paint-brush
Google’s Contracts Reduce Investment And Innovation Among Market Participantsby@legalpdf
120 reads

Google’s Contracts Reduce Investment And Innovation Among Market Participants

by Legal PDF: Tech Court CasesAugust 13th, 2024
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

When deciding whether to invest in improvements in search quality, Google compares the costs of investment to the benefits, such as likely returns in revenue or queries.
featured image - Google’s Contracts Reduce Investment And Innovation Among Market Participants
Legal PDF: Tech Court Cases HackerNoon profile picture

United States of America v. Google LLC., Court Filing, retrieved on April 30, 2024, is part of HackerNoon’s Legal PDF Series. You can jump to any part of this filing here. This part is 28 of 37.

B. Google’s Contracts Reduce Investment And Innovation Among Market Participants Including Google

1067. Market participants use a cost-benefit analysis when they think about search investments. Tr. 5838:13–5839:18 (Whinston (Pls. Expert)).


1068. When deciding whether to invest in improvements in search quality, Google compares the costs of investment to the benefits, such as likely returns in revenue or queries. UPX0891 at -884 (“The key proxy metric we will highlight for the purpose of this analysis is IS . . . . In 2019, finance estimated that a 0.1 IS metric gain may give approximately [redacted] in additional annualized gross revenue (based on the 2016 study).”); UPX0198 at -325 (“We need to do a cost vs benefit analysis of features at every launch, because we can’t launch everything due to resource constraints.”); UPX0215 at -953 (“It is anticipated that IS gains from 2020 launches in aggregate . . . will range between [redacted], which a study suggests will have a revenue impact of between [redacted].”); Des. Tr. 207:21–209:18 (Moxley (Google) Dep.) (Google considers “a number of factors in deciding what [projects] to prioritize,” including “business opportunity . . . [and] query volume” as well as the cost of a potential project, which could include the personnel allocation for a particular project). This cost–benefit analysis is the “basic model of what determines investment in economics.” Tr. 5837:4–5838:12 (Whinston (Pls. Expert)).


1069. Google’s rivals and potential entrants similarly use cost–benefit analyses when deciding whether to invest in search-quality improvements. Tr. 2642:23–2643:2 (Parakhin (Microsoft)) (search investment “fundamentally . . . boils down to what kind of a long-term revenue we can achieve”); Tr. 2344:3–20 (Giannandrea (Apple)) (“[I]f you were going to make a multibillion-dollar investment [in a search engine], you would need to have some business justification for it.”).


1070. Google’s contracts reduce incentives to invest and innovate because they “reduc[e] the benefits” of investing and innovating. Tr. 5837:4–5838:12 (Whinston (Pls. Expert)). With the contracts in place, the increase in traffic a rival will get from the investment or innovation is less than in a competitive world. Id. 5837:4–5838:12. This is why, with Google’s contracts in place, investments and innovation are less attractive to current rivals, Google, potential entrants, and distributors. Id.

1071. If firms lack incentives to invest to improve their quality, they will not improve and they will be weaker, less effective competitors. Tr. 5840:20–5841:3 (Whinston (Pls. Expert)).


1072. Google’s contracts reduce rivals’ incentives to invest in improving search. When asked what criteria he considered when making investment decisions for search, Mr. Parakhin testified that “fundamentally it boils down to what kind of a long-term revenue can we achieve,” listing factors including distribution and competitive position in terms of quality.


Tr. 2642:23– 2643:8 (Parakhin (Microsoft)). He explained that “distribution is basically an ability to get [general search] services in front of user[s];” without the “ability to effectively distribute, it’s almost meaningless to invest in [search].” Id. 2643:9–23; Tr. 3495:23–3497:12 (Nadella (Microsoft)) (Because Microsoft can get distribution on desktop devices, it makes more sense to invest in improving desktop search over mobile, where it cannot get distribution.); Des. Tr. 162:11–163:17 (van der Kooi (Microsoft) Dep.) (Investing in infrastructure and local search quality is a fixed investment, and “if [a general search service] lack[s] the volume of usage, it becomes prohibitive to participate and to make those fixed-costs investments.”).


1073. As Prof. Whinston explained, when DuckDuckGo thinks about how much to invest in improving quality, it considers how much it can possibly get as a benefit from that investment, and to the extent that these contracts are in place and tie up the market, there is less that DuckDuckGo can foresee as a benefit of investing. Tr. 5782:9–5783:14 (Whinston (Pls. Expert)).


1074. Lack of access to search users makes it more likely competitors will reduce search engine capabilities or exit the market altogether. According to the Chief Strategy Officer of Ask.com, Mark Stein, Ask’s inability to access users was a factor in Ask’s decision to stop crawling and indexing the web. Des. Tr. 270:10–271:7 (Stein (IAC) Dep.).


1075. Bing’s inability to access mobile users due to Google’s contracts has reduced Microsoft’s incentives to invest and improve search, particularly on mobile devices. As one Microsoft executive put it: “It is uneconomical for Microsoft . . . to invest more in mobile [search quality]” because “no amount of investment without securing some way to do distribution in mobile will result in any share gain.”


Tr. 2750:25–2751:11 (Parakhin (Microsoft)); Tr. 3496:17– 3497:12 (Nadella (Microsoft)) (After fixed cost investments to build out the infrastructure required for general search, “it quickly becomes how do you get enough queries to continuously improve search quality.”); Des. Tr. 112:9–25, 113:7–114:10 (Ribas (Microsoft) Dep.) (“[O]ne of the challenges we always had was, especially in mobile where we didn’t have the volume, how can we justify going to spend hundreds of millions of dollars or billions of dollars in – in more data centers and, you know, in more infrastructure when . . . we don’t really have the volume, where we’re locked from getting the volume?”).


1076. Microsoft has invested in improving desktop search because Microsoft has better access to distribution on desktop than on mobile. Tr. 3496:17–3497:12 (Nadella (Microsoft)) (Microsoft has focused on making both the necessary fixed costs and dynamic data acquisition on desktop where it has better access to consumers.); id. 3525:4–3526:9 (in the 2018 negotiations with Apple, Microsoft pointed out that on desktop, which is more open than mobile, Microsoft has won share and improved quality); UPX0236 at -340 (“We have shown that we can compete with Google on quality in [English] markets, and by attaching ton Windows PC distribution we managed to grow US share for 84 consecutive months.”)


1077. Although Microsoft has made significant investments in search, Mr. Nadella explained that for him to justify investing even more there must be an opportunity to compete for more mobile users. Tr. 3508:6–18 (Nadella (Microsoft)) (“Because, one, in order to start investing—we’re making all the fixed cost investments already. And for us to make additional fixed cost investments on, say, search relevance or search scale, we needed to get more of the dynamic data and mobile distribution.”).


1078. In fact, Microsoft had a “playbook” ready for the investments it would make if it could gain access to more mobile users. Tr. 3252:13–3255:3, (Tinter (Microsoft)) (“[W]e know what investments we need to make, we knew what engineering work we needed to do that we could improve relatively quickly.”); id. 3256:17–3258:13 (describing UPX0797).


Obtaining access to more mobile users would be “game changing” and lead to a significant investment. Tr. 2723:4–6 (Parakhin (Microsoft)) (discussing a potential distribution deal with Apple); Des. Tr. 94:11–19 (Ribas (Microsoft) Dep.) (A distribution deal with Apple would increase Microsoft’s incentives to invest in search.).

2. Google’s Contracts Reduce Google’s Own Incentives To Invest

1079. Google’s contracts reduce its own incentives to invest in search. Tr. 5782:9– 5783:14 (Whinston (Pls. Expert)) (“Google is going to run faster if it has more competition.”).


Google’s contracts directly reduce its incentives to invest because, with the contracts in place, Google is less likely to lose users if it fails to improve or maintain its search quality. Id. (Google doesn’t have to worry if its quality “isn’t as good as it might have been.”) Google’s distribution contracts also reduce competition from rivals which further reduces Google’s incentives to invest. Id. (“[H]aving weaker rivals also reduces Google’s concerns over losing customers.”).


1080. For example, Google invests in using massive amounts of user data because the benefits of doing so outweigh the costs. Supra ¶¶ 203–207 (§ III.E.5). But, as Dr. Nayak explained, Google decides how much data to use by engaging in a cost-benefit analysis, weighing the benefit of using more data and increasing quality against the cost of processing the data. Tr. 6337:6–6338:6 (Nayak (Google)) (“[A]s you get more data, it’s more expensive to process. So the cost of processing the data goes up . . . . The time to process it goes up . . . . And so there is this trade-off that we have in terms of amount of data that you use, the diminishing returns of the data, and the cost of processing that data. And so usually, there’s a sweet spot along the way where the value has started diminishing, the costs have gone up, and that’s where you would stop.”).


1081. If Google were not insulated from competition by its search contracts, competition would force Google to improve quality. Tr. 5862:18–5863:13 (Whinston (Pls. Expert)). Benefits that look marginal to a monopolist may suddenly look attractive—or necessary—to a firm facing rivalry because competition changes the costs and benefits. Id.


When Google faces more competition the “whole calculus changes, and now the rational business decision” is to invest more to improve quality. Id. 5862:18–5863:13. “[Competitive markets] force firms . . . to do things that are good for consumers.” Id. 5862:18–5863:13.


1082. Google’s investment in research and development is small compared to other firms in the software and computer industry. Tr. 5857:5–5858:21 (Whinston (Pls. Expert)) (discussing UPXD104 at 79 showing Google's R&D to sales ratio small compared to other software and computer services firms).


In fact, Google spends more on securing defaults than on all other search-related expenses combined, including product launches and improvements. UPX0249 at -556 (showing that Google's overall expenses declined per billion queries, even as its TAC expenses increased); Tr. 1677:14–21 (Roszak (Google)) (discussing UPX0249 at -556); Tr. 7575:3-7577:6 (Raghavan (Google)) (TAC is the largest single expense for Search + and greater than all other expenses combined; explaining UPX7002.A); id. at 7577:7–24 (Google spends much more on TAC than on R&D for search; discussing DXD-21 at .002).


1083. Google has underinvested in latency, crawl rate, index size, and other Search features. Tr. 6397:9–18 (Nayak (Google)) (as of 2020, the size of Google's index had declined, concerning engineers at Google); Des. Tr. 187:10–23, 189:4–13 (Google-PN 30(b)(6) Dep.) (Google does not track latency for other search engines but has done one experiment in the past with Bing, around 2017, which revealed a latency gap of about 300 milliseconds); UPX0249 at -547 ("Since 2014, index size down 48%... crawl rate down 35%, and processing down 33%."); UPX0752 at -017 (“As a consequence of efforts to drive cost efficiency (Search machines spend has declined from 2014 to 2017 despite 45% total increase in queries across that time), we have under-invested in Machines and seen impacts to product quality (smaller index, higher latency, blocked launches, serving errors”).


1084. For example, Google's performance on latency declined between 2011 and 2020, with latency rising from 150 milliseconds to 650 milliseconds in that period. UPX0223 at -122 (showing 500 mx increase in latency from 2011-2020), -086 (Bing beat Google on latency for five common queries); Tr. 7448:4–20 (Raghavan (Google)) (discussing UPX0223 at -122 and agreeing latency grew by a significant amount from 2011 to 2020). Despite this, a 2017 presentation prepared for Mr. Giannandrea, then head of Google Search, by the finance team, including Mr. Roszak, stated, “Budget only spent for ROI positive latency launches.” UPX0249 at -548.


1085. Google has been willing to sacrifice quality for profits. In 2018, Mr. Raghavan had an email exchange with then-VP of Finance, Carlos Kirjner. Mr. Kirjner was pushing Mr. Raghavan for a more aggressive revenue plan for the Ads team. Tr. 7555:22-7556:15 (Raghavan (Google)). Mr. Raghavan told Mr. Kirjner, "I've met enough engineers and PM who want to quit (and many are quitting) because they think we pay lip service to the user experience and squeeze out revenue, while pushing them to hit heroic monetization milestones."


UPX0734 at -509; Tr. 7553:16–7554:8 (Raghavan (Google)) (agreeing that he wrote this statement because he believed it). These engineers and PMs believed that Google's push for monetization was damaging the user experience. Tr. 7554:9-14 (Raghavan (Google)).


He went on to say, "I'm all for cleverer expanded match and auction pricing, but know that the former comes at some cost to users and the latter at a cost to advertisers; both come at a cost to our most critical product talent." UPX0734 at -509; Tr. 7555:4–20 (Raghavan (Google)) (If done poorly, expanded math "could come at a cost to either the user or the advertiser or both," and auction pricing was a reference to auction mechanisms or launches "that do things that are not right for the long term.”). Mr. Raghavan noted that what seemed different at this time was "more than one exiting person has given me the sense that ‘management' has held them to the 20% bar... and they feel pressure to perform... by reaching into user experience impact that they wouldn't have contemplated a couple of years ago." UPX0734 at -509; Tr. 7556:24-7557:8 (Raghavan)) (the 20% was a reference to a 20% OKR); supra ¶¶711-712.


In 2019, Mr. Gomes, then Sr. VP of Search, expressed concern that Search was "getting too close to the money," at the expense of quality-related issues like privacy and innovation. UPX2044 at -998. Mr. Gomes expressed his view that “we are getting too involved with ads for the good of the product and company,” and “[w]e need to think of other issues like DuckDuckGo and the privacy challenge and our innovation narrative.” UPX2044 at -998. Mr. Gomes was “concerned that growth is all we are thinking about.” Id. at -998; Tr. 8129:25–8130:15 (Gomes (Google)).


1086. In 2018, Google considered improving the quality and visibility of “Nav suggestion,” a general search services feature permitting a user searching google.com with a navigational query (e.g., “Facebook”) to go directly to the destination page without stopping at a SERP. The product team explained that the change would benefit users because it would “expedite user journ[ies] . . . increas[ing] the visibility of nav suggestions, and boost[ing] user confidence in them.” UPX0762 at -250; Des. Tr. 219:21–221:1 (Moxley (Google) Dep.).


Emily Moxley, then-Senior Director of Product Management, blocked this feature because she was “hesitant to go this direction,” which would reduce the number of SERPs served and advertising revenue collected, because “we are losing not just people with nav intent, but people that end up interacting with other units on nav queries.” UPX0762 at -249; Des. Tr. 219:12–222:5 (Moxley (Google) Dep.).


1087. When the restrictions imposed by Google’s exclusionary contracts are not in place, Google is forced to compete more vigorously by investing more in search. For example, a choice screen in Europe forced Google to increase investment in search to the benefit of consumers. Tr. 5844:5–5848:14 (Whinston (Pls. Expert)) (“Consumers were getting content that they weren’t getting before. Why? Because it cost Google money to get that content and why do it if you’re not threatened?”); UPX0764 at -077 (identifying feature, content, and trust improvements to Google Search in Europe following the choice screen).


1088. Following a July 2018 European Commission ruling, Google was forced to implement a choice screen in the European Union prompting users to select their default web browser and search app during initial device setup. UPX8091 at -504–05. The choice screen went into effect in early 2020, UPX8091 at -505, and the auction by which GSEs could seek to appear on the choice screen underwent several changes, most recently in September 2021. Tr. 10609:8–22 (Whinston (Pls. Expert)). With the rollout of the European choice screen, Google modeled projected revenue and market share loss. Tr. 8147:8-8148:12 (Gomes (Google)) (discussing UPX0749 at -081).


1089. Google believed that the European choice screen put its “revenue at risk,” UPX0749 at -276, and responded to the increased competition by launching its "Go Big in Europe” investments, which focused on driving daily active users. UPX0749 at -276 (describing Go Big in Europe and noting "choice screen revenue at risk”); Des. Tr. 204:17-205:17, 206:4-8 (Moxley (Google) Dep.) (“Go Big in Europe” focused on driving daily active users due to the potential revenue risk of the implementation of a choice screen in Europe); UPX0761 at -748 (“Go Big in Europe” Google's “response to the coming EU choice screen launch in Jan 2020.”). Google invested millions of dollars and assigned additional full-time engineers to launch new search features in certain European countries. Tr. 8152:12–25, 8154:1-15 (Gomes (Google)).


1090. These Go Big in Europe investments were "above and beyond' business as usual and were meant to “make sure Google is top of mind for EU users.” UPX1083 at -055 (emphasis in original). In response to the European choice screen, Google rolled out search improvements—including (a) “best-in-class or exclusive experiences,” e.g., enhanced sports highlights, and (b) features showcasing local content.


Tr. 8150:13–23; Tr. 8152:5–8 (Gomes (Google)) (discussing UPX0749 at -276); Des. Tr. 206:18–207:20 (Moxley (Google) Dep.) (Google prioritized initiatives to improve the search experience for users after the European choice screen was announced); UPX0764 at -077 (two pillars of Google’s Go Big in Europe investments were unique features and enhanced local content); UPX0765 at -311–12 (listing the highlights of Go Big in Europe product launches).


1091. Even more recently, Google responded to competitive pressure that Microsoft generated by integrating ChatGPT into Bing. Microsoft was “a huge investor” in the company that developed ChatGPT. Tr. 8269:8–16 (Reid (Google)); Tr. 3528:2–3529:2 (Nadella (Microsoft)). ChatGPT was announced to the world around November 2022. Tr. 8269:5–7 (Reid (Google)).


About three months later, Microsoft announced it would integrate ChatGPT’s technology into Bing to give the user a “copilot” experience with Bing search. Id. 8272:4–14. Anticipating Microsoft’s announcement, Google launched Bard, Google’s own chatbot, one day before Microsoft’s announcement. Id. 8272:1–7; id. 8277:8–11 (Google continues to watch what Microsoft does with AI products). Competitive pressure forced Google to modify its business plans.


1092. Finally, the fact that Google responds to regulatory action with “positive innovation-focused moments” belies the testimony of Google executives that Google innovates simply because of company ethos. UPX0792 at -487 (Nov. 2019 email stating, “[w]e’re taking a similar approach to the DOJ suit, landing a series of positive innovation-focused moments leading up to the AG complaint”).

3. Google’s Contracts Reduce Entry And Outside Investment In Search

a) Google’s Contracts Reduce Apple’s Incentives To Participate More Fully In Search


1093. Google recognizes that Apple is a potential entrant into general search. Google executives have “internally had conversations that [Apple] may be possibly building a search engine,” Tr. 7693:12–22 (Pichai (Google)), and understand that Apple has the resources to do so. Tr. 8094:11–8095:5 (Gomes (Google)). In 2018, Apple hired Mr. Giannandrea, who was then the head of Google search. Tr. 2164:18–2165:14 (Giannandrea (Apple)).


As Apple has recruited additional Google search employees, Mr. Pichai asked to be personally notified in “every individual case[]” when this occurs. UPX1092 at -456. In response to one Apple hiring event, Google employees discussed “Apple’s ability and likely plan to build web answers and a general search engine.” UPX0339 at -683. An internal Google memorandum written in 2020 discussed “the risk of Apple (potentially) launching its own search engine, replacing us a default.” UPX0002 at -390; Tr. 8694:24–8695:1 (Israel (Def. Expert)) (Google considers Apple a potential entrant that Google worries about).


1094. Google’s concern is well founded. Apple receives approximately [redacted] billion queries on its devices per week across Safari, Spotlight, and Siri. Tr. 2246:11–2247:9 (Giannandrea (Apple)). Apple “intercept[s]” each query that it receives and decides whether to send it to Google. Id. 2243:25–2244:10, 2281:5–10. By intercepting queries in this manner, Apple can instantaneously decide whether to answer a given query on its own or send the query to a different third party that may be a better fit than Google. Id. 2223:11–16 (“We’re intercepting every query you’re trying to do and trying to decide whether we can help.”); id. 2244:11–18 (Apple has the technical capability to send different queries to different search providers.).


1095. Apple is “serious” about search and has been investing in “[redacted]” needed for a search engine. Tr. 2244:19–2245:18 (Giannandrea (Apple)); UPX0659 at -213 [redacted]. Apple crawls the web to support its search capabilities and has steadily been increasing the number of pages it crawls. Tr. 2207:23–2208:16 (Giannandrea (Apple)). Apple also maintains a web index of about [redacted] billion pages and has been investing in growing the index. Id. 2212:9–23.


1096. Over time, Apple has considered ways to “disrupt Google Search.” UPX0626 at -729 (“[w]hat else would be helpful to discuss how we disrupt Google search?” with “high level themes” of “[a]nswer more queries” and “[m]ake spotlight a good starting point for all searches”); UPX0625 (goals for “Disruptive Search” include “make something better for our customers,” “hurts Google revenue engine,” and “revenue opportunity”).


1097. Upon joining Apple, Mr. Giannandrea led an effort to “[redacted].” Tr. 2228:19–2229:1 (Giannandrea (Apple)) (discussing UPX1123 at -511). Apple devoted approximately [redacted] to this effort and an annual investment of [redacted].” Id. 2227:18–2229:1. With this project—alternatively referred to as [redacted]—Apple sought to build [redacted] UPX1123 at -511.


1098. Apple understands that it has the capability to compete more directly with Google in general search. Tr. 2261:25–2262:2 (Giannandrea (Apple)) (Apple could build a search engine to compete with Google.); id. 2247:14–19 (agreeing that “Apple has the resources, financial and technological, to develop or acquire a general search engine.”); id. 2206:4–6 (Apple has not decided against developing its own GSE.).


1099. But Google’s revenue share payments make Apple far less likely to do so. As explained in a 2020 Google memorandum, Apple entering search “would have two different and separate sources of negative impact on [Apple's] free cash flows and valuation[:] the loss of revenues and the increased costs." UPX0002 at -392.


Apple would not only have to incur the cost of building and maintaining a search engine (which Google valued at $[redacted] B); it would also have to forego Google's revenue share payments under the ISA (which Google valued "in excess of $[redacted] B"). UPX0002 at -392-93; Tr. 5849:9-5850:10 (Whinston (Pls. Expert)) (discussing UPX0002 and Google's explicit acknowledgment that the "forgone payments" would help "dissuade Apple from wanting to come in"); UPX0339 at -683-84 (Apple could use Siri and Spotlight as "beachhead products" for broader entry into search, but "in the near term our deal may be lucrative enough (at least on Safari)" to keep Apple at bay). One way that Apple could enter search would be by acquiring Bing and investing to improve it.


Apple has considered this approach but recognizes that any benefits would have to be weighed against the cost of losing its ISA payments from Google. Tr. 2294:8-15 (Giannandrea (Apple); UPX0460 at -176 (illustrative modeling of Bing acquisition "Relative to Google Deal" under various sets of assumptions); Des. Tr. 182:12-14, 182:16–183:4, 196:2–23, 205:23–211:24, 212:23-215:20, 215:22-25 (Perica (Apple) Dep.) (discussing UPX0460).


1100. The revenue share payments that Apple receives from Google play an important role in Apple's decision-making about search. Tr. 2463:15–18 (Cue (Apple)). For example, in 2016, if Google had not agreed to Apple's minimum economic terms during the ISA negotiations, Apple would have walked away from the agreement. UPX0594 at -852 (Email from Eddy Cue to Tim Cook: “You and I need to decide what is our absolute min is. I told [Mr. Pichai] that him and I need to sit down alone next week and agree to the economic terms or we shouldn't move forward.”); Tr. 2465:8–19 (Cue (Apple)) (if Apple and Google could not agree on economic terms, "there's no deal").


1101. No existing GSE besides Google represents a “valid alternative” that Apple could set as the Safari default. Tr. 2464:8–2465:7 (Cue (Apple)). Consequently, if Apple were unable to reach a revenue share deal with Google, its next best option would likely be to build its own GSE. Id. 2540:15–2542:1 (in the 2016 negotiation, “it wasn’t a choice to pick any of the [other] existing search engines, so we probably would have been left with no other choice than potentially building our own”).


Thus far, Apple has chosen to accept Google’s revenue share payments rather than launch its own GSE. Tr. 8697:13–19, 8698:22–8700:5 (Israel (Def. Expert)) (Apple is making a “make-versus-buy decision about how to handle search,” and the ISA’s terms led Apple to “choose to buy, not make”).


1102. Apple’s participation in search need not be an all-or-nothing proposition. Because Apple reviews each query it receives, supra ¶ 1094, Apple could increasingly choose to divert additional queries away from Google by answering them itself or sending them to a third party, even if Apple did not launch a full-fledged general search service.


Tr. 2220:13–19 (Giannandrea (Apple)) (“every query that we provide an answer to is a query that doesn’t go to Google”); UPX0002 at -390 (Google memorandum explaining that “[l]aunching its own search engine is not the only step Apple could take to shift search traffic away from us. It could replace us as the Safari default by Bing, DuckDuckGo or a larger number of local search engines (eg, Yandex, Navr), totally or partially.


It could replace us for certain types of queries.”); UPX0798 at -905 (Apple email to Microsoft discussing an approach in which Apple would send some queries “to Google to maximize revenue/obey the ads contract” and send other queries to Bing to give Bing “mobile queries to build [its] search engine”).


1103. One way that Apple diverts queries away from Google is with its Suggestions feature in Safari. This feature offers users information and recommendations as they enter text into Safari's URL bar. Tr. 2208:22–2209:4 (Giannandrea (Apple)) (describing Suggestions); id. 2216:25-2218:5 (agreeing that Suggestions “attempt[s] to guess what the query is asking for and to provide the right path to the answer”).


1104. In some Suggestions, Apple provides its own answers directly to the user, and in others, Apple provides links to third-party websites. Tr. 2234:1–2235:4 (Giannandrea (Apple)) (responding to the Court's questions regarding UPXD007 and explaining that “we will either offer straight-up answers or suggestions like this navigation query or suggestions [that you] might want to get something from your device").


1105. Apple believes that the Suggestions feature offers "a much better user experience.” Tr. 2235:6–7 (Giannandrea (Apple)) (Suggestions saves people time and a trip to Google); id. 2219:25-2220:5 (“Our general approach is we think users of our devices are seeking answers, and so if we can provide the answer, we will do that rather than sending them off to a general search engine."); UPX0618 at -265 (“turn[ing] off all displacement of queries and fall[ing] back to Google” would be “worse for users").


1106. But although this feature offers benefits for users, it represents a threat to Google. Suggestions allows users to "bypass Google" and reach their destination directly. Tr. 2218:6–14 (Giannandrea (Apple)). In these cases, Google does not have an opportunity to monetize the user's query. Tr. 5003:3-5004:1 (Braddi (Google)). Thus, when Apple expanded its implementation of Suggestions in 2014, Google analyzed the potential impact to Google and concluded, "Bottom Line: It's bad."


UPX2010 at -527 (estimating that Apple “siphon[ing] queries away" would cause Google to lose 10-15% of its queries on Safari and 4–10% of its revenue from Safari queries). Google has told Apple that “reducing diversions, not showing both a top hit and a suggestion, and elevating the positioning of Google search results relative to Apple results . . . would be worth hundreds of millions of $s annually.” UPX0627 at -256.


1107. During the negotiation that led to the 2016 ISA amendment, Google considered various ways to limit Apple’s use of Suggestions. An internal Google email explained, “we are trying to build a structure that prevents [Apple] from diverting queries and destroying value.” UPX2011 at -001; Tr. 5010:19–5011:10 (Braddi (Google)) (confirming Google’s concern was with Apple diverting queries). In another exchange, Google discussed “listing the current triggering providers that [Apple] could continue to use” for Suggestions because Google was “trying to ensure this does not grow above what it is today.” UPX0965 at -102 (“Also, would be great if we can contain the number of results they present above ours over time (with probably the exception of ‘search history’)”).


In a term sheet that Google sent to Apple, Google included language stating that “Apple will not directly or indirectly take any action or make any omission that adversely impacts the expected economic benefit to Google . . . including, without limitation, by editing, filtering, truncating, appending terms to or otherwise modifying any Search Query originating from [Safari] (e.g., by using Apple’s ‘suggests’ algorithm in connection with Search Queries) or by altering [user interfaces].” UPX2012 at -006–07; Tr. 10026:12–18 (Murphy (Def. Expert)) (discussing UPX2012 and agreeing that “prevent[ing] Apple from shifting queries from Google” was “one of [Google’s] concerns” in these negotiations).


1108. After further discussion, Apple ultimately agreed to Google’s request for the “substantially similar” clause that appears in the 2016 ISA and remains in effect today. Tr. 7663:6–17 (Pichai (Google)) (agreeing Google requested “substantially similar” language.); supra ¶ 227. This clause prohibits Apple from making any product design change (including any change to Suggestions) that would cause Apple’s use of Google in Safari to no longer remain “substantially similar” to its use in 2016. Supra ¶¶ 227, 229.


1109. The “substantially similar” clause limits the extent to which Apple may use Suggestions to divert queries away from Google. In 2018, Ms. Braddi, who was actively involved in negotiating the 2016 ISA, explained “really what the agreement states” to a colleague at Google. UPX0309 at -823.


She wrote: “Up to about 3 yrs ago, [Apple] only referred user[s] to Wikipedia as a suggestion, the rest were provided by Google. However, ~2+ yrs ago we saw them increasingly offer the user other suggested redirections. This concerned us which is why we added into the agmt that they could not expand farther than what they were doing in Sept 2016 (as we did not wish for them to bleed off traffic).” UPX0309 at -823.


Similarly, draft Google prep materials for a December 2018 meeting between Mr. Pichai and Apple CEO Tim Cook explain: “We have been concerned for some time that search traffic cannibalization may be occurring on Apple devices via a few factors, primarily by Safari ‘Suggest’ but possibly also through Siri[.] . . . We even included a provision in the most recent agreement to protect against further cannibalization.” UPX2050 at -652.


1110. The “substantially similar” clause limits Apple’s use of Suggestions in several ways. For example, Apple must “keep their triggering for offering more and more suggestions to the user limited to . . . the trigger rates that they were doing as of September 2016.” Tr. 5017:9– 5018:23 (Braddi (Google)). At trial, Ms. Braddi recanted her deposition testimony on this point, id. 5076:7–11, but her ordinary-course documents confirm her initial testimony and undermine her credibility.


UPX0895 at -904 (“[Apple] has agreed not to implement any changes to the Safari Suggest product that could meaningfully change or divert the Safari traffic they refer to Google.com beginning in Sept. 2016. . . . We wish to go back an[d] categorize the search queries in Sept 2016 and measure the changes since to determine if they are meeting the obligations.”); UPX2014 at -896 (“We need to re-evaluate the changes to the % of non-conversion against the ‘suggest’ pattern that Apple Safari exhibited back in Sept 2016 to identify where they may not have ‘remain substantially similar in use.’


I know Chris took a baseline back in Sept 2016 of how often and in what categories Apple was initiating an ‘Apple Suggest’ to the user for possible redirection. Since this time, we have seen an increase of 5% to the number of queries that do not complete to Google for results. Can we determine the new categories or where the increase is due to them changing the experience away from ‘substantially similar’ to 2016?”).


1111. In addition, the “substantially similar” clause limits the way that Apple may “present” its Suggestions to users. Tr. 5016:25–5017:3 (Braddi (Google)) (responding to the Court’s question and testifying that the “substantially similar” clause applies to the “quality and presentation” of Suggestions). The “substantially similar” clause also limits Apple’s ability to make Suggestions that result in certain queries being sent to third-party websites other than Google.


Tr. 5077:10–5078:15 (Braddi (Google)) (the clause was intended to address Google’s concern that Apple might “sell[] off the traffic to other bidders” such as Amazon or Yelp); Tr. 7663:6–7664:2 (Pichai (Google)) (explaining that the clause was intended “to make sure, as we contemplate a longer term deal, that the notion of default was reasonably preserved in a similar way, particularly with respect to Apple being able to send queries to rival providers”).


1112. Each of these limitations restricts Apple’s flexibility to design its products in ways that may be good for consumers but would result in fewer queries for Google. Tr. 5850:12– 5851:4 (Whinston (Pls. Expert)) (discussing UPXD104 at -073 and Google’s response to the threat of Suggestions).


1113. Another way that Apple could compete more directly with Google is through Spotlight. Spotlight is a “universal search” feature on Apple devices that is primarily used to search for on-device content but can also be used to search for information on the web. Supra ¶ 8.


1114. As Apple has considered ways to "disrupt Google search," it has discussed "[m]aking spotlight a good starting point for all searches." UPX0626 at -729 (emphasis in original). Given Spotlight's potential to affect Google's business, Google has “pa[id] attention" to Apple's choices in this area. Des. Tr. 55:24–15, 57:5–19, 58:10–11, 58:14–59:10 (Fox (Google) Dep.) (“[W]e've been, I would say, paying attention what Apple has been doing with Spotlight. I don't know if I would call Spotlight a general search engine or not, but . . . it is addressing users' information needs, and so that's something that we've been paying attention to").


1115. Apple currently sells Search Ads that allow advertisers to promote iOS applications in response to user queries in the App Store. Tr. 2635:25-2636:1 (Cue (Apple)). Apple does not currently sell ads in response to user queries on Spotlight. Id. 2496:23–24.


1116. Members of Apple's advertising team have expressed a “strong desire” to expand the company's ad offerings to Spotlight, but they have recognized that the ISA would limit their ability to do so. UPX0959 at -177. In 2020, Apple's advertising team prepared to launch a trial in which they would show ads for relevant iOS apps (much like the ads run in the App Store) in response to queries entered into Spotlight and Safari. UPX0959 at -177–78. Before undertaking this trial, they discussed what "constraints" the company would face with respect to such an offering due to "the Google deal." UPX0959 at -178.


Winston Crawford, Apple's Global Senior Director of Business Management for Ad Platforms, noted that the right-of-first-refusal provision of the ISA does not apply to Safari. But with respect to Spotlight, he explained, “testing aside . . . . if we were to commercialize an ad offering in Spotlight we would first need to work with Google. This could open up discussions that lead to a form of renegotiations which I don’t think anyone wants to do right now. So while we have a strong desire to enter Spotlight, it is complex and carries some risk as it relates to the Google contract.” UPX0959 at -177–78.


1117. Google is well aware of the potential for Apple to sell ads on Spotlight. In a 2020 exchange, Google employees opined that although Apple does not currently offer ads in response to queries on Siri or Spotlight, “of course that could change.” UPX0339 at -684 (“One thing to consider is that Apple’s privacy narrative would be counter to advertising on Siri or Spotlight queries, as they have shown a strong position against products built on monetizing users through ads (of course that could change).”).


1118. Under the ISA, if Apple wishes to sell ads on Spotlight, it must give Google the right-of-first-refusal to supply the ads. Supra ¶ 230. If Google does choose to supply the ads, then the revenue share terms that apply to Safari queries would apply to the Spotlight queries as well. JX0033 at -796 (§ 2) (applying “the financial terms set forth in Section 4 of this Agreement” to any ads on Spotlight). In other words, Google would take 60% of the net revenue for the sale of these ads, and Apple would receive only 40%. Id. at -797 (§ 4); Tr. 10021:20– 10022:1 (Murphy (Def. Expert)) (agreeing that serving ads on Spotlight could be less valuable for Apple because they would have to share the money with Google).


1119. In this way, the ISA’s right-of-first-refusal provision reduces the money that Apple would make selling search ads in Spotlight, and accordingly, reduces its incentive to do so. Tr. 6063:1–5 (Whinston (Pls. Expert)) (“I think it keeps Apple potentially out of selling ads.”). This protects Google’s Search Ads monopoly.


b) Google’s Contracts Reduce Distributors’ And Nascent And Potential Competitors’ Incentives To Invest In Novel Search Products


1120. Google’s contracts also reduce distributors’ and nascent and potential competitors’ incentives to invest in innovative search tools. Tr. 5851:5–5852:5 (Whinston (Pls. Expert)) (discussing the effect of Google’s conduct on distributors’ incentive to partner with Branch).


As Dr. Ramaswamy explained, Google’s contracts effectively “freeze the ecosystem in place” to the detriment of nascent and potential competitors whose services could have eroded Google’s monopoly in general search services had they found traction with distributors. Tr. 3796:5–3798:22 (Ramaswamy (Neeva)); Tr. 5851:5–5852:5 (Whinston (Pls. Expert)) (though not a GSE, Branch’s app-search tool facilitated a type of search that could threaten Google’s monopoly).


1121. Google’s RSAs include prohibitions on distributors’ ability to preinstall alternative search services on their devices. Supra ¶¶ 261–263. Although distributors generally understand the RSAs provisions to prohibit preinstallation of services that directly compete against Google Search, like Microsoft’s Bing, the provisions are drafted broadly enough to encompass services like Branch’s app-search tool, even though it is not a GSE. Tr. 5851:5– 5852:5 (Whinston (Pls. Expert)); Des. Tr. 184:5–185:3 (Ezell (AT&T) Dep.) (describing alternative search services as including “typical web search that you would do through a search engine” and listing Bing as an example); id. 240:15–241:4 (Branch is not a substitute for a GSE like Bing).


As a result, Google’s RSAs discouraged OEMs and carriers from preloading Branch’s app-search tool without restrictions on its functionality; Branch has never reached a deal with an OEM or carrier to distribute its app-search tool with all of its intended features. Supra ¶¶ 836–848.


1122. Alarmed by the nascent threat of Branch’s program, Google acted to prevent distribution of Branch’s tools altogether. Those actions included informing AT&T that Google viewed AT&T’s RSA as prohibiting preinstallation of even the limited version of Branch’s appsearch tool if it required a connection to the internet. Supra ¶¶ 849–861. As Mr. Ezell explained, Google’s intervention raised the stakes for AT&T, which ultimately concluded that the risk to its revenue share outweighed the potential economic upside of working with Branch. Supra ¶ 860.


1123. Google’s contracts have also dissuaded Branch from continuing to invest in its own novel search services. Tr. 2948:22–2949:6 (Austin (Branch)).


1124. Venture capital firms view general search as a “no-fly zone.” Tr. 3510:24–3512:7 (Nadella (Microsoft)) (given the barriers to entry and cost to invest in a GSE, Silicon Valley investors consider general search to be the “biggest no-fly zone.”); UPX0240 at -507 (in 2018 email, Giannandrea wrote, “the reason a better search engine has not appeared is that it’s not a VC fundable proposition even though it’s a lucrative business”); Tr. 5848:15–5849:4 (Whinston (Pls. Expert)) (“[S]earch is a place that venture capital does not go.”). Venture capital firms’ hesitancy to invest in search startups creates difficulties for potential entrants; even start-ups with well-funded and talented teams, like Neeva’s, can be forced to shut down if they fail to gain enough traction quickly enough to justify additional investments. Tr. 3674:16–3675:6 (Ramaswamy (Neeva)) (discussing Neeva’s decision to shut down in part due to reduced interest from venture capital firms over time); id. 3723:22–3724:23 (“[I]f a well-funded and exceptionally talented team like Neeva could not even be a provider on most of the browsers, I don’t see that as the market working.”).


1125. Branch has poured hundreds of millions of dollars and countless employee hours into developing an app-search tool, but it now believes that the product has no path to distribution. Tr. 2948:22–2949:6 (Austin (Branch)). At least in part because of the implications for raising funds, it is likely that Branch will stop pursuing its original app-search vision. Id. 2923:1–21 (discussing need to show progress to investors on revenue generation).


If that happens, consumers will not only be deprived an innovative tool for navigating and discovering app content, but also a technology that could have threatened Google’s search monopoly by providing an alternative to web search. Tr. 5851:5–5852:5 (Whinston (Pls. Expert)) (Branch’s app-search tool could provide “a measure of one-stop shopping over apps that would be more like a general search engine.”).


Continue Reading Here.


About HackerNoon Legal PDF Series: We bring you the most important technical and insightful public domain court case filings.


This court case retrieved on April 30, 2024, storage.courtlistener is part of the public domain. The court-created documents are works of the federal government, and under copyright law, are automatically placed in the public domain and may be shared without legal restriction.