Reuters:

Google has been on its heels after OpenAI, a startup Microsoft is backing with around $10 billion, introduced software in November that has wowed consumers and become a fixation in Silicon Valley circles for its surprisingly accurate and well-written answers to simple prompts. …

Bard’s error was discovered just before the presentation by Google, based in Mountain View, California.

“While Google has been a leader in AI innovation over the last several years, they seemed to have fallen asleep on implementing this technology into their search product,” said Gil Luria, senior software analyst at D.A. Davidson. “Google has been scrambling over the last few weeks to catch up on Search and that caused the announcement yesterday (Tuesday) to be rushed and the embarrassing mess up of posting a wrong answer during their demo.”

Microsoft shares rose around 3% on Wednesday, and were flat in post-market trading.

Alphabet posted a short GIF video of Bard in action via Twitter, promising it would help simplify complex topics, but it instead delivered an inaccurate answer.[13]

While anecdotal, this case powerfully illustrates how advances in the AI sector can disrupt adjacent areas of the digital economy, including search services. Quantitative evidence supports this assertion, as well. For example, The Economist reports that 8% of Americans now consider ChatGPT to be their “go-to search engine”.^[14]

This shift may be attributed to the fact that, for certain types of queries, AI chatbots and search services are often more effective at delivering direct and concise answers, summarizing information, or addressing complex questions in a conversational manner. Consequently, this shift could diminish traffic to Google’s search engine, as users may be less likely to click on a list of search results when they can receive comprehensive answers directly from an AI system. Another possible advantage of these AI interfaces is that they do not (at least, not yet) feature advertising or promoted queries,^[15] which can be a nuisance for some consumers.^[16]

 FIGURE 1: Market Share of Leading Desktop Search Engines Worldwide (Jan. 2015-Jan. 2025)

SOURCE: Statista

Moreover, search services that currently lag behind Google could strengthen their competitive position by integrating AI into their offerings. The introduction of a chat feature in Bing, powered by OpenAI’s GPT-4, has already demonstrated success in attracting users. Reports indicate an increase in user engagement and browsing time on Bing following the launch of the new AI capabilities.^[17] According to Statista, Bing has seen its market share in search increase, albeit modestly. In January 2025, it reached a worldwide market share (including both mobile and desktop users) of 12.23%, up from 10.51% in January 2024 (see Figure 1).^[18]

Potential new entrants are another important source of competition for Google. As AI becomes more accessible, smaller companies and startups could develop alternative search engines or interfaces that challenge Google’s leadership. These potential competitors could come from adjacent “big tech” markets or from niche markets, such as specialized or “vertical” search engines.

Meta, for example, is reportedly developing an AI-based search engine to reduce its reliance on Google and Bing, which it currently uses to provide answers to its users.^[19] Likewise, Apple has the potential to create its own search engine and integrate it into its devices, potentially depriving Google of a crucial channel to reach billions of users.^[20] While Apple has denied any intention to construct a search engine, citing the substantial resources required and the associated competitive risks,^[21] the point is not that Apple is currently a competitor, but that it could become one. If Google were to attempt to leverage its market power against Apple, the dynamics surrounding the potential benefits and risks of developing a proprietary search engine would change significantly. Both Apple and Meta undoubtedly possess the resources and capabilities necessary to pursue such an initiative seriously.

Finally, there’s also the possibility that AI-powered “vertical” or specialized search engines could disrupt Google’s position. Disruption against leading companies often comes not from established actors, but from entrants in niche markets, which take advantage of the possibility to innovate more, remain more agile and focused, and cater to specific customer needs to gain traction and market share before entering into direct competition with more powerful incumbents.[22]

Traditional search engines like Google are designed to provide a broad range of results that can cover a wide array of topics. While Google uses sophisticated algorithms (such as RankBrain)^[23] to improve its understanding of context, it is still a generalized system. This means that—for highly specific or technical queries like medicine, law, or engineering—Google might struggle to provide the kind of nuanced, domain-specific knowledge users need.

Specialized AI search engines can address this gap by using models that have been specifically trained on vast datasets within a particular domain. Legal AI models, for instance, could be trained on case law, statutes, and legal documents, providing direct answers to complex legal queries. This could be invaluable for lawyers, law firms, or individuals seeking legal advice. Specialized engines would perform far better than Google at providing precise, contextually relevant answers to specific queries in their respective fields. In contrast, general search engines, even with the aid of AI, might struggle to maintain this level of specificity across diverse and complex topics.

II. Reassessing Traditional Barriers to Competition

The ITC also identifies several historical barriers to competition in search markets, including economies of scale, data advantages, and default positions. But even before the recent rise of AI-driven search alternatives, data accumulation alone did not create insurmountable barriers to entry, as competition in online markets was driven by such factors as engineering capabilities, business-model differentiation, and multi-homing by users. While these factors remain relevant, technological and market developments warrant a fresh assessment of their significance in 2025.

The CMA’s market study identified economics of scale, network effects, and data network effects (among other factors), as preventing rival search engines from “competing effectively with Google”.^[24] Indeed, it cites a number of market features that, it believes, prevent rival search engines from competing effectively with Google, including:

(a) economies of scale and scope – the infrastructure to search the web (a web index and crawlers) represents a major cost and is subject to significant economies of scale (paragraphs 3.53-3.63 & 3.87-3.91);

(b) network effects – users of search engines benefit from increased quality as the search engine acquires a greater number of users. This effect is driven by the importance of data (paragraphs 3.59 & 3.64);

(c) importance of data – the data on what queries users make and subsequently click on (click-and-query data) allow search engines to improve the quality and relevance of search results. The greater scale of queries Google sees compared to its rivals means it is able to deliver more relevant search results, in particular in relation to uncommon and new queries (paragraphs 3.64-3.89)…[25]

This echoes a widespread assumption—shared many other regulators—that firms operating digital platforms present some combination of increasing returns to scale, network effects, and data-related incumbency advantages. As we explain below, however, the evidence to support such claims is shaky, at best, as critics have failed to provide anything more than anecdotal evidence.^[26]

The CMA emphasizes Google’s historical data advantages, particularly in click-and-query data.[27] The assumption that data accumulation creates insurmountable competitive advantages, however, overlooks the fact that so-called “data network effects” are often just supply-side economies of scale or learning-by-doing effects, which diminish over time.[28]

An initial question concerns increasing returns to scale in the digital economy, in particular those related to the use of data.^[29] Indeed, critics often argue that digital platforms benefit from increasing returns to scale on all ranges of output. In other words, because of their scale, they require fewer economic inputs per-unit of output, which allegedly gives them an unassailable advantage over rivals. A report by the Stigler Center, for instance, concludes that:

Typically, information goods involve increasing returns to scale because their production requires a fixed cost and no or little variable cost …

The increasing returns to scale create barriers to entry: New firms cannot offer the quality of the incumbent without the same large-scale operation to pay for the fixed costs. But the firm can only achieve a large scale if quality is high. Thus, a potential entrant, foreseeing that it will not be profitable at the smaller scale, will not enter the market to challenge the incumbent.^[30]

In practice, however, the evidence for these increasing returns is particularly thin. In a paper published in George Mason Law Review, Geoffrey Manne and Dirk Auer show that financial data from tech companies does not support the notion that increasing returns to scale shields them from competition:

A look at the annual reports of many big tech firms is revealing in this regard. …

Google’s most recent 10-K, for example, shows that many of the company’s costs are unlikely to become smaller (in relative terms) as its output increases. This cuts against the existence of extreme returns to scale… In short, for most of Google’s expenditures, there is no obvious reason to believe that greater economic output would necessarily require a less than proportionate increase in economic inputs. …

The story is much the same for other big tech firms, such as Amazon and Facebook. Roughly three-quarters of Amazon’s expenditures involve “cost of sales” and “fulfillment” costs. These primarily include costs associated with the purchase and shipping of goods. There is little to suggest that either of these is a source of “extreme” returns to scale. They mostly involve the same capacity utilization challenges that brick-and-mortar retailers must contend with.^[31]

So, if they exist, from whence might big tech’s increasing returns to scale originate? One common suggestion is that they stem from the use of data. Critics sometimes point to the existence of so-called “data network effects” (although economies of scale and scope might be a more appropriate terminology^[32]).^[33] The argument goes that superior access to data allows firms to improve their products and gain more users. This then leads to even more data, thereby creating a self-reinforcing circle that eventually causes one firm to dominate the market. Take, for example, Google, which has become the poster child for unsophisticated “data network effects” arguments. In the words of Nathan Newman:

While there are a number of network effects that come into play with Google, [“its intimate knowledge of its users contained in its vast databases of user personal data”] is likely the most important one in terms of entrenching the company’s monopoly in search advertising. …

Google’s overwhelming control of user data . . . . might make its dominance nearly unchallengeable.^[34]

While the intuition is appealing, it has—to the best of our knowledge—neither been translated into a rigorous economic model nor established empirically. In fact, the anecdotal evidence that has often been used to support this naïve assertion merely shows that learning by doing plays an important role in the tech industry, just as it does in the rest of the economy.

To start, the existence of data-driven increasing returns to scale (or other data-related incumbency advantages) is not borne out by the empirical literature on the topic. Summarizing these empirical findings, Catherine Tucker concludes that “empirically there is little evidence of economies of scale and scope in digital data in the instances where one would expect to find them”.^[35]

There are numerous pieces of evidence to support this claim.^[36] For instance, economist Patrick Bajari and his co-authors use data from Amazon to show that (1) data on a wider range of products does not improve demand forecasting; and (2) increasing the timescale of data improves forecasting, but with diminishing returns.^[37] Likewise, in a paper co-authored with economist Lesley Chiou, Tucker finds that storing search-engine results for shorter periods does not affect the accuracy of subsequent search results.^[38] Again, this cuts against the existence of increasing returns to scale. In another paper, Tucker and her co-authors cast doubts on the overall accuracy of digital profiling, and thus the competitive edge that firms might obtain by acquiring larger amounts of data.^[39] Finally, a recent study argues that additional data improves algorithmic prediction with decreasing returns to scale.^[40] Using data from a large German news outlet, the authors show that the number of times a user visits a website improves the site’s prediction algorithm with decreasing returns (the algorithm optimizes the news articles that are presented to each individual user).^[41]

Likewise, a survey of the empirical literature by Geoffrey Manne and Dirk Auer concludes that “additional pieces of data are usually beneficial, but these benefits systematically entail diminishing marginal returns”.^[42] In short, available evidence suggests that claims of “extreme” returns to scale in the tech sector are greatly overblown. Not only are digital platforms’ largest expenditures unlikely to become proportionately less important as output increases, but empirical research strongly suggests that data itself does not give rise to increasing returns to scale, despite routinely being cited as the source of this effect.

III. Foreign Interventions: The EU Experience and Beyond

Building on experiences from other jurisdictions, the CMA must carefully distinguish between conduct that is genuinely anticompetitive and harmful to consumers, and conduct that—while not anticompetitive or harmful—conflicts with the CMA’s vision of how the online search market should be structured.

In defining the boundaries between “fair” and “unfair” or “informed” and “uninformed” choices, and “competitive” and “anticompetitive” conduct, the CMA confronts a pivotal question: does it seek to enable market-driven outcomes, or does it intend to impose its own? While it’s premature to assess the effectiveness of interventions in other jurisdictions, there are already some indications of where the line between regulators’ ambitions and consumer interests should be drawn.

A.  Self-Preferencing

The DMA’s self-preferencing ban has made it increasingly difficult for platforms to offer certain functionalities in Europe. For example, Google has been forced to remove features like maps, hotels bookings, and reviews from its search results. Until it can accommodate competitors who offer similar services (if this is even possible), these specialized search results will remain buried several clicks away from users’ general searches. Not only is this inconvenient for consumers, but it has important ramifications for business users.

Take hotel bookings, for example. Early estimates suggest that clicks from Google ads to hotel websites decreased by 17.6% as a result of the DMA. DMA implementation also sank 30% of clicks and bookings on Google Hotel Ads.^[43] As a result, the volume of direct bookings dropped as much as 36%, “increasing hotel dependence on intermediaries, which seriously damages their profitability”. In other words: while intermediaries benefit from the prohibition of self-preferencing, hotels lose out. In turn, it is not at all clear that intervention benefits consumers in any cognizable way, shape, or form.

By prohibiting Google from placing its own vertical services (Google Maps, Google Flights, and Google Hotel Ads) first, “the presentation of hotel offers to users based in DMA markets is less organised, clear and intuitive”.^[44] Previously, Google Search provided a direct display of hotels, featuring relevant details like prices, distance from the user, and images. Now, the top search results point to intermediaries like Booking.com and eDreams (see image below).

FIGURE 2: Post-DMA Google Search for Madrid Hotels

This sort of regulatory intervention does not make the market more “fair or contestable”. It merely robs Peter to pay Paul, while also robbing the consumer. As the study finds:

Prior to DMA, Google’s taxonomy of results was the result of decades of effort by the company to refine its results in order to provide an optimized search experience that would connect supply and demand in a way that was ideal for both.

This pre-DMA search experience offered hotels participating directly in the Google Hotel Ads product, the option to present their inventory (availability and room rates) in a way that was both efficient from the standpoint of distribution cost, and enriched for the user, as it integrated the experience of other services, e.g. Google Maps. This way of presenting information was clear, relevant and intuitive, and maximized purchasing decisions such as hotel bookings for those users who were so inclined.^[45]

Users have lost other functionalities following the implementation of Art. 6(5) DMA. They now face a less intuitive booking experience, with limited access to aggregated hotel offers, simplified calendar pricing, and streamlined tools like Google Travel. Frustrations include being redirected to search-engine results instead of the Travel section, and additional clicks required to complete actions.  Hotels have experienced a significant decline in visibility, with a 30% drop in website clicks and a 36% reduction in direct bookings in DMA-affected markets. The likely explanation is that lost clicks are being redirected to intermediaries like online travel agencies (e.g., Booking.com and Expedia) that have gained prominence in the new results format.

The irony, of course, is that Booking.com is also a designed a “gatekeeper” under the DMA. So, who has Art.6(5) really benefitted? Clearly not hotels: they have been subjected “to the toll of intermediation, strangling direct sales and holding users and hotels captive to less profitable, less independent business models”.^[46] It is also hard to see how users have benefitted from Google’s less integrated, less intuitive experience. Instead, the 30% of clicks lost are going to intermediaries like Booking.Com, Expedia, or eDreams.

Google has also removed other functionalities to comply with Art. 6(5). In March 2024, Google published an announcement titled “Complying with the Digital Markets Act” in which the company said it had “removed some features from the search results page which help consumers find businesses, such as the Google Flights unit”.^[47] Google noted that the DMA had produced unintended consequences, including a suboptimal user experience and impact to businesses.

We’ve always been focused on improving Google Search to help people quickly and easily find what they’re looking for. .. Rules that roll back some of these advances represent a fundamental shift in competition policy. We encourage other countries contemplating such rules to consider the potential adverse consequences — including those for the small businesses that don’t have a voice in the regulatory process.^[48]

These negative outcomes are not entirely unexpected. The notion that the ability to give preferential treatment to one’s products is inherently anticompetitive contradicts “over a century of antitrust jurisprudence, economic study, and enforcement agency practice” that have firmly established that “the competitive effects of a vertically integrated firm’s ‘discrimination’ in favor of its own products or services… generally produce significant benefits for consumers”.^[49]

It is also flatly contrary to a number of empirical studies showing that even the welfare of competitors (to say nothing of consumers) may often be improved by such self-preferencing.^[50] While enforcement of such provisions may benefit certain competitors in the short run (e.g., Booking.com or Expedia), they create perverse incentives over the long run for rivals, who may underinvest in ensuring their own viability due to such regulations inefficiently insuring them against their own business misjudgments.^[51] In addition, making Google Search less relevant and thus less attractive to users is likely to negatively affect small businesses who use Google to reach a wide, previously unimaginable audience — even if it does momentarily spur the sales of large intermediaries.

Is this a tradeoff that the CMA wishes to make? Further: is it a tradeoff that it is empowered to make under the Competition Act and the DMCC?

B. Choice Screens

In the EU, Android devices were required to offer choice screens even before the DMA took effect.[52] Yet despite these regulatory mandates, consumers have overwhelmingly stuck with Google—indicating that, even when presented with alternatives, they prefer Google Search. In fact, Google’s share of the general search market has remained largely unchanged since the introduction of choice screens, and still hovers around 91%.

The same cannot be said for other default services. In the United States, for example, Microsoft preloads its Edge browser and Bing search engine on Windows computers, yet Bing continues to trail far behind Google in U.S. desktop search. This suggests that consumers, including those who aren’t particularly tech-savvy, are willing and able to switch search engines if they choose. Simply put, Google’s dominance stems from consumer preference, not just default settings.

It does appear that choice screens have led to a small increase in the shares of rival search engines like DuckDuckGo and Bing. Since the Google Android decision, Google has lost about 5.6% market share in the EU, with significant variations across member states.^[53] But some of the search engines that have benefitted still license their results from Google and Bing. So, what have the choice screens really achieved? Perhaps funneling some of Google’s advertising revenue to “small” competitors like Bing, or smaller ones still like Ecosia and DuckDuckGo. But is this marginal redistribution of wealth worth the inconvenience inflicted on consumers? In addition, the change—however small—could affect businesses that have invested in optimizing their websites for Google’s search engine, not DuckDuckGo’s or Bing’s.

As for browsers, the DMA has given rivals significant installation jumps, including as much as a 164% increase for Opera^[54] and a 250% increase for Aloha.^[55] But was this due to the choice screens or the fanfare surrounding the DMA? If the latter, the surge in market share is likely to be fleeting. There are some indications that this is likely to be the case: the Google Android case, which mandated choice screens since 2018, had almost no impact on Google’s market share.^[56] (In fact, one of the few firms that benefitted from the choice screen was Russia’s Yandex).^[57]

Even with the examples of Opera and Aloha’s growing market share, does this confirm the European Commission’s supposition that users demand to be freed from the shackles of default settings, or merely that “mandated nagware” will inevitably drive some to try alternatives? The important question is, will users continue to stick to these services, or eventually return to the familiar market leaders? And what conclusions will regulators draw from these changes?

The CMA must thus ask itself the following question: if UK devices come with choice screens, would that make a difference? What sort of difference? If a very small percentage of users at the margin were to switch (say, to the highest option on the choice screen), would that make a significant competitive difference? Would it help the average consumer? Many consumers prefer a default—a highly functioning and simple package out-of-the-box.

In accordance with the DMCC, the CMA is obligated to impose conduct requirements that are proportionate^[58] and beneficial to consumers.^[59] There is, however, scant evidence to suggest that consumers actively seek choice screens; rather, it appears to be a preferred option of regulators rather than a demand from consumers themselves. Ultimately, regulators are mandated to safeguard consumer interests, not their own.^[60]

C. The US District Court’s Decision in Google Search

The CMA cites the recent Google Search decision by the U.S. District Court for the District of Columbia. It highlights that the court “found that Google had acted illegally to maintain its monopoly position in the markets for ‘general search services’ and ‘general search text advertising’ in the US”. The significance of these findings should, however, be put into the appropriate context.

First, the court’s monopoly finding isn’t surprising, or even particularly telling. Google has long held an extremely large share of general search in the United States. But as the U.S. Court of Appeals for the D.C. Circuit held in the watershed 2001 Microsoft decision, “merely possessing monopoly power is not itself an antitrust violation”. Rather, a plaintiff must also prove exclusionary conduct—a much taller order.^[61]

Second, just because a court determined that Google has monopoly power in search does not mean another court will agree it has monopoly power in online advertising or other markets.

Third, the decision is fundamentally flawed. The U.S. case concerns Google’s agreements to pre-load Google as the default search engine in internet browsers and on mobile devices. According to the U.S. Justice Department (“DOJ”), these agreements excluded Microsoft Bing (and other general search engines) from effectively competing because users rarely switch from the default.

There is, however, another explanation for Bing’s lack of success: maybe it just wasn’t as good. The court does not rule this out or show that Bing would have succeeded without Google’s deals. The court’s failure to thoroughly consider consumer preference as a factor in Bing’s lack of success raises concerns that the CMA should seek not to replicate.

As we have discussed extensively elsewhere,^[62] the court’s legal conclusion that the DOJ didn’t need to prove that the agreements were the cause of Google success rests on a misinterpretation of the Microsoft decision. To prevail, the government should have been required to prove that Google’s default distribution deals genuinely excluded Bing from the market to the detriment of consumers. Moreover, these default agreements are not exclusive, and users can and do switch to other search engines with relative ease. The real question is why users often choose not to switch. Is it due to the so-called “power of defaults”, or is it simply because they prefer Google’s service? Evidence points toward the latter.

As we pointed out earlier, in the EU, where Google has been prohibited from being offered as the default on Android devices since 2020, its market share has remained largely unchanged. On Windows devices, where Bing is the default, the majority of users still opt for Google. These examples indicate that, when alternatives are preferable, users are willing and able to change their default settings.

If consumer preference is, indeed, the primary reason for Google’s dominance, then the court’s decision may inadvertently hinder users from accessing their preferred services—a counterproductive outcome. This scenario underscores the importance of thoroughly understanding market dynamics before implementing regulatory interventions.

The U.S. court’s conclusion appears rooted in a presumption that large tech firms cannot become dominant simply by being better. Indeed, this same assumption permeates much of the world’s newly found antipathy toward big tech. Efforts to ban similar types of conduct (like Apple integrating iMessage with iPhones or Google “self-preferencing” its maps in its search results) without proof of harm to consumers risk stifling innovation and disregarding consumer preferences. The CMA should not tread this path without concrete evidence that consumers are being harmed by existing practices.

D. Requirements to Obtain User Consent for Data

Under the DMA, Google is required to obtain explicit user consent before combining data across its services; it also must share anonymized data with competitors. Evidence shows that these measures hamper the personalization and effectiveness of services such as search results, recommendations, and targeted advertising.

Take the EU’s General Data Protection Regulation (“GDPR”), for example. The impact of the GDPR became evident when platforms like Google and Facebook struggled to obtain consent for personalized ads, resulting in less data collection. This led to a decrease in ad revenue, as advertisers found less effective, non-targeted ads to be less profitable. A conduct requirement under the DMCC along these lines could have similar consequences, as forced segregation of data into “siloes” reduces platforms’ ability to offer seamless experiences and personalized services.

If Google were to show less personalized ads to UK consumers, it wouldn’t necessarily mean that it would show less ads. It would still show ads, just less relevant ones. This would be a mistake. Personalized ads drive sales.^[63] Small businesses, especially, rely on targeted advertising to reach consumers with more relevant information in the most cost-effective way.^[64] Targeted advertising also benefits consumers by specifically tailoring ads to match their interests.^[65] As one startup founder in the EU put it:

Targeted advertising is the main tool that allows a startup to validate its value proposition and get into a market at low cost, optimising marketing expenses that – at initial stages – must be super efficient. Any limitation would be a barrier to market entry for new players in the economy, which could pose a direct challenge to the entire startup ecosystem, and would result in favour of incumbents and established companies that are protected from challenges by the legislator.^[66]

E. Measures Relating to News-Publisher Remuneration

The CMA points out that “several jurisdictions have imposed measures with the aim of ensuring that news publishers receive fair payment terms for their content, including the News Media Bargaining Code in Australia and the Online News Act in Canada”.

In Canada, the Online News Act resulted in Meta banning all news from its platform.^[67] The news ban in Canada significantly disrupted the media landscape. Canadian news outlets experienced a sharp decline in online traffic, engagement, and revenue, with some losing 85% of engagement on Facebook and Instagram.^[68] The ban led to a 43% overall drop in engagement and a reduction of 11 million daily views. Smaller outlets, especially those relying heavily on social-media platforms, were hit hardest, with many seeing traffic plummeting by 50%.^[69]

The ban also affected non-news outlets, like university radio stations, which lost their social-media presence, hindering their ability to engage with audiences and secure funding. Rural and First Nations communities, in particular, faced challenges due to the loss of Facebook as a primary news source.^[70]

Meta also restricted sharing and viewing of news content on Facebook in Australia in response to the proposed News Media Bargaining Code. This led to a significant reduction in news consumption on the platform. One study found that, while some users sought alternative news sources, others experienced a decline in news consumption, potentially increasing exposure to misinformation.^[71] The Independent Media Alliance opined that the ban would be “terrible for not only the industry, but for Australian democracy”.^[72]

While Facebook eventually reversed the ban and reached a deal that allowed news sharing to resume, the situation had significant ramifications. Larger publishers negotiated deals for compensation from Facebook, but smaller news outlets faced sunk revenue losses. At the same time, Facebook has not reversed its ban in Canada. Thus, even in the Australian “best case” scenario, an agreement was reached only after painstaking negotiations and irreversible losses for smaller outlets.

As for Google, while it has shown more willingness to negotiate, this comes with a caveat. In Australia, Google agreed to pay news companies only after intense negotiations, similarly to Meta. Ultimately, Google secured terms more favorable to its business model. It negotiated payments on a case-by-case basis, rather than a fixed payment model with a common price across the board. While gigantic companies like Australia’s own News Corp can afford these transaction costs, smaller outlets may not. Google was also able to pick which content to display—and pay for—on its platform. Put simply, if you turn Google into a news buyer, it will shop around.

Ultimately, major media companies with significant bargaining power, like News Corp and Nine Entertainment, were the main beneficiaries of the agreements made under the News Media Bargaining Code. These large publishers offered more varied content that was valuable to Google because it attracted a larger audience and thus increased ad revenue. In addition, large publishers were able to command higher payments, making them more likely to receive favorable treatment in terms of visibility on Google’s platform. Conversely, smaller or independent news outlets that did not strike agreements with Google risked being excluded from Google’s news services or search results, or receiving much less exposure than they would have in a but-for world.^[73]

It is unclear how the over-representation of news from large corporations like News Corp and the under-representation of small, independent outlets benefits consumers, the market, or democracy.

IV. Conclusion: Assessing the CMA’s Proposed Interventions

The ITC outlines several potential interventions, ranging from mandated data sharing to restrictions on default agreements. While we appreciate the CMA’s thorough consideration of possible remedies, we respectfully submit that any interventions must be carefully evaluated against current market realities and the risk of unintended consequences.

A. Data-Sharing Requirements

The CMA suggests requiring Google to share its web index and click-and-query data with competitors.[74] While superficially attractive, mandatory data sharing raises complex practical and competitive concerns. Mandatory data-sharing requirements can create significant implementation challenges and may have ambiguous effects on competition.[75]

The value of historical search data diminishes rapidly in many categories, particularly for queries related to current events, prices, or trending topics. Recent empirical research by Chiou and Tucker demonstrated that the competitive advantage from historical search data may be significantly shorter-lived than previously assumed.[76] Moreover, sharing click-and-query data raises significant privacy considerations that could conflict with users’ expectations and data-protection requirements.[77]

More fundamentally, mandating data sharing could paradoxically reduce incentives for competition and innovation. Forced data sharing might encourage new entrants to focus on replicating incumbent approaches rather than developing novel solutions.[78] In a market where AI and specialized search services are already developing alternative approaches to information discovery, mandated data sharing might actually slow innovation by entrenching traditional search paradigms.

B. Default and Preinstallation Interventions

The CMA contemplates additional restrictions on default search positions and pre-installation agreements.[79] Such interventions, however, should be evaluated against existing measures and changing user behavior. The UK already has a functioning choice-screen mechanism for Android devices. Recent empirical work suggests that the effectiveness of choice screens depends heavily on their design and implementation, rather than merely their existence.[80]

Furthermore, default restrictions could have unintended consequences for competition. Many smaller search engines currently compete for default positions through revenue-sharing agreements with device manufacturers and browsers. With two-sided markets, however, restricting these agreements could paradoxically harm competition by removing a key mechanism through which alternative search engines currently reach users.[81]

C. Restrictions on Data Usage Across Services

The CMA proposes restricting Google’s ability to use data across its services.[82] Such restrictions, however, could harm user experience without necessarily promoting competition. Users often expect and benefit from integration across services. Cross-service data integration can generate significant consumer benefits through improved service quality and reduced search costs.

The challenge of balancing data protection with competition concerns has been extensively analyzed in recent literature. As Hal Varian notes, data should be understood in terms of access, rather than rigid ownership rules, and restrictions on data usage should account for the fact that advantages from data accumulation diminish over time.[83] Interventions in digital markets should carefully weigh immediate competitive benefits against the long-term incentives for innovation.

D. Conclusions and Recommendations

Drawing on both the empirical evidence and theoretical frameworks discussed above, the CMA should carefully reconsider whether an SMS designation is warranted at this time. The rapidly evolving nature of digital search markets suggests a more nuanced approach may be appropriate.

First, technological transitions—particularly those driven by AI—can rapidly reshape competitive dynamics in ways that traditional market-power analysis might miss.[84] The emergence of new search paradigms and business models suggests we are already in the midst of such a transition.

Second, traditional metrics of market power may not capture the current competitive reality. Market-share analysis can be particularly misleading during periods of technological change.[85] The CMA should consider developing new analytical frameworks that better capture the dynamic nature of competition in modern search markets.

If the CMA nonetheless proceeds with an SMS designation, we recommend the following principles for any subsequent interventions:

• Adopt an “innovation first” approach to remedies that preserves incentives for both incumbents and new entrants to develop novel search technologies.
• Focus on removing barriers to competition, rather than imposing detailed conduct requirements. Light-touch interventions often prove more effective than prescriptive regulation in fast-moving technology markets.
• Establish regular review periods to assess the continued appropriateness of any interventions.

The CMA has an opportunity to set a global standard for thoughtful regulation of digital search markets. By carefully considering the dynamic nature of competition and focusing on forward-looking analysis, the CMA can help ensure that UK consumers and businesses benefit from continued innovation in search technology.

[1] CMA Invitation to Comment on Strategic Market Status Investigation into Google’s General Search and Search Advertising Services, Compet. Mark. Auth. (14 January 2025), [hereinafter: “CMA Investigation”].

[2] Online Platforms and Digital Advertising Market Study Final Report, Compet. Mark. Auth. (1 July 2020), available at https://assets.publishing.service.gov.uk/media/5fa557668fa8f5788db46efc/Final_report_Digital_ALT_ TEXT.pdf.

[3] See Miles Kruppa, Jeff Bezos Bets on a Google Challenger Using AI to Try to Upend Internet Search, Wall Str. J. (4 January 2024), https://www.wsj.com/tech/ai/jeff-bezos-bets-on-a-google-challenger-using-ai-to-try-to-upend-internet-search-0859bda6.

[4] CMA Investigation, supra note 1, at 6.

[5] See Suzanne Vranica & Miles Kruppa, Google’s Grip on Search Slips as TikTok and AI Startup Mount Challenge, Wall Str. J. (6 Oct. 2024), https://www.wsj.com/tech/online-ad-market-google-tiktok-9599d7e8.

[6] Digital Markets, Competition and Consumers Act, c. 13 (U.K.).

[7] CMA Investigation, supra note 1, at 3.

[8] DMCC, supra note 6.

[9] CMA Investigation, supra note 1, ¶ 15.

[10] Id. at ¶ 39(a)

[11] Digital Markets Competition Regime: Guidance on the Operation of the New Regime, Competition and Markets Authority, Compet. Mark. Auth. (19 December 2024), § 2.10, available at https://assets.publishing.service.gov.uk/media/6762f4f6cdb5e64b69e307de/Digital_Markets_Competition_Regime_Guidance.pdf.

[12] Yutao Zhu et al., Large Language Models for Information Retrieval: A Survey, arXiv (4 September 2024), https://arxiv.org/html/2308.07107v4.

^[13] Martin Coulter & Greg Bensinger, Alphabet Shares Dive After Google AI Chatbot Bard Flubs Answer in Ad, Reuters (8 February 2023), https://www.reuters.com/technology/google-ai-chatbot-bard-offers-inaccurate-information-company-ad-2023-02-08.

^[14] Dismantling Google Is a Terrible Idea, The Economist (3 October 2024), https://www.economist.com/leaders/2024/10/03/dismantling-google-is-a-terrible-idea.

^[15] “That’s one clear reason a user might choose ChatGPT over Google Search: there’s no clutter of advertising or promoted queries pinned to the top. While Google makes significant revenues from search-result advertising, Fry said there are currently “no plans” for advertising in ChatGPT. Still, AI-powered search is more expensive to operate than traditional search, and it’s not yet clear how OpenAI will finance it for free users. Felix said that free users will have “some limits on how often they can use our latest search models”; see Kylie Robinson, OpenAI’s Search Engine is Now Live in ChatGPT, The Verge (31 October 2024), https://www.theverge.com/2024/10/31/24283906/openai-chatgpt-live-web-search-searchgpt.

^[16] While advertising is often described in the context of digital platforms as a cost to consumers, it also has benefits. Advertising reduces search costs for consumers and, therefore, reduces the total cost of purchasing goods and services.

^[17] Michael Kan, Microsoft’s Bing Tops 100 Million Users With ChatGPT Integration, PC Mag (9 March 2023), https://www.pcmag.com/news/microsofts-bing-tops-100-million-users-with-chatgpt-integration; Tom Warren, Microsoft Bing Hits 100 Million Active Users in Bid to Grab Share from Google, The Verge (9 March 2023), https://www.theverge.com/2023/3/9/23631912/microsoft-bing-100-million-daily-active-users-milestone.

^[18] Tiago Bianchi, Market Share of Leading Desktop Search Engines Worldwide from January 2015 to January 2025, Statista (23 January 2025), https://www.statista.com/statistics/216573/worldwide-market-share-of-search-engines.

^[19] Kalley Huang, Meta Develops AI Search Engine to Lessen Reliance on Google, Microsoft, The Information, (28 October 2024), https://www.theinformation.com/articles/meta-develops-ai-search-engine-to-lessen-reliance-on-google-microsoft.

^[20] Apple currently builds its own search engines for services like the App Store, Maps, Apple TV, and Apple News. Apple devices also have Spotlight, which allow users to find apps and documents across devices. See Mark Gurman, Apple Has What It Needs to Launch Its Own Google Replacement, Bloomberg (1 October 2023),  https://www.bloomberg.com/news/newsletters/2023-10-01/could-apple-replace-google-with-own-search-engine-it-s-possible-but-unlikely-ln7gywed.

^[21] Umar Shakir, Eddy Cue Explains Why Apple Won’t Make a Search Engine, The Verge (26 December 2024),  https://www.theverge.com/2024/12/26/24329148/apple-eddy-cue-search-engine-ai-google-doj-antitrust-intervene.

[22] See, e.g., Clayton M. Christensen, Michael E. Raynor, & Rory McDonald, What Is Disruptive Innovation? Harv. Bus. Rev. (December 2015), https://hbr.org/2015/12/what-is-disruptive-innovation.

^[23] Dave Davies, A Complete Guide to the Google RankBrain Algorithm, Search Engine J. (2 September 2020),  https://www.searchenginejournal.com/google-algorithm-history/rankbrain.

^[24] CMA Investigation, supra note 1, at 5.

[25] Id.

^[26] Jacques Crémer, Yves-Alexandre Montjoie, & Heike Schweitzer, Competition Policy for the Digital Era, Eur. Comm. (2019), 12, https://op.europa.eu/en/publication-detail/-/publication/21dc175c-7b76-11e9-9f05-01aa75ed71a1/language-en.

[27] CMA Investigation, supra note 1, at 5.

[28] See Hal Varian, Artificial Intelligence, Economics and Industrial Organization (Nat’l Bureau of Econ. Rsch., Working Paper 24839, July 2018), 17, available at https://www.nber.org/system/files/working_papers/w24839/w24839.pdf (“There is a concept that is circulating among lawyers and regulators called ‘data network effects.’ The model is that a firm with more customers can collect more data and use this data to improve its product. This is often true—but it is hardly novel. And it is certainly not a network effect! This is essentially a supply-side effect known as ‘learning by doing’ (also known as the ‘experience curve’ or ‘learning curve’.”).

^[29] For an introduction to the concept of increasing returns to scale, see, generally, Hal R. Varian, Microeconomic Analysis 16 (3d ed. 1992) (“[W]hen output increases by more than the scale of the inputs, we say the technology exhibits increasing returns to scale.”).

^[30] Fiona Scott Morton et al., Committee for the Study of Digital Platforms Market Structure and Antitrust Subcommittee Report, George J. Stig. Ctr. For The Study Of The Econ. And The State (1 July 2019), 36-37, available at https://research.chicagobooth.edu/-/media/research/stigler/pdfs/market-structure-report.pdf?la=en&hash=E08C7C9AA7367F2D612DE24F814074BA43CAED8C.

^[31] Geoffrey Manne & Dirk Auer, Antitrust Dystopia and Antitrust Nostalgia: Alarmist Theories of Harm in Digital Markets and Their Origins, 28 Geo. Mason L. Rev. 1279, 1339-1341 (2021).

^[32] See Catherine Tucker, Digital Data, Platforms and the Usual [Antitrust] Suspects: Network Effects, Switching Costs, Essential Facility, 54 Rev. Indus. Org. 683, 685 (2019). (“The middle two categories of network effects described by Grunes and Stucke (2016), are simply known as economies of scale and scope in economics terminology and are considered to be distinct from network effects.”).

^[33] See, e.g., Maurice E. Stucke & Allen P. Grunes, Big Data and Competition Policy 6–7 (2016) (arguing that data-driven industries are “subject to several network effects”, including: “traditional network effects, including social networks such as Facebook; network effects involving the scale of data; network effects involving the scope of data; and network effects where the scale and scope of data on one side of the market affect the other side of the market (such as advertising)”.); see also Jason Furman, Address at the FTC Hearings on Competition and Consumer Protection in the 21st Century, Fed. Trade Comm. (13 September 2018), (“I think the big empirical question that I do not know the answer to . . . is if you think there is diminishing returns to data then you are a lot less worried about it then [sic] if you think there is some region of increasing returns. There is [sic] some people that deal with computer science that say, with machine learning, when you get past a certain point you get to this place where you can, you know, do the AI in a certain way that you could not do before you get to that scale”.)

^[34] Nathan Newman, Search, Antitrust, and the Economics of the Control of User Data, 31 Yale. J. on Reg. 401, 420, 423 (2014).

^[35] See Tucker, supra note 32, at 686.

^[36] But see Maximilian Schaefer, Geza Sapi, & Szabolcs Lorincz, The Effect of Big Data on Recommendation Quality. The Example of Internet Search, (Düsseldorf Inst. for Competition Econ. Discussion Paper No. 284, 2018), 5, (showing that cookies that track user activity for longer periods of time improved the accuracy results on the Yahoo search engine). One potential objection to this study is that, although it broadly argues that obtaining more data about each user improves results, it says very little about the cumulative effect of obtaining data about multiple users. Ultimately, it is this second potential effect that is central to critics’ claims. See, e.g., Newman, supra note 34, at 421.

^[37] See Patrick Bajari, Victor Chernozhukov, Ali Hortaçsu, & Junichi Suzuki, The Impact of Big Data on Firm Performance: An Empirical Investigation (Nat’l Bureau of Econ. Rsch., Working Paper No. 24334, 2019), 5–6.

^[38] See Lesley Chiou & Catherine Tucker, Search Engines and Data Retention: Implications for Privacy and Antitrust (Nat’l Bureau of Econ. Rsch., Working Paper No. 23815, 2017), 3.

^[39] See Nico Neumann, Catherine E. Tucker, & Timothy Whitfield, How Effective is Third-Party Consumer Profiling and Audience Delivery?: Evidence from Field Studies, 38 Mktg. Sci. 918 (2019).

^[40] See Jörg Claussen, Christian Peukert, & Ananya Sen, The Editor vs. the Algorithm: Targeting, Data and Externalities in Online News (CESifo Working Paper No. 8012, 2019).

^[41] Id. at 11.

^[42] Manne & Auer, supra note 31, at 1343-44.

^[43]Javier Delgado, DMA Implementation Sinks 30% of Clicks and Bookings on Google Hotels Ads, Mirai (7 May 2024) https://www.mirai.com/blog/dma-implementation-sinks-30-of-clicks-and-bookings-on-google-hotel-ads.

^[44] Id.

^[45] Id.

^[46] Id.

^[47] Oliver Bethell, Complying with the Digital Markets Act, Google Blog (5 March 2024), https://blog.google/around-the-globe/google-europe/complying-with-the-digital-markets-act.

^[48] Adam Cohen, New Competition Rules Come with Trade-Offs, Google Blog (5 April 2024), https://blog.google/around-the-globe/google-europe/new-competition-rules-come-with-trade-offs.

^[49] See Geoffrey A. Manne, Against the Vertical Discrimination Presumption, Concurrences No. 2-2020 (2020), at 1. See also Jonathan M. Barnett, The Host’s Dilemma: Strategic Forfeiture in Platform Markets for Informational Goods, 124 Harv. L. Rev. 1861 (2011); Andrei Hagiu & Kevin Boudreau, Platform Rules: Multi-Sided Platforms as Regulators, in Platforms, Markets and Innovation (Annabelle Gawer, ed. 2009).

^[50] Manne, id., at 1-2 (citing examples from the literature showing that complementors and consumers alike often benefit from platform self-preferencing). See also Sam Bowman & Geoffrey A. Manne, Platform Self Preferencing Can be Good for Consumers and Even Competitors, Truth Mark. (4 March 2021), https://laweconcenter.wpengine.com/2021/03/04/platform-self-preferencing-canbe-good-for-consumers-and-even-competitors.

^[51] On self-inflicted dependence, see Geoffrey A. Manne, The Real Reason Foundem Foundered, Int’l Ctr. L. Econ. (2018), at 6, available at https://laweconcenter.org/wp-content/uploads/2018/05/mannethe_real_reaon_foundem_foundered_2018-05-02-1.pdf (“A content provider that makes itself dependent upon another company for distribution (or vice versa, of course) takes a significant risk. Although it may benefit from greater access to users, it places itself at the mercy of the other—or at least faces great difficulty (and great cost) adapting to unanticipated, crucial changes in distribution over which it has no control. This is a species of what economists call the ‘asset specificity’ problem.”).

[52] Commission Decision of 18 July 2018 in CASE AT.40099 Google Android.

^[53] Kevin Indig, Why Is Google Losing Market Share in the EU?, Search Engine J. (3 December 2024), https://www.searchenginejournal.com/why-is-google-losing-market-share-in-the-eu/534254.

^[54] Press Release, Opera Saw 164% Growth in the Inflow of New EU Users on iOs After the DMA-Enforced Ballot Screen, Opera Press (18 March 2024), https://press.opera.com/2024/03/18/opera-jump-in-new-eu-users-after-ballot-screen.

^[55] Supantha Mukherjee & Foo Yun Chee, Exclusive: EU’s New Tech Laws Are Working; Small Browsers Gain Market Share, Reuters (10 April 2024), https://www.reuters.com/technology/eus-new-tech-laws-are-working-small-browsers-gain-market-share-2024-04-10.

^[56] George Nguyen, Google’s Search Choice Screen Had Virtually No Effect on Search Market Share, Perhaps by Design, Search Engine Land (17 February 2021), https://searchengineland.com/googles-search-choice-screen-had-virtually-no-effect-on-search-market-share-perhaps-by-design-346167.

^[57] Id.

^[58] DMCC, S.19(5); S.21(b)(i).

^[59] DMCC, S.19(10) states: “Before imposing a conduct requirement or a combination of conduct requirements on a designated undertaking, the CMA must have regard in particular to the benefits for consumers that the CMA considers would likely result (directly or indirectly) from the conduct requirement or combination of conduct requirements.”; See also S.21(b)(ii).

^[60] Lazar Radic, Italy’s Google and Apple Decisions: Regulatory Paternalism and Overenforcement, Truth Mark. (10 December 2021), https://truthonthemarket.com/2021/12/10/italys-google-and-apple-decisions-regulatory-paternalism-and-overenforcement. (“Much like in the United States, where the Federal Trade Commission (FTC) has occasionally engaged in product-design decisions that substitute the commission’s own preferences for those of consumers, regulators around the world continue to think they know better than consumers about what’s in their best interests”).

^[61] U.S. v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001).

^[62] Geoffrey A. Manne, A Critical Analysis of the Google Search Antitrust Decision, Int’l Ctr. L. Econ. (14 August 2014), available at https://laweconcenter.org/wp-content/uploads/2024/08/Manne-Google-Search-Decision-Analysis-2024-08-14.pdf.

^[63] See, e.g., Megan Graham, Taco Bell and KFC’s Owner Says AI-Driven Marketing Is Boosting Purchases, Wall Str. J. (15 November 2024), https://www.wsj.com/articles/taco-bell-and-kfcs-owner-says-ai-driven-marketing-is-boosting-purchases-ab3a5f36.

^[64] Niklas Fourberg et al., Artificial Intelligence in the Digital Single Market: Opportunities, Challenges, and Risks for Small and Medium-Sized Enterprises, Eur. Parl. (June 2021), available at https://www.europarl.europa.eu/RegData/etudes/STUD/2021/662913/IPOL_STU(2021)662913_EN.pdf.

^[65] Id.

^[66] The Importance of Targeted Advertising for Startup Ecosystems in Europe, Allied for Startups (18 November 2021), https://alliedforstartups.org/2021/11/18/the-importance-of-targeted-advertising-for-startup-ecosystems-in-europe.

^[67] Indeed, while fair-remuneration laws can force a platform to pay a news publisher, it cannot force the platform to host the news publisher if it does not want to.

^[68] Jessica Paterson, How Meta’s News Ban Reshaped Canadian Media, Digital Content Next (12 September 2024), https://digitalcontentnext.org/blog/2024/09/12/how-metas-news-ban-reshaped-canadian-media.

^[69] Id.

^[70] Id.

^[71] Ying Gu, Stephanie Lee, & Yong Tan, News in the Dark: Effects of Facebook’s Australian News Ban on News Consumption, SSRN (5 April 2024), https://ssrn.com/abstract=4790864.

^[72] Josh Taylor, Facebook’s Potential News Ban Already Affecting Smaller Australian Media Outlets, Inquiry Told, The Guardian (21 June 2024), https://www.theguardian.com/media/article/2024/jun/21/facebooks-potential-news-ban-already-affecting-smaller-australian-media-outlets-inquiry-told.

^[73] Paul Karp, Amanda Meade, & Josh Butler, Meta, TikTok and Google Will Be Forced to Pay Australian News. What Does It Mean for You?, The Guardian (12 December 2024), https://www.theguardian.com/australia-news/2024/dec/12/meta-tiktok-and-google-to-be-forced-to-pay-for-australian-news.

[74] CMA Investigation, supra note 1, at ¶¶ 41-43.

[75] Report on the Application of the EU Rules on Public Procurement, Chapter 5: Data, Eur. Comm. (2019), https://op.europa.eu/en/publication-detail/-/publication/21dc175c-7b76-11e9-9f05-01aa75ed71a1/language-en.

[76] Lesley Chiou & Catherine Tucker, Search Engines and Data Retention: Implications for Privacy and Antitrust (NBER Working Paper 23815, 2017).

[77] See, e.g., Michael Gal & Daniel L. Rubinfeld, Data Standardization, 94 NYU L. Rev. 737 (2019).

[78] Jan Krämer, Shiva Shekhar, & Janina Hofmann, Regulating Digital Platform Ecosystems Through Data-Sharing and Data-Siloing: Consequences for Innovation and Welfare, EconStor (2022), available at https://www.econstor.eu/bitstream/10419/265645/1/Kraemer-et-al.pdf (“While mandated data sharing does increase the level of competition in the secondary market, it lowers the incumbent’s incentive to innovate in the primary market”).

[79] CMA Investigation, supra note 1, at ¶ 41(a).

[80] Omar Vasquez Duque, Active Choice vs. Inertia? An Exploratory Assessment of the European Microsoft Case’s Choice Screen, 19 J. Comp. L. & Econ 60. (2023).

[81] Erik Hovenkamp, The Competitive Effects of Search Engine Defaults, SSRN (14 November 2024), at 21, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4647211 (“If a potential entrant (if successful) can obtain a default, this increases its ex ante investment and raises the probability of entry. In this case, the default may raise dynamic consumer welfare”.)

[82] CMA Investigation, supra note 1 at ¶ 42.

[83] Varian, supra note 28 at 8.

[84] Crémer et al., supra note 26, at 39 (“Market boundaries might not be as clear as in the ‘old economy’. They may change very quickly. Furthermore, in the case of multi-sided platforms, the interdependence of the ‘sides’ becomes a crucial part of the analysis whereas the traditional role of market definition has been to isolate problems. Therefore, we argue that, in digital markets, we should put less emphasis on analysis of market definition, and more emphasis on theories of harm and identification of anti-competitive strategies”.)

[85] Id.