U.S. v Google Antitrust Trial: General Search Services In The United States Is A Relevant Market

A. General Search Services In The United States Is A Relevant Market

323. General search services are the services GSEs provide to consumers, allowing consumers to find information from across the internet. Tr. 4610:13–22 (Whinston (Pls. Expert)). The services provided by GSEs are significantly differentiated from other methods of searching for and accessing information, such as by using specialized search engines or social media platforms. Id. 4611:6–25. Accordingly, general search services is a relevant antitrust product market in the United States. Id. 4610:13–4612:13, 4616:5–25.

1. The Characteristics And Uses Of General Search Services Support Finding A General Search Services Market

324. The characteristics and consumer usage patterns of GSEs indicate that general search services constitute a relevant product market.

a) General Search Services Are Unique In The Breadth Of Queries They Serve

325. General search services are unique in the breadth of the queries they can serve. GSEs seek to return relevant results to any user query, no matter the topic. Tr. 8266:19–25 (Reid (Google)); Tr. 6511:5–23 (Hurst (Expedia Group)). Dr. Mark Israel, Google’s economic expert, conceded as undisputed that a GSE, “can handle virtually any type of query.” Tr. 8708:16–20 (Israel (Def. Expert)).

326. Specialized search engines cannot match the depth and breadth of results that are available on GSEs. Instead, specialized search engines focus on returning queries related to specific information categories, or “verticals.” Des. Tr. 181:5–184:16 (Moxley (Google) Dep.) (queries and results on a GSE differ from queries and results on verticals, such as Expedia, Yelp, Nike, and Amazon).

For example, a consumer can search “best place with a view in the Caribbean” and expect a result on Google, whereas a travel site like Expedia can only return results if the user provides specific information like “where you’re going, what dates you’re going, and how many people are going.” Tr. 6511:8–23 (Hurst (Expedia Group)); Tr. 4612:14– 4614:8, 4617:1–4618:3 (Whinston (Pls. Expert)) (comparing the SERPs of Google and Amazon, you get “a range of kinds of information [for Google] against just of a narrow, come buy something [for Amazon]”); id. 6149:11–6150:20 (Google SERP for the query “valentines day gifts” provides information not available on the Amazon SERP); Tr. 8677:25–8678:2 (Israel (Def. Expert)) (Google and Bing return results to a broader set of queries than Amazon).

b) Consumers Use General Search Engines As “One-Stop Shops”

327. Consumers use GSEs as “one-stop shops” to search the internet for answers to a wide range of queries, including non-commercial queries. Tr. 3670:19–23 (Ramaswamy (Neeva)) (GSEs provide a one-stop shop for all information needs.); Des. Tr. 168:20–22, 168:25–170:1 (Moxley (Google) Dep.) (Users prefer a seamless and quick experience to get information quickly.); Tr. 4612:8–13 (Whinston (Pls. Expert)). Indeed, Google’s mission statement is “to organize the world’s information and make it universally accessible and useful.” Tr. 7641:6–20 (Pichai (Google)); Tr. 8224:4–11 (Reid (Google)) (“They [Larry and Sergey] really, really had the desire that you could come here and get any question answered.”). As Prof. Whinston explained, if consumers did not care about one-stop-shops, then Google would be better off providing vertical search engines. Tr. 10466:1–10467:3 (Whinston (Pls. Expert)).

328. Many consumers begin their search sessions on GSEs. Tr. 3670:6–18 (Ramaswamy (Neeva)) ( “[A] general search engine is I guess best defined in contrast to a specialized search engine, which is everything that you are looking for, that’s the first place that you can turn to . . . for the vast majority of your information needs.”); Tr. 4614:9–4615:16 (Whinston (Pls. Expert)) (77% of consumers on Windows PCs begin their search sessions on GSEs.).

329. Starting a search on a GSE saves consumers the time and mental energy of determining the best place to search. For this reason, a significant volume of searches on Google are “navigational queries,” meaning the queries are for the purpose of finding a specific website known to the user. Tr. 8407:15–8410:24 (Israel (Def. Expert)) (defining “navigational queries” as queries on a GSE entered to go to another website); id. 8748:19–8749:1 (navigational queries are a significant volume of the searches on GSEs).

By contrast, a consumer relying only on specialized search engines may need to type their query in multiple places before finding the best source or the right answer. Id. 8716:20–8717:3, 8717:11–23 (“If I don’t know the best source, I may have to try various ones. . . . if I don’t know, I have to look . . . if I don’t know, I have to figure it out.”).

330. GSEs allow users to pursue all their different queries in one place. Tr. 4612:14– 4614:8 (Whinston (Pls. Expert)) (GSEs provide “a single location for all queries”); Tr. 8674:14– 20 (Israel (Def. Expert)) (“GSEs is kind of the Swiss army knife to the browser.”). A consumer relying only on specialized search engines may need to navigate to a different search engine for each query. Tr. 10470:8–10472:9 (Whinston (Pls. Expert)) (GSEs reduce consumers’ mental energy in finding relevant information, especially where users “may just not know about where things are”).

331. Google recognizes the value of being a one-stop shop to consumers. Tr. 4612:14– 4614:8 (Whinston (Pls. Expert)) (discussing a quote from Marissa Mayer (then-Google executive) on UPXD102 at 7 stating that “Google should be a Swiss Army knife: Clean, simple, the tool you want to take everywhere.”).

Google’s documents recognize that (1) consumers use GSEs for both commercial (monetizable) and non-commercial queries—reflecting the one-stop shop nature of the GSE—and (2) improving its response to non-commercial queries can increase consumer reliance on Google for commercial queries. UPX0334 at -084 (Dr. Varian presentation stating that the general purpose search engine business model is to provide good answers for non-commercial queries so users will use Google for commercial queries)); Tr. 188:14–19 (Varian (Google)) (agreeing that “Google answers non-commercial queries because it hopes, at some point, the user will also type in a commercial query and Google can make money off of it”).

For example, one Google presentation regarding how to increase commercial queries and accelerate monetization identified either growing search queries generally—because “commercial queries also increase in a constant proportion”—or by investing in better noncommercial search experiences—because “[i]nvesting in non-commercial verticals spills over into commercial opportunities.” UPX0331 at -413, -417.

332. Browsers use GSEs as the default in the address bar because consumers value one-stop shops. Des. Tr. 217:3–8 (Baker (Mozilla) Dep.) (Mozilla set a GSE as the default in the Firefox browser because providing a search engine that does not respond to all different types of user queries would “not be appropriate” and the “user experience” would “not be good”); Tr. 1032:4–24 (Higgins (Verizon)) (due to user preference, default search engines should be GSEs); Tr. 10468:2–24 (Whinston (Pls. Expert)).

333. Consumers’ reliance on these browser defaults, in turn, shows that they value onestop shops. Tr. 10468:25–10469:16 (Whinston (Pls. Expert)); id. 4612:8–4615:16 (discussing UPXD102 at 7); UPX0090 at -940 (users value the convenience and ease of access that the Google default provides on the Safari search bar).

c) Only General Search Engines Are Preset As Default Search Engines

334. Only GSEs are preset as default search engines. Tr. 7425:3–7427:4 (Raghavan (Google)) (TikTok, Facebook, and vertical search providers such as Amazon cannot be default search engines because none are search engines “that do traditional web search.”); Tr. 1032:4–24 (Higgins (Verizon)); Tr. 8625:21–8626:4 (Israel (Def. Expert)); Tr. 4622:1–12 (Whinston (Pls. Expert)). Mitchell Baker, CEO of Mozilla, noted that Mozilla would only consider GSEs for the default in the Firefox browser because a vertical provider focused on just one category of content would not be an appropriate default. Des. Tr. 217:9–12, 217:15–23 (Baker (Mozilla) Dep.). Because browser users prefer a GSE as the default, setting a SVP or social network as the default would “not be a good [user] experience.” Id. 217:24–219:2.

335. In Google’s Chrome browser, for example, the list of options a user may select to replace the default search engine includes only GSEs. Tr. 7425:17–20 (Raghavan (Google)).

336. In Safari, Apple only offers users the choice of changing their default search engine from Google to another GSE. Tr. 2169:17–21 (Giannandrea (Apple)); Des. Tr. 82:7–22, 83:13–24 (Apple-EC 30(b)(6) Dep.) (Apple considers “general search functionality” when evaluating which search engines to include as options in Safari). Accordingly, Apple users can only change their Safari default to Yahoo, Bing, DuckDuckGo, or Ecosia. Tr. 2169:17–21 (Giannandrea (Apple)); Des. Tr. 54:25–55:2 (Perica (Apple) Dep.) Apple only allows users to v

d) Choice Screens Offer Choice Among Only General Search Engines To Ensure A Good User Experience

337. The few choice screens for selecting search providers on mobile devices have only offered users choice among GSEs. Choice screens only include GSEs to “ensure a good user experience.” UPX8091 at -505; Tr. 7425:17–20 (Raghavan (Google)); Des. Tr. 217:3–8 (Baker (Mozilla) Dep.); Tr. 4622:13–4623:8 (Whinston (Pls Expert)).

338. On Google’s Android choice screen in Europe, for example, only GSEs are eligible for inclusion. UPX8091 at -505 (Only “general search services,” and not “specialized or ‘vertical’ providers,” are eligible to participate in the EU choice screen to set the Android search default.); Tr. 4622:13–4623:8 (Whinston (Pls. Expert)) (discussing UPXD102 at 18–19 (citing UPX8091 at -505)).

i. Both The Industry And Public Recognize General Search Services Comprise A Relevant Market

339. Industry participants, including Google, browser companies, specialized search companies, and industry analysts, recognize a market for general search services, comprised of GSEs.

e) Google Recognizes General Search Services As Its Own Market And Has Done So For More Than 10 Years

340. In contracts, internal documents, and conduct, Google recognizes that GSEs comprise a relevant general search market; indeed, Google has tracked its market share in that market for many years, and does not view specialized search engines as competing for general search queries.

i. Google Has Long Tracked Its Market Share As Compared With Other General Search Engines

341. Google has, for many years, calculated and tracked its market share in the general search services market. Tr. 199:2–202:25 (Varian (Google)) (referring to UPX0902 at -020) (describing four different ways Google used to measure general search share); Des. Tr. 21:25– 23:11 (Google-NF 30(b)(6) Dep.) (describing Google’s measurement of country-specific general search market share); id. 26:23–28:15 (Google developed internal data source to estimate general search market share); id. 30:5–31:16 (describing general search market share analysis); UPX0408 at -041, -053 (estimating general search market share); UPX0849 at -619 (same); UPX0348 at -759–60 (2009 general search market shares); UPX0327 at -350–51 (2011); UPX0351 at -775, -777–78 (2017 to 2018).

342. For more than four years, beginning in August 2009, employees at Google sent monthly emails updating Google executives on general search market shares. Tr. 203:21–204:5 (Varian (Google)) (a team led by Penny Chu measured search share monthly). Marissa Mayer, then Google’s V.P. of Search Products & User Experience, circulated these general search market figures “[i]n advance of the earnings call and board meeting,” describing them as the “latest market share analysis.” UPX0499 at -297–98. These email chains show that others within Google recognized that GSEs constitute a relevant market. UPX0499 at -297.

343. As early as 2009, Google concluded that its market share in general search was over 70%. UPX0499 at -297 (“Our numbers paint the following picture of the landscape: Google 71%, Yahoo 17%, Bing 7%”).

344. Between 2017 and 2019, high-level Google executives regularly received quarterly Factpacks, prepared by a team reporting to Ruth Porat, Google’s Chief Financial Officer, which included Google’s market share in the general search market. Tr. 3702:9–3704:15 (Ramaswamy (Neeva)); UPX0475 at -744 (“User Adoption Metrics” Table from Q2 2018 Factpack showing Google’s US “desktop search query share” of 84% and “mobile search query share” of 97%); UPX0476 at -668 (same, with shares of 84% and 98% respectively for desktop and mobile).

345. As recently as July 2021, Google’s Chief Business Officer, Philipp Schindler, received these general search market share figures in advance of meetings with Google’s Board of Directors. UPX0909 at -167–68 (“BOD Prep for Philipp – Jul 2021” reflecting the User Adoption Metrics Table from Q2 2021, including desktop search query share of 84% and mobile search query share of 97%).

346. Google did not include companies like Amazon or Facebook in its market share calculations even though it had the ability to do so. Tr. 207:16–208:10 (Varian (Google)).

347. When Microsoft’s Bing launched in 2009, Google immediately analyzed the effect of Bing’s entry on market shares in a market comprising only GSEs. Tr. 3548:4–5 (Nadella (Microsoft)) (Bing launched in 2009); Des. Tr. 18:16–19:22 (Google-NF 30(b)(6) Dep.) (Google analyzed general search market share when Bing first launched); UPX0347 at -370–76 (Google tracked Bing search shares in 2010).

348. Under Mozilla’s contract with Google, Mozilla may customize Firefox’s navigation bar provided the customization does not involve sending a search query to an “excluded service provider,” defined as “Baidu, Bing, DuckDuckGo, Yahoo, and Yandex” without forfeiting the Google revenue share. JX0031 at -613, -617 (Mozilla Sponsorship Agreement (2016)).

ii. Google’s Chief Economist And General Counsel Recognized That General Search Services Is A Relevant Market

349. Google’s chief economist, Dr. Varian, has recognized that general search is a relevant market. Tr. 375:17–376:1 (Google (Varian)) (when describing the loss of competition by going from three to two players in UPX0180, he was describing the loss of competition in the “general search market”); UPX0180 at -451 (email from Dr. Varian stating that a potential deal between Yahoo and Microsoft would create “a loss of competition from going from 3 to 2 players”); UPX0901 at -064 (Hal Varian (Google)) (differentiating “special purpose search engines” such as Amazon and Yelp which focus on “commercial search” from “general purpose search engines like Google, Bing, Baidu, and Yandex”).

Indeed, during public talks covering antitrust issues, Dr. Varian has presented materials referring to GSEs as a competitive product category, separate and distinct from specialized search engines, artificial intelligence, and social networks. Tr. 185:20–186:9 (Varian (Google)); UPX0334 at -085 (chart on slide titled “Competition,” from presentation by Dr. Varian, identifying “general purpose search engines” as a distinct competitive product category).

350. In 2005, Google’s then-chief legal officer David Drummond wrote a letter to Microsoft complaining—in a section titled “Impact on the Market for Search Services”—that Microsoft’s use of defaults in Internet Explorer would result in Microsoft “gain[ing] a large number of search users for reasons having nothing to do with the merits of Microsoft’s search offering.” UPX0172 at -732; Tr. 7680:8–7687:20 (Pichai (Google)) (discussing UPX0172). In that letter, Mr. Drummond described the approach of selecting default options among top general search providers “by market share” as “sensible.” UPX0172 at -731.

iii. Google Recognizes Other General Search Engines As Its Closest Competitors

351. Google’s closest competitors are other GSEs. Dr. Varian recognized, “if Google were to disappear, people would just switch to Bing.” UPX0340 at -059; Tr. 195:22–196:5 (Varian (Google)) (explaining that in UPX0340 he was considering “in this experiment of eliminating Google, then what would people do”); id. 194:7–194:22 (Dr. Varian has a practice of writing in-line responses); Tr. 4623:11–20 (Whinston (Pls. Expert)). Dr. Varian concluded that, without GSEs, the world would look like a “universal library[,] but with no card catalog,” suggesting that users could not simply substitute to other types of products to replace the services GSEs provide. Tr. 194:7–196:25 (Varian (Google)) (Dr. Varian has a practice of writing in-line responses).

352. In internal documents, Google identifies other GSEs as its competitors. UPX0278 (detailed presentation titled “Google Search Competitive Factpack” comparing Google Search only to Bing); UPX0408 at -030–45 (Google “Competitive Analysis” identifying “privacyfocus[]ed search engines,” namely, DuckDuckGo and Qwant, as Google’s only privacy-focused competitors). Google does not view SVPs as competing for general search queries. Tr. 2164:25– 2165:14, 2362:23–2363:3 (Giannandrea (Apple)) (when Giannandrea was an executive at Google, from 2010 to 2018, he did not view vertical providers as competing for general search queries); Tr. 8622:21–23 (Israel (Def. Expert)) (Google gets queries that Amazon does not compete for).

iv. Google Analyzes The Quality Of Its General Search Engines As Compared To Other General Search Engines Such As Bing And DuckDuckGo

353. Google analyzes the quality of its GSE as compared with other GSEs. Tr. 8099:14–8100:8 (Gomes (Google)); Tr. 6367:16–24 (Nayak (Google)) (Google compares its quality to Bing’s); Tr. 8677:18–8678:13 (Israel (Def. Expert)). Google regularly runs analyses to see how its search results and ranking compare, for example, to Microsoft’s Bing. Tr. 6466:14– 6466:18 (Nayak (Google)) (Google compared itself to Bing in side-by-side analyses); UPX2033 at -000 (Side-by-side analysis comparing Google and Bing’s COVID search results); Des. Tr. 175:25–177:25 (Moxley (Google) Dep.) (Google would analyze how its ranking compared to Bing’s); 4621:2–12 (Whinston (Pls. Expert)) (commenting on UPX0268 at -182); UPX0268 at -132–33, -182 (presentation showing Google evaluating its search quality against other GSEs.).

354. Google compares its search product with other GSEs across axes of quality such as speed/latency. Tr. 7771:7–25 (Pichai (Google)); Tr. 6457:13–21 (Nayak (Google)) (Google’s slower latency than Bing had been raised and Google instituted a response); UPX2022 at -590 (Google comparing its latency with Bing’s). Privacy is another area in which Google compares itself to other GSEs. Tr. 4163:12–19 (Juda (Google)) (in UPX0811 at -420 Google compares itself with DuckDuckGo); UPX0811 at -420 (presentation on potential privacy initiatives in which Google compares its privacy offerings with DuckDuckGo’s).

355. Google’s comparisons with vertical search engines or Facebook are separate analytics for Google’s verticals. Tr. 8099:14–8100:8 (Gomes (Google)) (Google compares its IS quality scores against Bing and DuckDuckGo, but not against Amazon or Facebook); UPX0483 at -295 (Google “Competitor Intelligence” document separating analysis for verticals (including Amazon and Booking) from analysis for “traditional Search engines” (including Bing, Yandex, and DuckDuckGo).

v. Google’s Distribution Agreements Differentiate Between General Search Engines And Specialized Search

356. Google’s distribution agreements differentiate between GSEs and specialized search engines. Google’s exclusive agreement with Verizon between 2011 and 2014 defined “General Web Search” as “search functionality that produces search results by searching a large proportion of indexable web sites, . . .” and explicitly distinguished, “[f]or clarity, vertical and customizable search functionality . . . is not General Web Search.” JX0016 at -678; UPX6026 at -553 (written 30(b)(6) response: “‘[G]eneral web search,’ meaning services that ‘search[] the web for . . . web results to general queries.’” (second set of brackets in original)); UPX6030 at -621 (written 30(b)(6) response: “Examples of such services that deliver search results by searching the Internet in a manner substantially similar to Google Search include Bing and Yahoo Search.”).

357. “Alternative Search Service” in Google’s contracts “means any search service that is substantially similar to Google Search (as determined by Google in its reasonable discretion).” JX0095 at -689 (§ 1.3) (T-Mobile RSA (2021)); JX0062 at -177 (§ 1.7) (Motorola MIA (2020)). The “Alternative Search Services” prohibited under Google’s RSAs do not include specialized search services or social media.

Des. Tr. 185:19–25, 186:7–187:14 (Ezell (AT&T) Dep.) (the Facebook and Amazon Shopping apps, as verticals and not general search services, could be preloaded in compliance with the RSA); Tr. 8688:25–8689:9 (Israel (Def. Expert)) (the RSAs prohibit the preinstallation of GSEs like Bing, but do not prohibit the preinstallation of specialized verticals like Amazon and Yelp or social networks like Facebook, Instagram, and TikTok); Des. Tr. 260:20–261:3, 261:5–262:2, 262:5–7 (Levine (Google) Dep.) (the RSA Escalation Council determined the Amazon Shopping app would only be considered an alternative service once it incorporated full Alexa functionality (meaning some access to “general web search”); UPX0567 at -914–15 (RSA Escalation Council meeting notes stating that Amazon App may be preloaded only until it incorporates Alexa functionality, after which it will be treated as an alternative assistant service).

358. The 2014 Joint Cooperation Agreement (JCA), through which Google and Apple amended the ISA in 2014, recognizes GSEs as comprising a relevant market. UPX6026 at -555 (written 30(b)(6) response: “Google understands ‘other general search engine’ in the context of the JCA to refer to other general purpose search engines that may be referenced in third-party reporting service market share reports for individual countries.”); Des. Tr. 59:25–60:17, 60:19– 23, 60:25–61:19, 61:21–23 (Apple-EC 30(b)(6) Dep.); JX0024 at -822 (2014 Apple-Google Joint Cooperation Agreement). The JCA allows Apple to request a carveout from the parties’ agreement for any country where “Google’s usage share compared with other general search engines only declines to [redacted]% or less for [redacted] consecutive years.”

UPX6026 at -555 (written 30(b)(6) response: “Google’s understanding of ‘usage share’ as used in the JCA is that ‘usage share’ refers to measurements of general-purpose search engine usage in commercially available third-party reporting services, such as comScore.”); Des. Tr. 55:10–56:7 (Apple-EC 30(b)(6) Dep.); JX0024 at -822.

vi. Google’s Android RSAs Restrict Preinstallation Of Other General Search Engines, But Not Specialized Search Engines Or Social Networks

359. Google’s Android RSAs restrict preinstallation of other GSEs but not products such as specialized search engines or social networks. Tr. 8688:25–8689:9 (Israel (Def. Expert)); Tr. 4621:13–25 (Whinston (Pls. Expert)); e.g., JX0071 at -405 (§ 9.3(a)) (Samsung RSA (2020)) (exempting vertical search from contract provisions applying to alternative search services, unless the service is “incorporating a set of vertical searches”); UPX0322 at -879 (Google’s proposal for Samsung allowing vertical providers like Expedia to serve results to certain Bixby queries).

360. For example, Google’s exclusive agreement with Verizon between 2011 and 2014 provided that the “placement and General Web Search Access Point requirements only pertain to applications and services that provide 'General Web Search' services." JX0016 at -678 (limiting to GSES the placement requirement that Google be set "as the default search provider for all General Web Search access points" on Verizon devices); Tr. 1035:14–1037:13 (Higgins (Verizon)) (discussing JX0016).

f) Other Industry Participants And Industry Analysts Recognize General Search Engines As Its Own Market

361. Other industry participants recognize general search services as a relevant product market. Tr. 4761:6–4762:17, 4614:9–4616:8, 5744:21–5745:20 (Whinston (Pls. Expert)).

i. Industry Analysts Recognize The General Search Services Market

362. Industry analysts calculate market shares for the general search services market. Tr. 199:12-22 (Google (Varian)) (ComScore sells market share data about search engines); Tr. 4619:21-4620:7 (Whinston (Pls. Expert)); Tr. 8684:7-8685:1 (Israel (Def. Expert)); UPX0336 at -474 (referring to ComScore's measure of Google's market share “compared to other general search engines only"). Industry participants rely on calculations of GSEs' market share, for example from StatCounter, ComScore, and others, and consider these reliable. Tr. 3832:6-25, 3834:2-18 (Lowcock (IPG)) (discussing UPX0450, an IPG client document providing a graph listing “search engine market share” in the U.S. among Google, Bing, Yahoo, DuckDuckGo, Ecosia, and Yandex).

ii. Other General Search Engines Recognize The General Search Services Market

363. Other (non-Google) GSEs recognize GSEs as providing a product and service distinct from specialized search engines. Des. Tr. 25:22–26:14 (Ramalingam (Yahoo) Dep.) (search engines focused on providing particular results were called "verticals" and they are separate and apart from general search).

364. Further, other (non-Google) GSEs describe GSEs, not specialized search engines, as their closest competitors. Microsoft considers Google to be Bing’s only competition. Tr. 3098:6–3099:3 (Tinter (Microsoft)) (when thinking about Bing’s search competitors “it’s one company, it’s Google”); Tr. 6221:5–6223:2 (Barrett-Bowen (Microsoft)) (Bing does not view vertical partners as “competitors” even though they may have user overlap in some instances). DuckDuckGo’s CEO similarly testified that its competitors include other GSEs, but not specialized search engines such as Amazon. Tr. 1942:11–1942:21 (Weinberg (DuckDuckGo)).

365. Other (non-Google) GSEs compare themselves to GSEs. For example, as testimony from Microsoft executives confirms, Bing compares itself with Google and other GSEs, including in side-by-side comparisons in various markets, among them the United States. Tr. 2718:19–2719:4 (Parakhin (Microsoft)); UPX0832 at -501 (comparing OneSearch’s features to Google, Bing, and Yahoo Search); DX0524 at -762 (comparing Bing’s search engineering team size against Google’s); PSX01148 at -390 (the “Bing Challenge” is a side-by-side comparison done in the United States).

iii. Consumers Recognize General Search Engines As Comprising A Relevant Product Market

366. Based on usage, consumers recognize GSEs as a relevant product market. Tr. 4614:9–25 (Whinston (Pls. Expert)) (empirical evidence shows that 77% of first searches and sessions are on a GSE and therefore consumers are using it as a gateway to the entire internet).

iv. Other Market Participants Recognize The General Search Market

367. Google’s partners recognize GSEs are distinct from specialized search engines. Tr. 1030:13–1032:2 (Higgins (Verizon)). Apple, for example, recognizes that specialized search engines that focus on a particular subject are not GSEs. Des. Tr. 59:25–60:19 (Apple-EC 30(b)(6) Dep.). Carriers compare the relative quality of GSEs and do not consider specialized search engines as substitutes for the default position on their devices. For example, when Verizon investigated its options for replacing Google on Verizon’s devices, it evaluated Google’s quality as compared with only Yahoo and Bing. UPX0622 at -064–65 (noting that “Bing does outperform Google” in certain areas).

Carriers do not see the value in making specialized search engines the default search engine on devices because users prefer GSEs. Tr. 1032:3–24 (Higgins (Verizon)). And as an AT&T executive testified, AT&T viewed Bing as the only real alternative to Google. Des. Tr. 240:6–241:4 (Ezell (AT&T) Dep.).

368. Browser companies compare the relative quality of GSEs and do not consider specialized vertical search engines as a substitute. Des. Tr. 63:17–22 (Baker (Mozilla) Dep) (explaining Mozilla in 2014 compared the quality of Yahoo, Google, and Bing); Tr. 2167:21– 2169:11 (Giannandrea (Apple)) (explaining the differences between GSEs (i.e., Bing, Yahoo, you.com, and DuckDuckGo) and specialized search engines (e.g., Facebook, Amazon, and TikTok); id. 2170:21–2171:13 (Safari’s URL bar is intended to default to a GSE, and not a specialized search engine).

v. Specialized Search Engines Recognize They Participate In Markets Separate From General Search Engines

369. Specialized search engines identify other specialized search engines, not GSEs, as their competition. Tr. 5346:1–14 (Dijk (Booking.com)) (Google Search is not a competitor of Booking.com, but specialized verticals like Google Travel are); Des. Tr. 95:15–16, 95:18–96:3, 96:5–10, 96:12–25, 284:15–285:7 (Dacey (TripAdvisor) Dep.) (TripAdvisor competes with OTAs, hotels, and Google’s travel product, not Google’s general search).

2. General Search Services Require Unique Production Facilities

370. The uniqueness of a GSE stack supports a finding that general search services comprise a relevant product market; specifically the ability to provide responses based on independent or syndicated capabilities including crawling and indexing a large proportion of the web and retrieving information relevant to general queries from across the broader internet. Supra ¶¶ 65–72 (§ III.A.1).

3. A Monopolist Of General Search Services Would Be Able To Sustain Quality Significantly Below The Competitive Level

371. A GSE with a monopoly position in the general search services market would be able to sustain prices significantly above, or decrease quality significantly below, the competitive level. Tr. 4603:7–4616:25 (Whinston (Pls. Expert)).

a) Google’s User Experiments Support The Finding Of General Search Services As A Relevant Product Market

372. Google has conducted quality degradation experiments, intentionally increasing latency, or otherwise decreasing quality for some users, to study their responses. UPX1082 at -294.

These experiments show little user responsiveness to decreases in Google’s quality, suggesting that potential search alternatives like specialized search engines or social networks are not a strong competitive constraint on Google. UPX1082 at -294; Tr. 4770:23–4772:24 (Whinston (Pls. Expert)) (when Google creates significant reductions in search engine quality, substitution to other alternatives is limited); id. 10469:17–10470:5.

This low user responsiveness to decreases in quality indicates that a GSE monopolist could sustain quality significantly below the competitive level. UPX1082 at -294 (Google presentation summarizing quality degradation experiments showing that a reduction in 1 IS point of quality is expected to cause a [redacted] decline in search volume).

4. General Search Engines Have No Reasonably Interchangeable Alternatives

373. Firms such as Amazon, Facebook, Instagram, and TikTok are not reasonable substitutes for GSEs. Tr. 4618:4–4619:20, 6153:12–18 (Whinston (Pls. Expert)). Notably, such firms, including specifically Amazon, have endeavored to create offerings competitive with Google Search and have been unsuccessful. UPX0432 at -345 (Google competitive analysis stating "Amazon tried and failed to create a competitive search offering"); Des. Tr. 78:17-79:21 (van der Kooi (Microsoft)). Further, the growth of vertical search over the past decade has not dented Google Search's extraordinary revenues and profits. UPX7002.A at -774 (booked revenue for Google Search increased from $46 billion in 2014 to $146 billion in 2021).

a) Social Media Is Not a Reasonable Substitute For General Search

374. Social media is not a reasonable substitute for GSES. UPX0346 at .001 (search users on YouTube differ from those on google.com because they are in a different mindset); Des. Tr. 218:19-219:2 (Baker (Mozilla) Dep.) (Firefox and its users would not consider Facebook a substitute for a GSE); Tr. 5241:12–5243:17 (Dijk (Booking.com)) (Facebook, Instagram, and Tiktok are not GSEs); Tr. 4618:4-4619:20 (Whinston (Pls. Expert)).

375. Social media platforms do not provide information directly from the open web. Tr: 4618:12-4619:20 (Whinston (Pls. Expert)). Unlike GSES, social media platforms search only their own content. Tr. 2168:5–2168:22 (Giannandrea (Apple)); Des. Tr. 78:9–79:21 (van der Kooi (Microsoft) Dep.) (describing sites such as Facebook as "catalog search," distinct from "generic internet search"); UPX0334 at -085 (chart by Dr. Varian in which Google and Microsoft, but not Facebook are identified as operating general purpose search engines). For example, "if you do a search on TikTok, what you get back is TikTok content." Tr. 7420:22-25 (Raghavan (Google)).

376. On social media platforms, users typically do not enter a query or receive a SERP; instead, users are presented with an information feed. UPX0033 at -120 (Facebook, Instagram, TikTok, and Twitter are “[1]ean-back, low intent content consumption feed platforms”); UPX2114 at -761 (in 2021, Facebook's Vice President of Ads Engineering wrote: "people don't come to core [Facebook] products with intent (unlike search), from the beginning, we knew we were about demand generation, not demand fulfillment”); UPX0445 at -507–08. As Dr. Israel conceded, TikTok does not provide a SERP, and therefore would not be a substitute for “any user that wants a SERP.”

Tr. 8675:20–8676:3 (Israel (Def. Expert)). Accordingly, social media platforms provide consumers with different and less reliable information than GSEs. Tr: 4618:12–4619:20 (Whinston (Pls. Expert)); id. 6153:12–18 (comparing DXD-15 at .006 to DXD-15 at .016 and describing differences between Google and Facebook results for the query “running shoes”).

377. Search users do not substitute between Facebook and Google. This is because “[s]earch users are typically looking for something specific,” while “social users are interested in what social communities or their peers are doing, saying, buying or debating.” UPX0409 at -254. An internal Google study showed that users who increase their use of Facebook use Google more often, rather than less. UPX0902 at -020; Tr. 197:4–199:1 (Varian (Google)) (unaware of more recent studies on the impact of Facebook on Google Search than UPX0902); UPX0904 at -148– 56 (Facebook usage positively correlated with Google searches).

b) Specialized Vertical Providers (e.g., Amazon, Yelp) Are Not Reasonable Substitutes For General Search Engines

378. Specialized Vertical Providers (SVPs) are also not reasonable substitutes for GSEs. Des. Tr. 141:11–17 (Miller (Google)). Where a GSE provides answers without any specific domain restriction, an SVP will either be focused on a particular domain, or on a particular aspect or vertical. Des. Tr. 27:25–28:1; 28:3–28:10 (Ramalingam (Yahoo) Dep.).

The results returned to a query entered into an SVP would be different than those returned to the same query entered into a GSE. Des. Tr. 29:16–19, 29:21–30:3 (Ramalingam (Yahoo) Dep.). Google’s internal documents recognize that the core competencies of SVPs differ from those of Google’s GSE. UPX0329 at -378.004–378.005 (Dr. Varian’s chart showing the capabilities across industries for Apple, Google, Amazon, Facebook, and Microsoft; Google and Microsoft are the only two companies identified as competing as “general purpose search engines,” while Amazon, Apple, Google, and Microsoft are identified as competing in a separate “special purposes search engines” category); UPX0332 at -673. Also, there are many queries that an SVP simply cannot answer.

Des. Tr. 276:25–278:7 (Dacey (TripAdvisor) Dep.) (TripAdvisor cannot answer various sets of queries (e.g., news, sports scores, stock prices, clothing, home repair, gas prices, historical info, movie reviews, health info); Tr. 8715:25–8716:11 (Israel (Def. Expert)) (discussing Yelp’s provision of a null response to a query).

In fact, specialized search engines can be complementary to Google’s GSE. UPX0344 at -058 (“our [Google’s] analysis indicates that [users active on Amazon and large online retailers’ sites are] likely incremental to rather than cannibalizing their activity on G.com”); UPX0345 at -792 (stating Google “could not see any significant negative incremental impact from Amazon”); Tr. 8736:5–9 (Israel (Def. Expert)) (There are “elements of complementarity between” Amazon and Google.).

379. Unlike GSEs, many specialized search engines only permit searches in predefined ways. Tr. 6511:8–23 (Hurst (Expedia)) (Expedia Group “only works with structured queries,” which include information like “where you’re going, what dates you’re going, and how many people are going.”). On the other hand, Google “can respond to any query that you type into its text box” and is “returning results from anything it has scraped on the Internet.” Id. 6511:8–23; Tr. 4616:9–25 (Whinston (Pls. Expert)) (searches on many specialized search engines “have to be done in predefined ways” while “[i]n a general search engine, you can put anything in”).

380. Specialized search providers like Yelp, Expedia, or Kayak, search only “a very narrow set of content.” Tr. 8096:23–8097:14 (Gomes (Google)) (Google and Bing serve broader information interests than Amazon, which services only product listings); Tr. 8266:19–8268:23 (Reid (Google)) (unlike GSEs, Expedia, Yelp, Amazon, and DoorDash are not aiming to answer all queries); Tr. 9144:4–12 (Holden (Google)) (“Metasearch engines are … another level of aggregation, you could say, in—where a consumer can go to look for travel specific information”); Tr. 2168:23–2169:11 (Giannandrea (Apple)); Des. Tr. 181:6–182:15 (Stein (IAC) Dep.) (Angi is focused on connecting consumers with service professionals).

381. Many specialized search engines, including Amazon, only search their own content and not the worldwide web. Tr. 2168:15–17, 2169:3–2169:8 (Giannandrea (Apple)); Des. Tr. 278:8–13 (Dacey (TripAdvisor) Dep.) (TripAdvisor does not crawl or index the web, and has no plans to); Des. Tr. 182:8–15 (Stein (IAC) Dep.) (Angi does not crawl or index the web, nor does it have any plans to do so). As one Microsoft executive testified, sites such as Amazon and YouTube provide “catalog search,” distinct from “generic internet search.” Des. Tr. 78:9–79:21 (van der Kooi (Microsoft) Dep.); Tr: 4616:9–25 (Whinston (Pls. Expert)) (many specialized search engines are “not sending you off to other sites, they don’t have a broad index of the web”).

382. Even in the shopping vertical, consumers often use Google and Amazon for different, complementary tasks. Tr. 7434:22–7435:14 (Raghavan (Google)) (there is a correlation between Google search usage and Amazon Prime membership because “Prime members who in any way intend to shop at Amazon might come to Google and do a lot of research before they do it”); Tr. 8097:23–8098:1 (Gomes (Google)) (users can search for a plumber on Google but not Amazon); PSX00267 at -258–59 (Google presentation titled “Google and Amazon in US Retail” showing Amazon Prime members are more likely to use Google, suggesting Google search is a complement to searches on Amazon).

383. Most specialized search engines serve only commercial queries. Des. Tr. 39:15– 41:19 (Google-EM 30(b)(6) Dep.) (“the results from Expedia would not be useful” for generalinformation queries); Des. Tr. 110:10–24 (Fox (Google) Dep.) (defining a commercial query as one where the user is likely “looking to conduct . . . a transaction for a product or service”); Tr. 8396:16–8398:17 (Israel (Def. Expert)). For Google, however, commercial queries make up only 20% of all queries. Id. 8725:15–8726:8; UPX0010 at -053 & n.6 (As of January 2020 “the vast majority of queries [–about 80%–] don’t show any ads at all.”); Tr. 5876:21–5877:11 (Whinston (Pls. Expert)).

384. Industry participants recognize that specialized search engines such as Amazon are not GSEs. Tr. 7426:15–20 (Raghavan (Google)) (Amazon is not a “general purpose search engine”); Tr. 183:16–18 (Varian (Google)) (Amazon does not provide a GSE); Des. Tr. 37:2–19 (Google-EM 30(b)(6) Dep.) (identifying only Google, Bing, DuckDuckGo, Ecosia, Baidu, and Qwant as the companies Google refers to as “search engines”); Tr. 5230:20–23 (Dijk (Booking.com) (Booking.com is not a GSE, but rather an e-commerce platform); id. 5241:12– 5243:20 (stating he does not consider Facebook to be a GSE); Tr. 2167:5–2169:11 (Giannandrea (Apple)); UPX0911 at -875 (2019 email written by Joan Braddi (Google) stating that “Amazon is not considered a search site”).

385. Google’s analyses show that search users do not substitute between Google and shopping specialized search engines such as Amazon. One analysis, code-named Project Charlotte, found loyalty club membership (e.g. Amazon Prime) and other engagement with online retailers correlated with more, not fewer, queries on Google. Tr. 7430:2–7435:20 (Raghavan (Google)) (“Prime members who in any way intend to shop at Amazon might come to Google and do a lot of research before they do it. So we will see that correlation.”) (discussing UPX0344 at -058-60); Des. Tr. 141:11-17 (Miller (Google) Dep.) (internal Google research shows "more active users on Amazon are also more active users on [Google] Search.").

Project Charlotte followed an earlier 2018 analysis concluding that Google observed a $[redacted] per user uplift in search revenue among signed-in U.S. Android users who installed Amazon's app and were likely to be active on it during the following 28 days. UPX0335 at -687.

386. A second study found a positive correlation between Amazon app use and query volume, concluding Amazon app users were more likely to be regular and frequent Google users. Tr. 8733:1-8734:13 (Israel (Def. Expert) (discussing PSX00562 at -966). The study went on to conclude that a user's adoption of any of six major apps—Amazon, eBay, Walmart, Pinterest, Spotify, or Twitter—was correlated to increased revenue and queries on Google mobile, and no significant change on desktop behavior. Id. 8737:24-8738:19 (discussing PSX00562 at -976-77); Des. Tr. 146:7-147:23 (Miller (Google) Dep.) (“[P]eople who installed these apps saw increased usage on Google both for mobile queries and for shopping queries.") (discussing PSX00562 at -977).

387. Specialized search engines are among the top advertisers on Google, but Google Search does not advertise on specialized search engines. Tr. 193:6–18 (Varian (Google)); Des. Tr. 220:21-221:6 (Soo (OpenTable) Dep.) (OpenTable does search engine marketing on Google.); Tr. 4614:9-4615:16 (Whinston (Pls. Expert)) (Specialized search engines are the biggest advertisers on Google and Bing.); Tr. 8741:16–23 (Israel (Def. Expert)) (Amazon and Yelp clearly believe it is effective to advertise on Google.).

Indeed, Amazon is the top advertiser on Google whereas Google Search does not advertise on Amazon. DX1121 (data file titled “RFP Request 42(b) - Top 250 Search Advertisers (2012–2021).xlsx") (Amazon was the top advertiser on Google in 2020 and 2021); Tr. 7441:1–7442:1 (Raghavan (Google)) (As one of the largest advertisers on Google, Amazon spends billions of dollars).

c) Apple’s Suggestions Is Not A Reasonable Substitute For General Search Engines

388. Apple’s Suggestions is not a substitute for a GSE. As an initial matter, like TikTok and other social media firms, Apple Suggestions do not produce a SERP. Tr. 2223:9–10 (Giannandrea (Apple)). In addition, Apple’s Suggestions only appear on [redacted] of user queries. Id. 2250:8–2251:9.

389. Similarly, Apple’s Spotlight is not a reasonable substitute for a GSE. Spotlight’s primary focus is locating content on the user’s device. Tr. 2204:23–2205:3 (Giannandrea (Apple)) (Spotlight focuses on finding content on the device first and then falling back to search); id. 2205:16–21 (because it does not search the web in the first instance, Spotlight is not a GSE).

d) Voice Assistants Including Apple’s Siri Are Not A Reasonable Substitute For General Search Engines

390. Voice assistants, including Apple’s Siri, are not a reasonable substitute for GSEs. As an initial matter, voice assistants such as Apple’s Siri do not provide users with a SERP. Tr. 2237:22–23 (Giannandrea (Apple)). Further, and unlike GSEs, voice assistants primarily enable users to conduct tasks, such as sending messages. Id. 2236:23–2237:10 (Mostly, Siri plays music, sends texts, or turns on the living room lights.); Des. Tr. 292:23–293:11 (Connell (Microsoft) Dep.) (Siri and Cortana are used to “command the device.”); UPX0320 at -621 (Google Assistant “helps you get things done.”). Voice assistants, and Siri specifically, do not compete with GSEs today. DX0609 at -326 (“Siri doesn’t compete directly against Google search.

It competes against Alexa and Google Assistant. The query mix is very different.”); Tr. 2236:13–15 (Giannandrea (Apple)) (Siri is Apple’s voice assistant.); id. 2237:20–21 (Siri is not a GSE.); Des. Tr. 155:9–156:4, 280:20–281:13 (Ezell (AT&T) Dep.) (Google’s contracts allow Samsung’s voice assistant—Bixby—to be preloaded on devices as long as it uses Google’s GSE by default.).

e) Generative AI Systems Are Not A Reasonable Substitute For General Search Engines

391. Generative AI systems, like ChatGPT, are not a reasonable substitute for GSEs. These systems can be used to summarize results into a single answer, but they are not good at identifying which results need to be summarized. Tr. 3696:15–3697:21 (Ramaswamy (Neeva)) (“AI [lets] you do . . . things like summarization, presenting a single answer in ways that, honestly, search engines of old could not do. But the middle problem of figuring out what are the most relevant pages for a given query in a given context still benefits enormously from query click information. And it’s absolutely not the case that AI models eliminate that need or supplant that need.”).

392. Google and Bing’s Generative AI tools, like Bard and Bing Chat, still rely on traditional search systems to provide answers to user’s questions and queries; thus, they do not replace traditional search. Tr. 2670:19–2671:9 (Parakhin (Microsoft)) (“The large language model [in Bing Chat] is used for reasoning and for providing the answer, but the base information is coming from search.”); id. 2670:10–18 (Bing Chat marries the functionality of ChatGPT and Bing); Tr. 8331:18–24 (Reid (Google)) (Google relies on the search index to verify or confirm Search Generative Experience (AI) responses against the results that it gets from Google search.); Tr. 3529:25–3530:10 (Nadella (Microsoft)) (Bing Chat “absolutely” still relies upon traditional search.).

393. Generative AI is a nascent technology. Tr. 8219:6–24, 8276:10–24 (Reid (Google)); Tr. 7531:15–7532:8 (Raghavan (Google)) (“And I view this as a journey, not as something that happened overnight. . . . And that’s a journey that we’re still early on.”); Tr. 8333:16–8334:4 (Reid (Google)) (Google’s “[Search Generative Experience] is very much in early stage.”); id. 8270:23–25, 8283:14–20 (agreeing that Google’s chatbot, Bard, is “experimental”); UPX2065 at -424 (Blog post by Google’s CEO Mr. Pichai: Bard is “experimental.”).

It is unlikely that Generative AI tools will displace the need for traditional search. Tr. 7528:25–7530:8 (Raghavan (Google)) (does not believe that in 10 years people will be doing everything through chatbots and large language models); id. 7530:25–7531:8 (AI through chatbots and large language models have not created a whole new world and caused the old world to go away.) (discussing UPX2040 at -299).

394. There are several drawbacks to using Generative AI-based tools instead of traditional search. First, because of the time and expense of training Generative AI systems, those systems lack the ability to provide fresh information. Tr. 8279:19–21 (Reid (Google)) (agreeing that chatbots “are not trained super frequently”); id. 8281:13–16 (agreeing that people in the chatbot industry criticize chatbots “as having the ability to be stale because it takes such a long time and is so expensive to train them”).

As Dr. Israel conceded, the information available on ChatGPT is approximately a year old, in contrast to the information available on Google, which is current. Tr. 8706:14–8707:1 (Israel (Def. Expert)). Second, Generative AI systems often produce inaccurate information. Tr. 8219:6–24 (Reid (Google)) (“[Google’s chatbot] does have some challenges.

It’s not always accurate because it is fundamentally predicting what it should say. And it has some error problem[s], and so we have to be thoughtful about how we roll it out.”); id. 8285:9–12 (agreeing that “[Large Language Model] experiences, Bard included, can hallucinate and present inaccurate information as factual”).

Thus, Bard carries a disclaimer that says “Bard may display inaccurate or offensive information.” Id. 8283:21–25; UPX2067 at -445–46 (listing “[k]nown limitations” of chatbots, including “misrepresent[ing] facts,” “inaccurately identify[ing] insights” and “[b]ias”). Third, Generative AI systems, in contrast to GSEs, respond to user queries slowly. Tr. 8334:5–8 (Reid (Google)) (recognizing that “[s]peed is a limitation” of chatbots).

395. Users have not replaced their use of traditional search with chatbot-based search tools, like Google’s Bard or Microsoft’s Bing Chat. Tr. 8326:18–24 (Reid (Google)) (Google’s chatbot receives only a “small portion” of Google’s total daily users, at around 8 to 10 million.)

5. The United States Is The Relevant Geographic Market For General Search Services

396. The United States is the relevant geographic market for GSEs. Tr. 4654:3–21 (Whinston (Pls. Expert)). Google recognizes that the United States is a relevant geographic product market for the provision of general search services. Des. Tr. 21:25–23:11 (Google-NF 30(b)(6) Dep.) (describing instances where Google would measure country-specific general search market share). The product provided to the consumer—the SERP—differs materially as between the United States and countries abroad.

For example, search results provided outside of the United States are often in languages other than English. Id. 21:25–23:11 (describing the importance of adapting to the Russian language to develop a good product in Russia).

Consumer intent when entering a specific search query often differs abroad, and the SERP returned to a consumer in London, England, may include different content than the SERP provided to a customer in the United States. Tr. 7360:25–7362:2 (Raghavan (Google)) (a user in London entering the query “mousetrap” may be “looking for theater tickets to a long-running play in London called ‘Mousetrap’).

397. Internal documents at both Google and Bing, including the contracts at issue, recognize that the United States is a relevant geographic market for general search services. For example, the 2014 JCA between Google and Apple contemplated calculating GSE usage shares by country, and [redacted]. JX0024 at -822 (§ 1) (Apple JCA (2014)); JX0097 at – 357 (§ 1) (Apple ISA (2021 amend.)). Microsoft documents also make assumptions and calculate shares specifically in the United States. UPX0115 at -142 (Microsoft assumptions based on market shares calculated in the U.S. general search services market).