From Casetext: Smarter Legal Research

Williams v. Apple, Inc.

UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION
May 28, 2021
Case No. 19-CV-04700-LHK (N.D. Cal. May. 28, 2021)

Opinion

Case No. 19-CV-04700-LHK

05-28-2021

ANDREA M WILLIAMS, et al., Plaintiffs, v. APPLE, INC., Defendant.


ORDER GRANTING IN PART AND DENYING IN PART MOTION FOR CLASS CERTIFICATION

Re: Dkt. No. 76, 77

Named Plaintiffs Andrea Williams and James Stewart bring this putative class action against Defendant Apple, Inc. for breach of contract. Before the Court is Plaintiffs' motion for class certification. ECF No. 76-4 ("Mot."). Having considered the submissions of the parties, the relevant law, and the record in this case, the Court GRANTS IN PART and DENIES IN PART Plaintiffs' motion for class certification.

I. BACKGROUND

A. Factual Background

Apple is a corporation incorporated under the laws of California and has its principal place of business in Cupertino, California. First Amended Complaint ("FAC") ¶ 8, ECF No. 38. Apple's cloud storage service is called iCloud. See id. ¶¶ 2, 13-17. iCloud allows subscribers to "utilize certain Internet services, including storing your personal content (such as contacts, calendars, photos, notes, reminders, documents, app data, and iCloud email) and making it accessible on your compatible devices and computers, and certain location based services." FAC ¶ 53.

According to Plaintiffs, "[o]wners of Apple devices are granted up to 5 GB of iCloud storage for free. If an Apple device user wishes to store more than 5 GB of data on the cloud through iCloud, then that user must subscribe to iCloud's paid service." FAC ¶ 26. Plaintiffs allege that "[i]n order to subscribe to iCloud, a user must agree to the iCloud Terms of Service Agreement." Id. ¶ 28. The relevant portion of the iCloud Terms of Service Agreement ("iCloud Agreement") provides the following:

Apple is the provider of the Service , which permits you to utilize certain Internet services, including storing your personal content (such as contacts, calendars, photos, notes, reminders, documents, app data, and iCloud email) and making it accessible on your compatible devices and computers, and certain location based services, only under the terms and conditions set forth in this Agreement. iCloud is automatically enabled when you are running devices on iOS 9 or later and sign in with your Apple ID during device setup, unless you are upgrading the device and have previously chosen not to enable iCloud. You can disable iCloud in Settings. When iCloud is enabled, your content will be automatically sent to and stored by Apple , so you can later access that content or have content wirelessly pushed to your other iCloud-enabled devices or computers. ¶
Id. ¶ 28 (emphasis in original). This language appears in a September 16, 2015 version of the iCloud Agreement and a September 19, 2019 version of the iCloud Agreement. Id. ¶ 29; FAC, Exs. 1-2.

Named Plaintiffs Andrea Williams and James Stewart sue Apple on behalf of United States iCloud subscribers (excluding Apple, its employees, and its directors) who paid for an Apple iCloud subscription sometime during the class period September 16, 2015 through October 31, 2018 (collectively, "Plaintiffs"). FAC ¶ 44; see Mot. at 1. Williams is a resident and citizen of Florida who, in January 2016, "subscribed to Apple's iCloud service, paid money to Apple for her iCloud subscription, and used iCloud to store her data on the cloud." FAC ¶ 11. Stewart is a resident and citizen of California who, in August 2015, "subscribed to Apple's iCloud service, paid money to Apple for his iCloud subscription, and used iCloud to store his data on the cloud." Id. ¶ 15. Both Williams and Stewart continue to subscribe to iCloud. Id. ¶¶ 11, 15. Moreover, Williams and Stewart "expect to continue [subscribing] for the immediate foreseeable future, given that they have concerns as to the fate of their already stored data if they were to terminate their paying subscriptions." Id. ¶ 50.

The FAC alleges that Apple failed to inform Williams and Stewart that their data was being stored on "non-Apple remote servers and facilities" despite alleged assurances to the contrary. Id. ¶¶ 12-13. Specifically, Plaintiffs allege that they "bargained for, agreed, and paid to have Apple—an entity they trusted—store their data." Id. ¶ 38. According to the FAC, however, Apple's representations were false. "Apple lacked the facilities needed to readily provide the cloud storage space being sold to class members through iCloud." Id. ¶ 33. "Unable to provide the cloud storage space . . . , Apple breached its iCloud agreement with its subscribers and had these users' data stored not by Apple on Apple facilities, but instead turned the users' digital files to other entities, like Amazon and Microsoft[,] for them to store on their facilities." Id. ¶ 34.

Elsewhere, however, the FAC contradicts Plaintiffs' allegation that they "bargained for" this specific provision. Specifically, the iCloud Agreements that Plaintiffs attached to the FAC are form contracts that could not be modified and merely allowed Plaintiffs to click an "AGREE" button. FAC, Ex. 1 at 1; FAC, Ex. 2 at 1. Nowhere do the FAC or the iCloud Agreements allege or establish that the iCloud Agreements could be modified.

The FAC alleges that "[h]ad Apple disclosed that, contrary to its contractual representation, Apple was not the provider of the cloud storage," putative class members "would not have subscribed to Apple's iCloud service or would have not agreed to pay as much as [they] did for the service." Id. ¶¶ 12-13. The FAC claims that other companies, such as Microsoft and Google, offer cheaper cloud storage services than Apple and that Apple's "price premium" harmed putative class members who would have otherwise utilized these cheaper cloud storage alternatives. Id. ¶¶ 40-43.

B. Procedural History

On August 12, 2019, Plaintiffs filed the instant putative class action. ECF No. 1 ("original complaint"). Plaintiffs' original complaint, like the operative FAC, alleged three causes of action against Apple: (1) breach of contract, (2) violations of California's False Advertising Law ("FAL"); and (3) violations of California's Unfair Competition Law ("UCL"). Compare Original Compl. ¶¶ 45-66 (three causes of action), with FAC ¶¶ 51-75 (same). Plaintiffs allege that Apple agreed to be the "provider of the [iCloud] Service" and to store putative class members' content on Apple's servers. FAC ¶¶ 48-50. Yet Apple allegedly breached this promise because "storage was provided by non-Apple third parties with whom neither [Named] Plaintiffs nor class members had bargained." Id. ¶¶ 55-57. Plaintiffs also claimed that Apple violated the FAL and UCL by making the following "false and misleading" claim: "Apple was the provider of the iCloud cloud storage service and [] class members' data would be stored on the cloud by Apple." Id. ¶ 62.

On March 27, 2020, the Court granted in part and denied in part Apple's motion to dismiss the original complaint. ECF No. 34. Specifically, the Court granted with leave to amend Apple's motion to dismiss (1) Plaintiffs' prayer for injunctive relief; and (2) Plaintiffs' FAL and UCL claims. Id. at 26. The Court denied Apple's motion to dismiss Plaintiffs' breach of contract claim. Id.

On April 27, 2020, Plaintiffs filed the operative FAC. ECF No. 38. Apple moved to dismiss the FAC on May 11, 2020. ECF No. 39. On November 17, 2020, the Court granted in part and denied in part Apple's motion to dismiss the FAC. ECF No. 65. Specifically, the Court dismissed Plaintiffs' FAL and UCL claims with prejudice but did not dismiss Plaintiffs' prayer for injunctive relief. Id. at 1. Accordingly, only Plaintiffs' breach of contract claim for damages and injunctive relief remains.

On January 8, 2021, Plaintiffs filed their instant motion for class certification. ECF No. 76-4 ("Mot." or "Motion"). On January 29, 2021, Apple filed its opposition to the instant motion. ECF No. 81-2 ("Opp'n"). On February 13, 2021, Plaintiffs filed their reply supporting class certification. ECF No. 91-2 ("Reply"). On February 22, 2021, Apple filed a sur-reply. ECF No. 98-5 ("Sur-Reply"). On February 24, 2021, Plaintiffs filed a response to Apple's sur-reply. ECF No. 100-3 ("Pls. Sur-Reply").

The key deposition—which is discussed at length in Plaintiffs' reply in support of class certification—took place after Apple filed its opposition to class certification. ECF No. 82-2 at 1 n.1 (discussing deposition of Dane Aconfora, Apple's Director of Service Forecasting and Efficiency). Thus, the Court GRANTS Apple's motion for leave to file sur-reply, ECF No. 99, and considers Apple's Sur-Reply and Plaintiffs' response to the Sur-Reply for purposes of this Order.

II. LEGAL STANDARD

Class actions are governed by Rule 23 of the Federal Rules of Civil Procedure. Rule 23 does not set forth a mere pleading standard. To obtain class certification, Plaintiffs bear the burden of showing that they have affirmatively met each of the four requirements of Rule 23(a) and at least one subsection of Rule 23(b). Zinser v. Accufix Research Inst., Inc., 253 F.3d 1180, 1186, amended by 273 F.3d 1266 (9th Cir. 2001). "A party seeking class certification must affirmatively demonstrate . . . compliance with the Rule[.]" Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 350 (2011). "A district court that 'has doubts about whether the requirements of Rule 23 have been met should refuse certification until they have been met.'" Olean Wholesale Grocery Coop., Inc. v. Bumble Bee Foods LLC, 993 F.3d 774, 793 (9th Cir. 2021) (quoting Brown v. Electrolux Home Prods., Inc., 817 F.3d 1225, 1233-34 (11th Cir. 2016)).

Rule 23(a) provides that a district court may certify a class only if: "(1) the class is so numerous that joinder of all members is impracticable; (2) there are questions of law or fact common to the class; (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class; and (4) the representative parties will fairly and adequately protect the interests of the class." Fed. R. Civ. P. 23(a)(1)-(4). That is, the class must satisfy the requirements of numerosity, commonality, typicality, and adequacy of representation to maintain a class action. Mazza v. Am. Honda Motor Co., Inc., 666 F.3d 581, 588 (9th Cir. 2012).

If all four prerequisites of Rule 23(a) are satisfied, the Court must also find that Plaintiffs "satisfy through evidentiary proof at least one of the three subsections of Rule 23(b). Comcast Corp. v. Behrend, 569 U.S. 27, 33 (2013). Rule 23(b) sets forth three general types of class actions. See Fed. R. Civ. P. 23(b)(1)-(b)(3). As relevant here, Plaintiffs seek certification under Rule 23(b)(3) and 23(b)(2). A class may be certified under Rule 23(b)(3) if a court finds that "questions of law or fact common to class members predominate over any questions affecting only individual members, and that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy." Fed. R. Civ. P. 23(b)(3). A class may be certified under Rule 23(b)(2) if a court finds that "the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive or corresponding declaratory relief is appropriate respecting the class as a whole." Fed. R. Civ. P. 23(b)(2).

"[A] court's class-certification analysis must be 'rigorous' and may 'entail some overlap with the merits of the plaintiff's underlying claim[.]'" Amgen Inc. v. Conn. Ret. Plans & Trust Funds, 568 U.S. 455, 465-66 (2013) (quoting Dukes, 564 U.S. at 351). This "rigorous" analysis applies to both Rule 23(a) and Rule 23(b). Comcast, 569 U.S. at 34. (discussing how Congress included "addition[al] . . . procedural safeguards for (b)(3) class members beyond those provided for (b)(1) or (b)(2) class members (e.g., an opportunity to opt out)," and how a court has a "duty to take a 'close look' at whether common questions predominate over individual ones"). When conducting this rigorous analysis, the Court may be required to "judg[e] the persuasiveness of the evidence presented." Ellis v. Costco Wholesale Corp., 657 F.3d 970, 982 (9th Cir. 2011).

Nevertheless, "Rule 23 grants courts no license to engage in free-ranging merits inquiries at the certification stage." Amgen, 568 U.S. at 466. "Merits questions may be considered to the extent—but only to the extent—that they are relevant to determine whether the Rule 23 prerequisites for class certification are satisfied." Id. If a court concludes that the moving party has met its burden of proof, then the court has broad discretion to certify the class. Zinser, 253 F.3d at 1186.

III. DISCUSSION

Plaintiffs seek certification of a Rule 23(b)(3) class ("Damages Class") and a Rule 23(b)(2) class ("Injunctive Class"). The Damages Class comprises:

All persons in the United States who paid for a subscription to iCloud at any time during the period September 16, 2015 until October 31, 2018. Excluded from this Class definition are all employees, officers, or agents of Defendant Apple Inc. Also
excluded from this Class definition are all judicial officers assigned to this case as well as their staff and immediate families.
Mot. at 1. The Injunctive Class comprises:
All persons meeting the foregoing Rule 23(b)(3) class definition who are current paying subscribers of iCloud in the United States as of the date the Court enters its order granting Plaintiffs' motion for class certification.
Id. For both classes, only one claim is at issue: breach of contract under California law. Mot. at 1. Specifically, Plaintiffs allege that Apple breached the iCloud Agreement (a standardized form contract) by storing Plaintiffs' data on third-party servers rather than Apple's servers. Mot. at 3. In other words, Plaintiffs claim that Apple promised "fully in-house" iCloud storage, not "partly outsourced" storage. Mot. at 23.

Apple opposes Plaintiffs' motion for class certification on six grounds. First, Apple argues that Plaintiffs cannot prove predominance because adjudicating breach of contract would require individualized inquiries into (1) class members' interpretation of the iCloud Agreement; and (2) whether class members' data was in fact outsourced. See Opp'n at 7-8, 11-13. Second, Apple argues that Plaintiffs cannot prove predominance because adjudicating injury would require individualized inquiry into purchases on iCloud family plans. See id. at 9-11. Third, Apple argues that the Named Plaintiffs—Andrea Williams and James Stewart—are atypical and inadequate class representatives. See id. at 24-25. Fourth, Apple argues that the affirmative defenses of waiver and laches defeat predominance. See id. at 14-15. Fifth, Apple argues that Plaintiffs' model of classwide damages fails to satisfy Comcast Corporation v. Behrend, which requires Plaintiffs to "establish[] that damages are capable of measurement on a classwide basis." 569 U.S. 27, 34 (2013); see Opp'n at 15-22 (analyzing damages model). Lastly, Apple argues that Plaintiffs have waived their injunctive class because "Plaintiffs' Motion does not address the standards governing the elements required to certify a [Rule 23](b)(2) class nor apply the record of this case to show how such elements are met." Opp'n at 25.

The Court addresses Apple's six arguments in turn. As detailed below, the Court agrees in part with Apple's first, third, and last arguments. Those arguments identify Plaintiffs' problems with common proof during part of the class period, Williams' inadequacy as a class representative, and Plaintiffs' failure to support certification of the Injunctive Class. The Court disagrees with Apple's other arguments. Ultimately, the Court (1) certifies the Damages Class but limits it to the class period September 16, 2015 to January 31, 2016; (2) holds that Stewart is the sole class representative; and (3) denies certification of the Injunctive Class.

A. Common issues of contractual breach predominate for the class period September 16, 2015 to January 31, 2016.

Apple first argues that Plaintiffs cannot prove Rule 23(b)(3) predominance because adjudicating breach of contract would require two individualized inquiries. The first alleged inquiry would be into each class member's interpretation of the iCloud Agreement. The other alleged inquiry would be into whether each class member's data was in fact outsourced, at least in part, during the class period September 16, 2015 until October 31, 2018. Below, the Court explains why (1) class members' interpretation of the iCloud Agreement is immaterial; but (2) Plaintiffs lack common proof of outsourcing for the portion of the class period after January 31, 2016.

1. Class members' interpretations of the iCloud Agreement do not defeat predominance.

In this breach of contract action, it is undisputed that all putative class members agreed to a standardized iCloud Agreement containing the challenged language at issue. Mot. at 2 (citing APL-ICSTORAGE_00000309 (Agreement revised September 19, 2019); Ex. 4 to Katriel Decl., at APL_ICSTORAGE_00000272 (Agreement revised Sept. 17, 2018); Ex. 5 to Katriel Decl., at APL-ICSTORAGE_00000328 (Agreement revised March 1, 2017); Ex. 6 to Katriel Decl. APL-ICSTORAGE_00000291 (Agreement revised October 20, 2014)). Despite the uniformity of the contract language at issue, Apple argues that "[d]etermining class members' intent requires individualized inquiries." Opp'n at 11. Specifically, Apple argues that "what any individual class member knew about Apple's approach to iCloud storage prior to accepting the [contracts] would inform their interpretation of the disputed provision." Id. at 12.

Apple's argument is meritless. It is an "axiom of contract law[] that when there is a standardized agreement like the form contract at issue in this case, the agreement 'is interpreted wherever reasonable as treating alike all those similarly situated, without regard to their knowledge or understanding of the standard terms of the writing.'" Bally v. State Farm Life Ins. Co., 335 F.R.D. 288, 302 (N.D. Cal. 2020) (emphasis added) (quoting Restatement (Second) of Contracts § 211(2)) (collecting cases). Rather than weigh each class member's subjective understanding, "courts in construing and applying a standardized contract seek to effectuate the reasonable expectations of the average member of the public who accepts it." Restatement (Second) of Contracts ("Restatement") § 211(2) cmt. e. Applying this settled rule to class certification, many courts have rejected the argument Apple makes here. See, e.g., Bally, 335 F.R.D. at 302, 304 (finding predominance and certifying class); Ewert v. eBay, Inc., No. 07-CV-02198-RMW, 2010 WL 4269259, at *7 (N.D. Cal. Oct. 25, 2010) (certifying class and holding that "in construing the form contract . . . , the court need not delve into the actual knowledge of individual class members"); Vaccarino v. Midland Nat. Life Ins. Co., No. 11-CV-5858-CAS, 2013 WL 3200500, at *19 (C.D. Cal. June 17, 2013) ("[E]ven assuming that the contractual provision at issue were ambiguous, the subjective expectations of an insured class member would have little if any bearing on the breach of contract analysis." (quoting Yue v. Conseco Life Ins. Co., 282 F.R.D. 469, 476 (C.D. Cal. 2012))).

In response to this settled authority, Apple offers two unsatisfactory responses. The first is that "this Court need not accept the Restatement as binding." Opp'n at 13. For support, Apple cites a district court case that asserted, without explanation or citation, that "the Restatement is not binding upon th[e] [c]ourt." Id. (citing Monaco v. Bear Stearns Companies, Inc., No. CV 09-05438 SJO JCX, 2012 WL 10006987, at *7 n.4 (C.D. Cal. Dec. 10, 2012)). Yet Monaco did not acknowledge decisions by California Courts of Appeal, which a court sitting in diversity must follow "unless there is convincing evidence that the highest court of the state would decide differently." Daniel v. Ford Motor Co., 806 F.3d 1217, 1222 (9th Cir. 2015) (quoting In re Schwarzkopf, 626 F.3d 1032, 1038 (9th Cir. 2010)). California Courts of Appeal have held—with no evidence that the California Supreme Court would hold differently—that the Restatement "is entitled to great consideration" "in the absence of a contrary statute or decision in this state." Lake Almanor Assocs. L.P. v. Huffman-Broadway Grp., Inc., 178 Cal. App. 4th 1194, 1201 (Ct. App. 2009) (quoting Canfield v. Security-First Nat. Bank, 13 Cal.2d 1, 30-31 (Cal. 1939)). Here, Apple has failed to present a statute or California decision contrary to Restatement § 211. Nor is the Court aware of any such statute or decision. To the contrary, California Courts of Appeal apply Restatement § 211 to the interpretation of "standardized forms of agreement" like the iCloud Agreements here. Ellena v. Dep't of Ins., 230 Cal. App. 4th 198, 213 (Ct. App. 2014) (relying on Restatement § 211). Thus, under California contract law, the iCloud Agreements should be interpreted "without regard to [subscribers'] knowledge or understanding of the standard terms." Bally, 335 F.R.D. at 302 (quoting Restatement (Second) of Contracts § 211(2)).

Apple's other response to Restatement § 211 is that applying it would be unreasonable here. For support, Apple cites another district court case, which stated that "there are times where, even with regard to a form contract, a court will need to inquire into the parties' subjective understanding of an ambiguous term." Opp'n at 13 (quoting Rodman v. Safeway, Inc., No. 11-CV-03003-JST, 2014 WL 988992, at *7 (N.D. Cal. Mar. 10, 2014), aff'd on other grounds, 694 F. App'x 612 (9th Cir. 2017)). Yet Rodman is unpersuasive here in three ways. First, Rodman's only authority for its statement on form contracts was Monaco—which, as discussed above, ignored decisions by California Courts of Appeal. Rodman, 2014 WL 988992, at *7. Second, Rodman's statement about form contracts was dicta. The Rodman Court denied class certification and distinguished Monaco to confirm that denial was appropriate. Id. at *7-8.

Third, subsequent Ninth Circuit precedent confirms that Rodman is inapposite. In Risinger v. SOC LLC, the Ninth Circuit held that where "the evidence needed to resolve the [contractual] ambiguity is common to the class, individual issues will not predominate." 708 F. App'x 304, 306 (9th Cir. 2017) (emphasis added). Applying that holding, the Risinger Court affirmed the certification of a breach of contract class, even though the alleged breach at issue turned on the ambiguous term "customary [duties]." Risinger v. SOC LLC, No. 12-CV-00063-MMD, 2015 WL 13670894, at *6 (D. Nev. Sept. 30, 2015), aff'd, 708 F. App'x at 306. Here, by Apple's own admission, the evidence for resolving any contractual ambiguity comprises "widespread public disclosure[s] by Apple and numerous media platforms." Opp'n at 15 (emphasis added). These "widespread" disclosures "[t]hroughout the Class Period" allegedly informed class members that Apple stored some iCloud data on third-party servers. Id. at 3-4 (collecting disclosures). In short, these "widespread" disclosures are "common to the class." Risinger, 708 F. App'x at 306. Thus, Apple's cited disclosures do not defeat predominance. Id. In sum, individual class members' interpretations of the iCloud Agreement do not defeat predominance.

2. Plaintiffs lack common proof of iCloud outsourcing for the portion of the class period after January 31, 2016.

Apple's main argument against class certification is that "Plaintiffs have not come forward with common proof." Opp'n at 7. Specifically, Apple asserts two failures of common proof. First, Plaintiffs allegedly lack "proof that every member of the class had their iCloud data placed on third-party servers during the Damages Class Period [i.e., September 16, 2015 until October 31, 2018]." Id. (emphasis added). Second, Apple argues that Plaintiffs lack proof that could "determine which U.S. paid iCloud subscribers may have had some data stored historically on third-party servers." Id. (emphasis in original).

Plaintiffs' response to Apple splits the Damages Class Period into two. For the Damages Class Period before February 2016 (September 16, 2015 to January 31, 2016), Plaintiffs argue that all class members had their iCloud data stored on third-party servers. Reply at 6-7; Pls. Sur-Reply at 1-3. As for the rest of the Damages Class Period (February 1, 2016 to October 31, 2018), Plaintiffs argue that there is common proof that (1) shows all class members could have had their iCloud data stored on third-party servers; and (2) identifies which class members had data stored historically on third-party servers. Reply at 7-8; Pls. Sur-Reply at 3-5.

Below, the Court analyzes the two parts of the Damages Class Period. Ultimately, the Court partially agrees with both parties in part. Plaintiffs have presented common proof that, before February 2016, all class members had their iCloud data stored on third-party servers. Yet for the rest of the Damages Class Period, Plaintiffs have failed to present common proof.

a. Plaintiffs present common proof for the Damages Class Period from September 16, 2015 to January 31, 2016.

For the portion of the Damages Class Period before February 1, 2016, Apple argues that "Plaintiffs are essentially asking the Court to ignore the clear and overwhelming evidence that Apple began storing iCloud user data on its own servers in 2015." Sur-Reply at 3. Apple is incorrect. Plaintiffs ask nothing of the sort. Rather, Plaintiffs show that before February 1, 2016, all iCloud data stored on Apple's own servers was also stored on third-party servers. Thus, before February 1, 2016, all class members had their iCloud data stored on third-party servers in alleged breach of the iCloud contract.

Apple's records show this storage on third-party servers in two ways. First, an internal presentation on Apple's first in-house iCloud servers (codenamed "Project McQueen" or "McQueen") discusses the "dual writing" of iCloud data on both McQueen servers and Amazon's Simple Storage Service (Amazon "S3"). Ex. 1 to Pls. Sur-Reply, ECF No. 100-5 (APL_ICSTORAGE_00008181). Specifically, in a "McQueen Update" presentation dated May 1, 2015, a slide titled "Software Roadmap" states that Apple is "[i]nitially dual writing to S3" and running "[f]acilities testing in parallel"—but "targeting turning off dual writes this Q2 FY'16." Id. The second quarter of Apple's 2016 fiscal year spans December 27, 2015 through March 26, 2016. See Securities & Exchange Commission, Form 10-Q for Apple Inc. at 44 (filed Apr. 27, 2016). Thus, as of the date of the May 1, 2015 slide presentation, Apple was not only storing iCloud data on third-party servers, but also Apple planned to continue doing so until sometime between December 27, 2015 and March 26, 2016.

Second, consistent with Apple's roadmap, many internal documents state that Apple's McQueen servers launched in February 2016. Three documents are especially revealing:

• In an undated internal presentation with data through December 2016, two slides bear the title "Project McQueen[.] Internal iCloud Storage Alternative: Launched February 2016." Ex. 2 to Pls. Sur-Reply, ECF No. 100-6 (APL_ICSTORAGE_00004357-58) (emphasis added). One slide shows the percentage of total storage on and daily uploads to third-party servers. The other slide graphs, over time, iCloud storage across five different storage providers—of which "Apple" is only one source. The "Apple" line on the graph does not start growing until after January 2016.
• In an internal presentation dated November 30, 2015 and titled "McQueen Update," Apple "[t]arget[s] go-live in [e]nd of Feb 2016" for in-house storage. Ex. 4 to Pls. Sur-Reply, ECF No. 100-8 (APL-ICSTORAGE_00016596) (emphasis added). Apple thus stated that, until the second quarter of Apple's 2016 fiscal year, McQueen would have zero "End User Demand" and zero "Storage Required." APL-ICSTORAGE_00016604.

• In an internal email dated March 26, 2019, Apple's Senior Director for Finance Bill Van Dyke wrote that "iCloud's launch in 2011 was largely built on [Amazon's] S3 storage and has grown over time. In 2016 we launched our own internal storage solution (McQueen)." Ex. 5 to Pls. Sur-Reply, ECF No. 100-9 (APL-ICSTORAGE_00016133) (emphasis added).
Together, Apple's internal documents show that before February 1, 2016, Apple stored all class members' iCloud data on third-party servers.

In addition, these internal documents are consistent with supposedly contrary evidence that Apple cites in its Sur-Reply. Specifically, Apple cites the deposition testimony of Dane Aconfora (Apple's Director of Service Forecasting and Efficiency) and a declaration by Ahmed Bashir (Apple's Director of Engineering). Sur-Reply at 2. Aconfora testified that McQueen launched in 2015, and "scale[d]-out" in 2016. Id. (quoting Aconfora Dep. at 33:22-25, 37:3-13). Similarly, Bashir declares that "[i]n 2015, Apple launched its own in-house object storage." Bashir Decl. ¶ 5, ECF No. 93.

Yet Aconfora and Bashir conspicuously do not address Apple's documented dual-writing onto third-party servers. Rather, Aconfora and Bashir's statements are consistent with the timeline set forth in the May 1, 2015 Software Roadmap and later documents. As detailed above, Apple "[i]nitially dual wr[ote]" to Amazon S3 and "target[ed] turning off dual writes" between December 27, 2015 and March 26, 2016. APL_ICSTORAGE_00008181. Then "[i]n 2016 [Apple] launched [its] own internal storage solution (McQueen)." APL-ICSTORAGE_00016133. McQueen had zero "End User Demand" until it was "[l]aunched [in] February 2016." APL_ICSTORAGE_00004357-58, 00016604. Thus, Apple is incorrect that there is no common proof of third-party storage before February 1, 2016.

b. However, Plaintiffs fail to present common proof for the Damages Class Period from February 1, 2016 to October 31, 2018.

Apple is correct, however, that Plaintiffs fail to offer common proof of third-party storage for the portion of the class period starting February 1, 2016 (February 1, 2016 to October 31, 2018). Below, the Court first analyzes Apple's arguments against common proof. The Court then explains why Plaintiffs' counterarguments are unpersuasive.

i. Starting February 1, 2016, some putative class members likely had their data stored solely on Apple servers.

For the portion of the Damages Class Period starting February 1, 2016, Apple makes two arguments against common proof of an alleged contractual breach. First, Apple argues that "it is likely that some paid subscribers would have had data stored only on Apple servers" between November 2016 and August 2018. Second, Apple argues that its "iCloud storage records do not and cannot identify all U.S. paid iCloud subscribers whose data was stored using third-party servers." Opp'n at 4 (capitalization altered); accord Sur-Reply at 4 (arguing same).

The Court agrees with Apple. As explained below, Plaintiffs have failed to meet their burden to show by "a preponderance of the evidence" that the issue of third-party storage would predominate across the class. Olean Wholesale Grocery Coop., Inc. v. Bumble Bee Foods LLC, 993 F.3d 774, 784 (9th Cir. 2021) (setting preponderance standard). Nor have Plaintiffs showed that the number of uninjured class members would be "de minimis"—where "5% to 6% constitutes the outer limits of a de minimis number." Id. at 792 (quoting In re Rail Freight Fuel Surcharge Antitrust Litig., 934 F.3d 619, 624-25 (D.C. Cir. 2019)). Rather, given (1) the likelihood that a non-de minimis number of class members had their data stored exclusively on Apple servers and (2) the gaps in Apple's iCloud subscriber records, the putative class would "degenerat[e] into a series of individual trials." Gene & Gene LLC v. Biopay LLC, 541 F.3d 318, 326 (5th Cir. 2008); see also, e.g., In re New Motor Vehicles Canadian Exp. Antitrust Litig., 522 F.3d 6, 20 (1st Cir. 2008) ("Under the predominance inquiry, a district court must formulate some prediction as to how specific issues will play out in order to determine whether common or individual issues predominate in a given case."). The Court addresses both issues in turn.

First, as Apple's Director of Service Forecasting and Efficiency avers, "it is likely that a number of paid U.S. iCloud subscribers could have never had data stored using third-party servers." Aconfora Decl. ¶ 11. Supporting Aconfora's declaration are four internal presentations which show that, between November 2016 and August 2018, "Apple servers stored between 57% and 99% of iCloud storage requests originating in the [United States]." Id. ¶ 6. Moreover, as of approximately March 2019, Apple stored about 40% of all iCloud data on Apple servers. APL-ICSTORAGE_00016133 (Van Dyke email). Altogether, these statistics suggest that a substantial percentage of class members did not have their iCloud data stored on third-party servers. Any non-de minimis percentage of these class members defeats predominance. Olean Wholesale, 993 F.3d at 792 (collecting circuit precedent).

Second, Apple's records cannot isolate those class members whose iCloud data was never stored on third-party servers. Apple's Director of Engineering, Ahmed Bashir, details the limitations of Apple's records in a sworn declaration under penalty of perjury. Bashir avers that Apple keeps two sets of records regarding the location of iCloud data. The first set is a log of "successful upload events (i.e., when an object is initially stored using iCloud or relocated to a different server)." Bashir Decl. ¶ 12. The second set of records is "the current location of all of a user's iCloud objects." Id. ¶ 13. Neither set of records can "identify all U.S. iCloud paid subscribers [1] for whom Apple stored objects on third party servers when such data was uploaded to iCloud or [2] for whom Apple currently stores objects using third party servers." Id. ¶ 14. Nor could Apple's records "determine the total set of U.S. iCloud paid subscribers who had some portion of their iCloud objects stored by Apple using third-party servers" during the Damages Class Period. Id. ¶ 16.

Two gaps in the records account for Apple's inability to identify injured class members. To start, the records "do not allow Apple to track individual objects over time (i.e., they cannot be used to determine what portion of a user's objects may have historically been stored using third party servers)." Id. ¶ 12. Given that iCloud subscriber's data is "constantly being deleted and/or relocated, any snapshot of the current location of an active U.S. user's objects would likely become inaccurate over a period of time." Id. ¶ 13. The other gap in the records is that they "do not include information about the subscriber's location or their paid or unpaid status." Id. ¶ 14. Thus, "it is not feasible to be able to use these sources to identify all users who were U.S. iCloud paid subscribers at the time their data was stored." Id.

In sum, not only is it likely that the putative class contains many members who never suffered the alleged contractual breach, but also it is infeasible to determine who those injured class members are. Accordingly, Plaintiffs fail to show that they could determine—on a classwide basis—which class members in fact had their data stored on third-party servers. See Olean Wholesale, 993 F.3d at 784 (holding that plaintiffs must prove predominance by a preponderance of the evidence).

ii. For the putative class period starting February 1, 2016, Plaintiffs offer unpersuasive arguments for common proof of iCloud outsourcing.

Plaintiffs respond with three arguments, none of which is persuasive. First, Plaintiffs assert that "it is probability-defying for there to have been any U.S. paying iCloud subscribers who never had iCloud content stored on non-Apple servers." Pls. Sur-Reply at 5. Yet the premise of Plaintiffs' assertion is based on incorrect speculation. Plaintiffs speculate that iCloud data is stored randomly and uniformly across five entities' servers (Apple, Amazon, Google, Microsoft, and AT&T) over time, so that the probability that data remains on Apple servers "is akin to rolling a five-numbered die thousands or millions of times and having it land on the same number every time." Reply at 9.

iCloud data was not stored in the way Plaintiffs suggest. Apple's iCloud algorithm intelligently allocated data between in-house and third-party storage "based on the cost and availability of" such storage. Bashir Decl. ¶ 9. Moreover, how the algorithm determined "storage location[s] changed over the course of the [Damages] Class Period." Id. This algorithmic allocation of data explains why, at one point during the Damages Class Period, 99% of all iCloud data was stored on in-house servers. Aconfora Decl. ¶ 6 (discussing November 10, 2016 presentation). This disproportionate allocation of storage to Apple servers would itself be improbable under Plaintiffs' incorrect theory of iCloud storage. All told, without more, Plaintiffs' speculation about storage probabilities fails to prove predominance by "a preponderance of the evidence." Olean Wholesale, 993 F.3d at 784.

Second, citing Bashir's declaration at ¶¶ 12 and 14, Plaintiffs assert that Apple's records identify which paying iCloud subscribers had their data stored on third-party servers. Pls. Sur-Reply at 5. Plaintiffs mischaracterize Bashir's declaration. Specifically, Plaintiffs selectively cite only the first clause of one of Bashir's sentences. The full sentence clarifies that Apple's records do not track the paying subscribers at issue. Bashir avers that although "data sources may be searched by individual subscriber ID, because the sources do not include information about the subscriber's location or their paid or unpaid status, it is not feasible to able [to] use these sources to identify all users who were U.S. iCloud paid subscribers at the time their data was stored." Bashir Decl. ¶ 14 (emphasis added). Similarly, ¶ 12 emphasizes that Apple's records on iCloud uploads "do not identify whether the user associated with the object was a U.S. iCloud paid subscriber at the time of the upload." Id. ¶ 12 (emphasis added).

Lastly, Plaintiffs argue that the mere fact that "every putative class member was exposed to Apple's [storage] algorithm" supports predominance. Reply at 11 (emphasis added). As support, Plaintiffs cite Torres v. Mercer Canyons, Inc., 835 F.3d 1125, 1136 (9th Cir. 2016). Yet Torres is inapposite. In Torres, defendant Mercer allegedly had a common policy of "failing to inform domestic farm workers of the availability of H-2A work that paid $12 per hour," in violation of federal and state worker protection laws. Torres, 835 F.3d at 1130. The Torres plaintiffs moved to certify a Rule 23(b)(3) class of workers who were not informed of the H-2A work. Id. at 1131, 1132. In response, Mercer argued that even if zero workers received legally required information, the presence of some "fortuitous[ly] non-injur[ed]" workers defeated predominance. Id. at 1137. Specifically, Mercer "conten[ded] that those class members, even had Mercer provided them with the omitted H-2A job information, would not have been any better off because they were not looking for work at the time, would not have been ultimately hired by Mercer, or for other individualized reasons." Id.

The Ninth Circuit rejected Mercer's argument on two grounds. To start, Mercer's theory of non-injury merely "highlight[ed] the potential for unlawful conduct in the absence of harm." Id. Moreover, the Torres district court was "well situated to winnow out those non-injured members at the damages phase of the litigation, or to refine the class definition." Id.

Neither ground for the Torres Court's decision applies here. First, the Torres plaintiffs argued—and defendant Mercer accepted for the sake of argument—that every class member was exposed to "unlawful conduct." Id. Mercer merely argued that some class members would not have been damaged by that unlawful conduct. Id. Yet as the Ninth Circuit explained, mere exposure to unlawful conduct was enough to show "informational injury" under the "remedial statute" at issue (the Washington Consumer Protection Act). Id. at 1135. Thus, because Mercer had exposed every class member to unlawful conduct, Mercer had injured practically every class member. This commonality of injury supported predominance. Id. at 1136.

Here, Plaintiffs' only surviving claim—a breach of contract claim—does not allege that mere exposure to Apple's storage algorithm was unlawful. See Williams v. Apple, Inc., No. 19-CV-04700-LHK, 2020 WL 6743911, at *2 (N.D. Cal. Nov. 17, 2020) (order granting in part motion to dismiss). Rather, the allegedly unlawful conduct was Apple in fact storing Plaintiffs' data on third-party servers. Specifically, Plaintiffs allege that "Apple materially breached its iCloud agreement with Plaintiffs and the class members because . . . instead of Apple being the provider of the cloud storage of class members' data, such storage was provided by non-Apple third parties." FAC ¶ 55.

In other words, if Apple were making Mercer's meritless argument, Apple would be starting from different premises. Apple would be arguing that even if all iCloud subscribers had their data stored on third-party servers (the "unlawful conduct" at issue), some iCloud subscribers were uninjured for "individualized reasons" related to those subscribers' circumstances—not Apple's conduct. Torres, 835 F.3d at 1137. Apple in fact avers, however, that "a number of paid U.S. iCloud subscribers could have never had data stored using third-party servers." Aconfora Decl. ¶ 11.

Thus, the instant case is closely analogous to a recent Ninth Circuit decision that distinguished Torres. In Andrews v. Plains All American Pipeline, L.P., 777 F. App'x 889 (9th Cir. 2019), the Ninth Circuit reversed a district court's grant of class certification. The Ninth Circuit reasoned that "[t]he district court relied on inapposite cases in which exposure to the alleged misconduct was itself the injury or was the sole cause of the injury." Id. at 892 (emphasis added) (citing Torres and other cases). Those cases were inapposite because, in Andrews, "causation and injury [were] necessary elements of the class's claims." Id. Given those "key elements," class members may not have "suffered any injury at all" despite their "exposure" to common conduct. Id.

Here too, causation and damages are "essential elements" for breach of contract. Codding v. Pearson Educ., Inc., 842 F. App'x 70, 72 (9th Cir. 2021) (causation); Behnke v. State Farm Gen. Ins. Co., 196 Cal. App. 4th 1443, 1468 (Cal. Ct. App. 2011) (damages). Similarly, Plaintiffs have not shown common proof of causation or injury. Rather, the iCloud algorithm likely kept many putative class members' data on Apple servers only. Aconfora Decl. ¶ 11. Whether the iCloud algorithm stored a particular class member's data on third-party servers (i.e., causation) and harmed that class member (i.e., injury) would require individualized inquiry. Thus, the first ground for the Ninth Circuit's Torres decision does not apply here.

The other ground for the Ninth Circuit's ruling in Torres was that the district court was "well situated to winnow out those non-injured members at the damages phase of the litigation, or to refine the class definition." Torres, 835 F.3d at 1137. Here, by contrast, Apple cannot "winnow out" putative class members whose data was never stored on third-party servers. As detailed above, Apple avers "it is not feasible" to identify who was a paid U.S. subscriber at the time his/her data was stored. Bashir Decl. ¶¶ 12, 14. Apple's iCloud storage records simply "do not include information about the subscriber's location or their paid or unpaid status." Bashir Decl. ¶ 14.

Nor would "refining" the class definition address gaps in common proof here. Torres, 835 F.3d at 1137. In Torres, the class definition could be "refined" because it "fit the theory of legal liability." Id. at 1138. That is, the Torres class definition "track[ed] the group of job-seekers who were subject to, and therefore could be harmed by, Mercer's allegedly unlawful failure to disclose." Id. at 1139. Here, by contrast, Plaintiffs' theory of liability is that Apple stored Plaintiffs' data on third-party servers. FAC ¶ 55. Thus, the class definition must track the group of U.S. paid iCloud subscribers who (1) had their data stored on third-party servers and (2) could have been harmed by such storage. As detailed above, Apple's records are not so precise.

In sum, whether Plaintiffs offer common proof of iCloud outsourcing depends on the portion of the class period at issue. For the period before February 1, 2016 (September 16, 2015 to January 31, 2016), Plaintiffs do have common proof that all class members had their iCloud data stored on third-party servers in alleged breach of the iCloud contract. Yet for the period starting February 1, 2016 onward (February 1, 2016 to October 31, 2018), Plaintiffs fail to provide common proof that would support predominance.

B. Before February 1, 2016, actual injury is classwide despite iCloud's Family Sharing feature.

Apple next asserts that "[d]etermining actual injury requires individualized inquiries." Apple offers three arguments why. Opp'n at 10-11. The first two arguments have already been addressed in the Court's analysis above. Specifically, Apple argues that (1) Plaintiffs lack "common evidence that class members' data was stored using third-party servers (i.e., breach); and (2) class members would have been uninjured if they encountered "widely circulated articles spanning the Class Period that disclosed Apple's practice." Opp'n at 10. As analyzed above, Apple's argument against common evidence does not defeat predominance for the part of the Damages Class Period before February 1, 2016. See Section III-A-2 (analyzing common proof of third-party storage across two parts of the Damages Class Period). As for Apple's reliance on public disclosures of third-party storage, the Court has explained how settled contract law and Ninth Circuit precedent preclude those disclosures from defeating predominance. See Section III-A-1 (analyzing, e.g., Restatement (Second) of Contracts § 211(2) and Risinger v. SOC LLC, 708 F. App'x 304 (9th Cir. 2017)). Thus, only Apple's third argument against the predominance of actual injury remains.

Apple's remaining argument is that iCloud's Family Sharing feature requires individualized inquiries. This argument proceeds in three parts. First, Apple notes that the putative Damages Class only includes subscribers who "paid for a subscription." Opp'n at 11 (emphasis in original). Specifically, the Damages Class comprises:

Family Sharing is a feature that "allows up to six individuals to share an iCloud storage plan." Opp'n at 11 (citing Evan Krasts Decl. ¶¶ 25-27).

All persons in the United States who paid for a subscription to iCloud at any time during the period September 16, 2015 until October 31, 2018. Excluded from this Class definition are all employees, officers, or agents of Defendant Apple Inc. Also excluded from this Class definition are all judicial officers assigned to this case as well as their staff and immediate families.
Mot. at 1 (emphasis added). Second, Apple notes that some iCloud users are on family plans, which take payment from only one family member at a given time. Opp'n at 11. Third, Apple argues that it is infeasible to determine who in a family plan "paid for [the] subscription" and thus is a member of the Damages Class. As an example, Apple cites Named Plaintiff Andrea Williams. Id. Williams shares a family plan with her son and used his credit card to pay for the plan for at least three months. Williams Dep. at 85:22-24, 87:3-88:11. In Apple's view, Williams illustrates how "determining [who] paid for [iCloud]"—and so determining who could be a Damages Class member—"is a genuine concern." Opp'n at 11.

Binding Ninth Circuit precedent forecloses Apple's argument. In Briseno v. ConAgra Foods, Inc., 844 F.3d 1121 (9th Cir. 2017), the Ninth Circuit rejected an argument like Apple's here. In Briseno, plaintiffs moved to certify classes that contained persons "who ha[d] purchased" a certain cooking oil. Id. at 1124. Briseno defendant ConAgra "opposed class certification on the ground that there would be no administratively feasible way to identify members of the proposed classes because consumers would not be able to reliably identify themselves as class members." Id. Specifically, ConAgra argued that it would be infeasible to identify who "purchased" the cooking oil "because consumers do not generally save grocery receipts and are unlikely to remember details about individual purchases of a low-cost product." Id. at 1125. In short, ConAgra argued the proposed classes were unascertainable. Id. at 1124 nn.3-4.

The Ninth Circuit disagreed that ascertainability was an element of class certification. The Ninth Circuit held that an ascertainability requirement would not only contravene the text of Rule 23, but also ignore principles underlying the Rule. See id. at 1124-32 (explaining flaws with ascertainability requirement). Among other things, the Ninth Circuit explained that an "affidavit describing [one's] purchases" in a claims administration process would suffice to show membership in a class of persons "who ha[d] purchased" a product. Id. at 1132. The Ninth Circuit further explained that ConAgra would have "opportunities to individually challenge the claims of absent class members if and when they file claims for damages." Id. at 1131. Thus, the Ninth Circuit affirmed certification of eleven classes of persons "who ha[d] purchased" the product at issue. Id. at 1124.

Here too, Plaintiffs move to certify a class of purchasers: a subset of persons "who paid for a subscription to iCloud." Mot. at 1. Similarly, Apple argues that identifying "those users [who] paid for [iCloud] . . . is a genuine concern." Opp'n at 11. Yet as in Briseno, the administrative feasibility of identifying who paid for the product at issue—and thus could be a member of the Damages Class—is immaterial to class certification here. "[Apple] may prefer to terminate this litigation in one fell swoop at class certification rather than later challenging each individual class member's claim to recovery, but there is no due process right to 'a cost-effective procedure for challenging every individual claim to class membership.'" Briseno, 844 F.3d at 1132 (emphasis in original) (quoting Mullins v. Direct Digital, LLC, 795 F.3d 654, 669 (7th Cir. 2015)). Rather, like in Briseno, even if consumers lack "receipts" of their iCloud purchases, consumers could eventually proffer "affidavit[s] describing [their] purchases" to prove class membership. Id. Apple would also have the "opportunity[y] to challenge the claims of absent class members if and when they file claims for damages." Id. at 1131. In short, the asserted difficulty of identifying "who paid for a subscription to iCloud" is no bar to predominance.

C. Named Plaintiff James Stewart is an adequate and typical class representative, although Named Plaintiff Andrea Williams is not.

Apple argues that Named Plaintiff James Stewart is an atypical class representative and that Named Plaintiff Andrea Williams is an inadequate class representative. See Opp'n at 24-25. The Court analyzes Stewart first, followed by Williams. Ultimately, the Court concludes that Stewart may serve as a class representative, but that Williams cannot.

1. Stewart is not preoccupied with the unique defenses of laches and waiver.

According to Apple, Stewart "is atypical because he is subject to the doctrines of laches and waiver and thus will be preoccupied with his own unique defenses." Opp'n at 24. The Court disagrees. Neither laches nor waiver render Stewart atypical.

"[L]aches is not available as a defense to [Plaintiffs'] claim for breach of contract seeking money damages" under California law. Wyler Summit P'ship v. Turner Broad. Sys., Inc., 235 F.3d 1184, 1193 (9th Cir. 2000); accord, e.g., Heetebry v. Oakander, No. F077652, 2020 WL 5884658, at *5 (Cal. Ct. App. Oct. 2, 2020) ("[L]aches is not a valid defense to an action at law seeking damages for breach of contract."). The reason is that "laches is available as a defense only to claims sounding in equity, not to claims at law." Wyler Summit, 235 F.3d at 1193. "[A] suit to recover damages for [] breach of contract is an action at law[.]" Raedeke v. Gibraltar Sav. & Loan Assn., 10 Cal. 3d 665, 671 (Cal. 1974).

Here, Plaintiffs' only claim against Apple is for breach of contract under California law. See Williams v. Apple, Inc., No. 19-CV-04700-LHK, 2020 WL 6743911, at *11 (N.D. Cal. Nov. 17, 2020) (dismissing other claims with prejudice). Moreover, as explained at the end of this order, money damages is the only remedy that Plaintiffs currently seek. See Section III-F, infra (explaining how Plaintiffs abandoned their injunctive relief class). Thus, laches is unavailable as a defense against Plaintiffs or Stewart.

The defense of waiver is also inapplicable to Stewart. Under California law, a party to a contract waives a contractual right by "indicating an intent to relinquish the right." Wind Dancer Prod. Grp. v. Walt Disney Pictures, 10 Cal. App. 5th 56, 78 (Ct. App. 2017) (quoting DuBeck v. California Physicians' Service, 234 Cal. App. 4th 1254, 1265 (Ct. App. 2015)). Apple argues that Stewart waived his right to in-house storage by agreeing to the iCloud contract despite reading other documents that "disclosed the exact conduct at issue." Opp'n at 12 (citing Stewart Dep. at 95:23-96:2).

Apple fails to meet the high bar for showing waiver at the class certification stage. "The burden [] is on the party claiming a waiver of a right to prove it by clear and convincing evidence that does not leave the matter to speculation, and 'doubtful cases will be decided against a waiver.'" DRG/Beverly Hills, Ltd. v. Chopstix Dim Sum Cafe & Takeout III, Ltd., 30 Cal. App. 4th 54, 60 (Ct. App. 1994) (quoting City of Ukiah v. Fones, 410 P.2d 369, 371 (Cal. 1966)). Moreover, "[w]aiver is ordinarily a question of fact unless 'there are no disputed facts and only one reasonable inference may be drawn.'" Wind Dancer, 10 Cal. App. 5th at 78 (quoting DuBeck, 234 Cal. App. 4th at 1265).

Apple's cited deposition testimony presents disputed facts and more than one reasonable inference. Specifically, Apple asserts that Stewart admitted at his deposition that he read statements that "disclosed the exact conduct at issue"—that is, disclosed that iCloud data could be stored on third-party servers. Opp'n at 12 (citing Stewart Dep. 95:23-96:2). Yet Stewart admitted no such thing. Stewart in fact testified that "I don't recall reading this portion [of the document] with the reference to third-party partners' servers. . . . In those other documents I read, I did not see this." Stewart Dep. at 96:12-19. Thus, the testimony presents a factual dispute as to whether Stewart knew of iCloud's third-party servers. In turn, if Stewart lacked knowledge of iCloud's third-party servers, then he could not have ""intentional[ly] relinquish[ed]" a "known right" to fully in-house iCloud storage. DRG/Beverly Hills, 30 Cal. App. 4th at 60 (quoting City of Ukiah, 410 P.2d at 370).

All told, Apple's weak case for waiver shows that Stewart is not atypical. For Stewart to be atypical, it must be "predictable that a major focus of the litigation will be on an arguable defense unique to [Stewart]." Hanon v. Dataproducts Corp., 976 F.2d 497, 508 (9th Cir. 1992) (emphasis added) (quoting Rolex Employees Retirement Trust v. Mentor Graphics Corp., 136 F.R.D. 658, 664 (D. Or. 1991)). Waiver is unlikely to be a "major focus of the litigation" because, as discussed above, Apple will struggle to prove waiver. Nor is waiver a defense "unique" to Stewart. Other class members presumably could have read the "widespread disclosure[s]" that, in Apple's view, support the waiver defense. Opp'n at 13 n.14. However, as explained above, such disclosures do not defeat predominance. See Section III-D, infra. Thus, Apple's only argument against Stewart serving as a class representative fails. Accordingly, the Court finds that Stewart is a typical and adequate class representative under Rule 23(a).

2. Williams is an inadequate representative because she is the sister-in-law of class counsel.

Apple is correct that Williams is an inadequate class representative, however. Williams "is the sister-in-law of class counsel Azra Mehdi," who is one of two class counsel. Opp'n at 24; accord Reply at 15 (conceding same); Mot. at 26 (listing class counsel). This in-law relationship "create[s] a grave conflict of interest" between Williams and the class she seeks to represent. Eubank v. Pella Corp., 753 F.3d 718, 724 (7th Cir. 2014) (reversing class settlement on this basis, among others). The conflict arises because "where there is a close familial bond between a class counsel and a class representative, . . . there is a clear danger that the representative may have some interests in conflict with the best interests of the class as a whole when making decisions that could have an impact on attorney fees." Petrovic v. Amoco Oil Co., 200 F.3d 1140, 1155 (8th Cir. 1999) (affirming disqualification of class counsel on this basis). "[T]he larger the fee award to class counsel"—and thus the smaller the common fund available to the class—"the better off [the class representative's] . . . in-law would be financially." Eubank, 753 F.3d at 724. In other words, an in-law relationship creates "significant personal and financial ties" that present "an incentive for [Williams] to place the interests of [class counsel] above those of the class." London v. Wal-Mart Stores, Inc., 340 F.3d 1246, 1255 (11th Cir. 2003) (reversing class certification based on personal and business relationship). Thus, Williams is an inadequate class representative.

Plaintiffs' only response to the above weight of authority is a perfunctory citation to one case, Darisse v. Nest Labs, No. 14-CV-01363-BLF, 2016 WL 4385849, at *7 (N.D. Cal. Aug. 15, 2016). Reply at 15. Darisse is unavailing. There, the district court in fact held that the class representative was inadequate. Darisse, 2016 WL 4385849, at *7 (denying class certification). The fact that the class representative was brother-in-law of class counsel was one of several reasons for the Darise Court's finding of inadequacy. Id. The Darise Court merely noted in dicta—and without citation or explanation—that the in-law relationship "on its own . . . would not be persuasive." Id. Yet given the Darise plaintiff's other defects as a class representative (such as "weaknesses of his particular case and [] unique defenses," id.), the Darise Court had no reason to consider the caselaw on how in-law relationships render a class representative inadequate. Thus, here, the Court follows the holdings of the several courts of appeals rather than the dicta in Darise. See, e.g., Eubank, 753 F.3d at 724 (disapproving in-law relationship); Petrovic, 200 F.3d at 1155 (same); London, 340 F.3d at 1255 (disapproving friendship and past business relationship). Williams is an inadequate class representative.

D. Apple's affirmative defenses of laches and waiver fail to defeat predominance.

Relatedly, Apple briefly argues that the affirmative defenses of laches and waiver raise individualized issues that defeat predominance. Opp'n at 14-15. However, as discussed above, laches is inapplicable because Plaintiffs' only remaining claim is a breach of contract claim for damages. See Section III-C-1, supra (analyzing laches).

Nor does waiver defeat predominance. As both the United States Supreme Court and Ninth Circuit have stressed, a court may certify a Rule 23(b)(3) class "even though [] important matters will have to be tried separately, such as some affirmative defenses peculiar to some individual class members.'" True Health Chiropractic, Inc. v. McKesson Corp., 896 F.3d 923, 931 (9th Cir. 2018) (emphasis added) (original alterations omitted) (quoting Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442, 453 (2016)). In True Health, for instance, the Ninth Circuit reversed in part the district court's denial of class certification, which had been based erroneously on an affirmative defense peculiar to some class members. Id. at 932-33. The Ninth Circuit held that unless defendant "ha[d] presented evidence" that an affirmative defense would require non-"'speculati[ve]'" individualized inquiries, the district court lacked discretion to deny class certification based on that affirmative defense. Id. at 932 (quoting Bridging Communities Inc. v. Top Flite Fin. Inc., 843 F.3d 1119, 1125 (6th Cir. 2016)).

Here, Apple's evidence fails to present more than mere "speculation and surmise" that individualized adjudications of the waiver defense would defeat predominance. Id. (quoting same). Indeed, Apple fails to identify even one class member who waived a breach of contract claim. Apple points to Plaintiff James Stewart, but as detailed above, Apple has presently failed to present "clear and convincing evidence" that Stewart waived anything. See Section C-1, supra (quoting DRG/Beverly Hills, Ltd., 30 Cal. App. 4th at 60, and other cases).

The First Circuit rejected a similar argument in Smilow v. Sw. Bell Mobile Sys., Inc., 323 F.3d 32 (1st Cir. 2003). Specifically, the First Circuit held that waiver does not necessitate individualized inquiries where class members all received similar documents from the defendant and all signed "a standard form contract" with an integration clause. Id. at 34 & n.7. Moreover, the First Circuit held that "[e]ven in the unlikely event that individual waiver determinations prove necessary, the proposed class may still satisfy the predominance requirement." Id. at 39. The First Circuit thus reversed the district court, which had decertified a class of consumers who had "signed a standard form contract." Id. at 34; accord, e.g., Kivett v. Flagstar Bank, 333 F.R.D. 500, 506 (N.D. Cal. 2019) (adopting Smilow's analysis in breach of contract class action and certifying class).

Here too, a waiver defense does not defeat predominance as to Plaintiffs' breach of contract claim. Rather, as in Smilow, a key common issue predominates here: all class members "signed a standard form contract" with the same challenged language and an integration clause. Smilow, 323 F.3d at 34; see Ex. 6 to Katriel Decl., at APL-ICSTORAGE_00000307 (integration clause); Section I-A, supra (detailing iCloud Agreement). Accordingly, Apple's affirmative defense of waiver does not defeat predominance.

E. Plaintiffs' damages model satisfies Comcast.

To model classwide damages, Plaintiffs rely on a conjoint survey. The survey extrapolates from (1) survey respondents' relative preferences for six cloud storage attributes to estimate (2) respondents' willingness to pay for "fully in-house" cloud storage. Swain Rep. ¶¶ 32, 72, ECF No. 76-8. Plaintiffs then apply the survey's estimates to Apple's subscription data to yield classwide damages. See Mangum Rep. ¶¶ 52-54.

Apple argues that Plaintiffs' damages model fails to prove predominance under Comcast Corporation v. Behrend. Opp'n at 15-24. In Comcast, the United States Supreme Court held that plaintiffs must provide a damages model which shows that "damages are susceptible of measurement across the entire class for purposes of Rule 23(b)(3)." Comcast, 569 U.S. at 35. If plaintiffs do not offer a plausible damages model that matches their theory of liability, "the problem is not just that the Court will have to look into individual situations to determine the appropriate measure of damages; it is that Plaintiffs have not even told the Court what data it should look for." In re MyFord Touch Consumer Litig., 2016 WL 7734558, at *15 (N.D. Cal. Sept. 14, 2016) (collecting cases). "It is thus necessary for courts to consider 'the degree to which [damages] evidence is reliable in proving or disproving' whether a common question of law or fact predominates over the class members." Olean Wholesale Grocery Coop., Inc. v. Bumble Bee Foods LLC, 993 F.3d 774, 791 (9th Cir. 2021) (emphasis in original) (quoting Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442, 455 (2016)).

Apple argues that Plaintiffs' damages model fails Comcast in two ways. First, Apple argues that Plaintiffs' damages model calculates the wrong measure of contract damages. Second, Apple argues that Plaintiffs' damages model is unreliable. The Court disagrees. Plaintiffs' damages model satisfies Comcast. The Court reaches this conclusion in three steps below. First, for context, the Court summarizes Plaintiffs' damages model. Second, the Court explains how Plaintiffs' measure of contract damages is proper. Third, the Court rebuts Apple's arguments against the reliability of Plaintiffs' damages model.

1. Background on Plaintiffs' damages model: conjoint survey by Dr. Scott Swain and resulting damages estimate by Dr. Russell W. Mangum III

Plaintiffs' proposed method of measuring classwide damages relies on two proffered expert reports. The first and most important report describes a conjoint survey conducted by Dr. Scott D. Swain, Associate Professor of Marketing at Clemson University. Mot. at 22 (describing Swain Report). The second report applies Swain's survey results to Apple's subscription data to estimate class damages. Mot. at 25 (discussing report by Dr. Russell W. Mangum III of Nathan Associates). The Court summarizes each report in turn.

Swain's conjoint survey focused on the product attribute at issue in the instant case: the storage location of iCloud data. In Hadley v. Kellogg Sales Co., this Court explained that, as a general matter, conjoint surveys attempt to determine the value of a product's specific attributes:

The general idea behind conjoint analysis is that the market value for a particular product is driven by features or descriptions of features embodied in that product. Survey respondents are therefore asked to choose between different sets of product attributes, the responses are aggregated, and statistical methods are then used to determine the value (often termed "partworth") that consumers attach to each specific attribute.
324 F. Supp. 3d 1084, 1103 (N.D. Cal. 2018) (internal quotation and citation omitted). Swain's survey tried to estimate the "price premium that class members paid" for iCloud storage that was "fully in-house" versus "partly outsourced." Swain Rep. ¶¶ 29, 45.

To estimate this price premium, Swain asked respondents to choose from a hypothetical set of cloud storage options. The hypothetical storage options differed along six attributes: (1) brands; (2) monthly price; (3) storage location; (4) storage size; (5) family share; and (6) compatible operating systems for multi-device sync. Id. ¶ 63. Swain selected these six attributes after consulting with Plaintiffs' counsel, "conduct[ing] a scan of competing cloud storage services," and interviewing an online focus group. Id. ¶¶ 35-36. Within each of the six attributes, respondents were presented with one possibility (or "level") from a set of two to four possibilities. For the "storage size" attribute, for instance, respondents saw either 50 GB, 200 GB, or 2 TB. For the crucial "storage location" attribute, respondents saw either "fully in-house" or "partly outsourced." The survey's description of the different attributes and levels is reproduced below:

The cloud storage services you will see can differ in terms of their features, Review each feature as you normally would when shopping in real life. Check the boxes to indicate that you've reviewed each feature.

Features

Check

Brands: the names of the different cloud storage services.• Dropbox• Google Drive• iCloud (Apple)• OneDrive (Microsoft)

[ ]

Monthly price: price you pay each month for the cloud storage service.• You'll see monthly prices that range from $0.69 to $12.99.

[ ]

Storage Location: where the brand stores your encrypted cloud content.• Fully in-house: all of your cloud content is stored exclusively on the brand'scomputer servers located at the brand's facilities.• Partly outsourced: the brand diverts some of your cloud content for storageon other companies' servers located at other companies' facilities.

[ ]

Storage Size: amount of cloud storage space you get (GB = Gigabytes, TB =Terabytes).• 50 GB• 200 GB• 2 TB (equals 2000 GB)

[ ]

Family Share: number of users you can add to your cloud storage account.• Only 1 user• Up to 6 users

[ ]

Multi-device Sync: ability to sync multiple devices across different operatingsystems.• Syncs iOS, MacOS, Windows• Syncs iOS, MacOS, Windows, and Android

[ ]

Swain Rep. at 25 (Figure 2).

With these attributes and levels in mind, respondents were then asked to choose from a randomized hypothetical set of cloud storage options. Each option displayed a certain level for each attribute. An example of survey question is reproduced below.

If these were your only options, which would you choose?

Brand:

Google Drive

iCloud(Apple)

Dropbox

OneDrive(Microsoft)

NONE: Iwouldn't chooseany of these.

Multi-device Sync:

Syncs iOS,MacOS,Windows, andAndroid

Syncs iOS,MacOS,Windows

Syncs iOS,MacOS,Windows, andAndroid

Syncs iOS,MacOS,Windows

Storage Location:

Partlyoutsourced:Google divertssome of yourGoogle Drivecontent forstorage on othercompanies'servers locatedat othercompanies'facilities.

Fully in-house:all of youriCloud content isstoredexclusively onApple computerservers locatedat Applefacilities.

Partlyoutsourced:Dropbox divertssome of yourDropboxcontent forstorage on othercompanies'servers locatedat othercompanies'facilities.

Fully in-house:all of yourOneDrivecontent is storedexclusively onMicrosoftcomputerservers locatedat Microsoftfacilities.

Family Share:

Only 1 user

Up to 6 users

Up to 6 users

Only 1 user

Storage Size:

50 GB

2 TB

200 GB

50 GB

Monthly Price:

$0.99

$12.99

$1.99

$0.99

Select

Select

Select

Select

Select

Id. at 26 (Figure 3).

After collecting all survey responses, Swain ran a statistical regression to estimate "the relative utility or consumption satisfaction" (or "partworth") associated with each level of each product attribute. Swain Rep. ¶ 72. Focusing on the partworth for the "storage location" attribute, Swain concluded that consumers' relative willingness to pay for cloud storage decreased substantially if the storage was "partly outsourced" versus "fully in-house." Specifically, Swain concluded that willingness to pay for "partly outsourced" storage is 34.8% lower for iCloud's 50 GB offering; 41.7% lower for the 200 GB offering; and 46.2% lower for the 2 TB offering. Swain Rep. ¶ 93.

Plaintiffs' other proffered expert, Dr. Russell W. Mangum III of Nathan Associates, then multiplied Swain's partworth estimates with Apple's subscription data to yield classwide damages. Mot. at 25. That is, Mangum multiplied iCloud's historical revenue—split across 50 GB, 200 GB, and 2 TB iCloud plans—with Swain's partworth estimates for the "storage location" attribute. Mangum Rep. ¶¶ 52-54. In sum, Plaintiffs' damages model hinges on Swain's estimate of consumer's willingness to pay (percentage "partworth") for "fully in-house" cloud storage over "partly outsourced" storage.

Accordingly, Apple's arguments against Plaintiffs' damages model focus on Swain's survey, and the validity of Mangum's report rises and falls with Swain's survey. See Opp'n at 20 (arguing that Mangum's report is flawed for same reasons that Swain's survey is flawed). Below, the Court addresses Apple's two main sets of arguments against Plaintiffs' damages model. The Court concludes that (1) Plaintiffs' damages model calculates an appropriate measure of contract damages; and (2) Plaintiffs' damages model is sufficiently reliable to prove predominance and thus satisfy Comcast.

2. Plaintiffs' damages model calculates an appropriate measure of contract damages.

Apple first argues that Plaintiffs analyze the wrong measure of damages. Opp'n at 15-16. Specifically, Plaintiffs' damages model calculates "the difference between price paid for a product and value received," or "price premium" for short. Id. at 15. Apple asserts that rather than calculate a price premium, Plaintiffs should have calculated "traditional measures of contract damages." Id. at 16.

Tellingly, though, Apple does not specify the purportedly correct "traditional measures of contract damages." Apple is vague because its argument lacks merit. "The difference between price paid for a product and value received" is in fact the main measure of contract damages. Both the Restatement (Second) of Contracts and California case law make this clear. The Restatement provides that the "measure of damages in general" is the injured party's "loss in the value to him of the other party's performance caused by its failure or deficiency, plus" any other losses. Restatement § 347 (emphasis added); see also Section III-A-1, supra (discussing how the Restatement "is entitled to great consideration" in California contract law). Similarly, California case law holds that a proper measure of contract damages is "the difference between the actual value of what plaintiff has received and that which he expected to receive." Overgaard v. Johnson, 68 Cal. App. 3d 821, 823 (Ct. App. 1977) (emphasis added).

Furthermore, to the extent Apple implies that Plaintiffs' theory of damages has changed, Plaintiffs' theory of damages has been consistent in Plaintiffs' experts' reports and testimony. For example, Stewart testified that he sought damages "[i]n regards to the differences based on what was paid and the reduced level of service." Stewart Dep. at 148:6-7. His "thought of diminished service or the value of that service was the original document [that] stated Apple would do all of these things [i.e., store data in-house]." Id. at 148:10-12. Yet "in separate documents, Apple was not necessarily doing what was stated in the original whole Terms of Service agreement." Id. at 148:13-15. Similarly, in Plaintiffs' damages model, Swain and Mangum estimated the reduced willingness to pay attributable to iCloud storage being "partly outsourced" versus "fully in-house." See Section III-E-1, supra (summarizing Swain and Mangum's damages' model). Thus, all told, Plaintiffs analyze an appropriate measure of contract damages.

3. Plaintiffs' damages model is sufficiently reliable to prove predominance.

Apple's other Comcast argument is that Plaintiffs' damages model is unreliable. Opp'n at 16-22. Specifically, Apple focuses its attacks on Swain's conjoint survey. Yet Apple does not appear to dispute that "conjoint analysis is a well-accepted economic methodology." Hadley v. Kellogg Sales Co., 324 F. Supp. 3d 1084, 1107-09 (N.D. Cal. 2018) (quoting In re Dial Complete Mktg. & Sales Pracs. Litig., 320 F.R.D. 326, 331 (D.N.H. 2017)) (collecting cases holding same). Rather, Apple claims to take issue with the particular design and results of Swain's conjoint survey. Apple asserts that Swain's survey is unreliable in three ways. First, Apple argues that the survey displayed the wrong product attributes and brand comparators. Opp'n at 16-17. Second, Apple argues that the survey improperly excluded supply-side considerations. Id. at 18-20. Third, Apple argues the survey's results are not only allegedly irrational, but also incorrectly assume that "at least some portion of all class members' data was stored on third-party servers during the Damages Period." Opp'n at 22. The Court addresses Apple's three arguments below and concludes that none has merit.

a. Swain selected appropriate attributes and brand comparators.

Apple first argues that "Swain's election of features (or attributes) are not based on features consumers value, but rather are Swain-selected features designed to skew results in Plaintiffs' favor." Opp'n at 16. Specifically, as detailed above, Swain selected six attributes: (1) brands; (2) monthly price; (3) storage location; (4) storage size; (5) family share; and (6) compatible operating systems for multi-device sync. Swain Rep. ¶ 63. In Apple's view, Swain should have also included other attributes "such as automatic sync, comprehensive back up, the ability to restore data, and the seamless integration of iCloud across their Apple devices." Opp'n at 17. Apple further argues that Swain's "brands" attribute was futile. Opp'n at 23. Although Swain presented survey respondents three alternative brands to iCloud—Google Drive, DropBox, and Microsoft OneDrive—Apple asserts that "these products are not comparable . . . Indeed, no other product on the market provides the same range of functionality as iCloud."

Apple's criticism of Swain's attributes fails to defeat predominance. Swain's damages model is still sufficiently "persuasive" and "reliable" to establish predominance. Olean Wholesale, 993 F.3d at 786, 791 (first quoting Ellis v. Costco Wholesale Corp., 657 F.3d 970, 982 (9th Cir. 2011), and then quoting Tyson Foods, 136 S. Ct. at 1046). In fact, in two previous cases, the Court has rejected arguments like Apple's here. In Hadley v. Kellogg Sales Co., 324 F. Supp. 3d 1084, 1108 (N.D. Cal. 2018), defendant Kellogg argued that the "proposed conjoint survey fail[ed] to include other 'critical' attributes 'that consumers consider' when purchasing breakfast products." The Court disagreed. The omitted attributes neither defeated predominance nor made the conjoint survey unreliable. Id. at 1108, 1117. The Court explained, after reviewing the case law, that "it is well-established that these types of critiques merely go to the weight, but not to the admissibility, of survey-based analyses." Id. at 1108 (collecting cases); accord, e.g., In re Dial Complete Mktg. & Sales Pracs. Litig., 320 F.R.D. 326, 337 (D.N.H. 2017) (rejecting argument against attributes and holding that the conjoint survey proves predominance).

In In re Arris Cable Modem Consumer Litig., 327 F.R.D. 334, 368, 372 (N.D. Cal. 2018), defendant Arris argued that the conjoint survey "fail[ed] to properly define the attribute [] at issue" and improperly "fram[ed] [] the survey options" for respondents. The Court disagreed. The conjoint survey was not only "sufficiently reliable from a methodological standpoint," but also consistent with plaintiffs' theory of damages under Comcast. Id. at 370. The Court first explained that "[t]o the extent that [d]efendant asserts that the conjoint survey overstates the magnitude or frequency of [the attribute at issue], that is a merits argument about the proper amount of damages, not a mismatch between [p]laintiffs' damages model and theory of liability." Id. The Court further explained that "critiques of [plaintiffs'] survey design—specifically, the framing of survey options"—did not render the conjoint survey unreliable. Id. at 372 (citing Fortune Dynamic, Inc. v. Victoria's Secret Stores Brand Mgmt., Inc., 618 F.3d 1025, 1038 (9th Cir. 2010)).

Similarly here, Apple argues that Swain's conjoint survey (1) "fail[ed] to include other 'critical' attributes," Hadley, 324 F. Supp. 3d at 1108; and (2) improperly "fram[ed] [] the survey options" for the "brand" attribute. Arris, 327 F.R.D. at 372. As in Hadley and Arris, these arguments are unpersuasive. Apple's arguments fail for three reasons.

First, it is not apparent that Swain failed to include critical attributes. Swain picked his six attributes after reviewing competing cloud storage services and interviewing an online focus group. Swain Rep. ¶¶ 35-36. By estimating the value of "fully in-house" storage and "partly outsourced" storage, these attributes create "a sufficient nexus between Plaintiffs' representative evidence and their [] theory of liability." Olean Wholesale, 993 F.3d at 789. Moreover, after reviewing Apple and Plaintiffs' "battle of the experts," the Court finds that Swain's selection of attributes is sufficiently reliable to prove predominance at class certification. Id. at 793 (quoting Ellis, 657 F.3d at 984).

Second, as Apple admits, omitting attributes from a conjoint survey merely "inflates the importance of" the other attributes, such as the "storage location" attribute at issue. Opp'n at 17 (quoting MacDougall v. Am. Honda Motor Co., No. 17 CV-01079-JGB, 2020 WL 5583534, at *7 (C.D. Cal. Sept. 11, 2020)). "To the extent that [Apple] asserts that the conjoint survey overstates the magnitude or frequency of [third-party storage], that is a merits argument about the proper amount of damages, not a mismatch between Plaintiffs' damages model and theory of liability." Arris, 327 F.R.D. at 370. Thus, Swain's selection of attributes is still sufficiently reliable to prove predominance at class certification. Id. at 793.

Lastly, Swain presented the correct survey options for the "brand" attribute. Apple's own documents identify Google Drive, DropBox, and Microsoft OneDrive as products analogous to iCloud. Ex. 10 to Katriel Reply Decl., at APL_ICSTORAGE_00033711. In sum, Swain's selection of attributes suffices to support predominance.

b. Apple's "supply-side" argument lacks merit, and regardless, Swain considered supply-side factors.

Apple next argues that "Swain's conjoint survey improperly excludes supply-side considerations." Opp'n at 18. Specifically, Apple asserts Swain only accounted for "the value iCloud users supposedly place on their data being stored only using Apple's servers" and not Apple's willingness to supply in-house iCloud storage. Id. In Apple's view, Plaintiffs should have calculated "market price in the but-for world—i.e., the price of [iCloud] without the disputed language." Id. at 20 (emphasis added).

The Court disagrees with Apple on two grounds. The first is conceptual. To measure contract damages, Plaintiffs need not hypothesize Apple's willingness to supply iCloud in a "but-for world" where the iCloud Agreement lacks the disputed language. Indeed, it is unclear how Plaintiffs would model Apple's theoretical supply of a hypothetical product. See Swain Dep. at 150:4-151:1 (discussing the difficulty of determining "what Apple thinks it might have done or imagines that it might do"). Rather, where "defective or partial performance is rendered, the loss in value caused by the breach is equal to the difference between [1] the value that the performance would have had if there had been no breach and [2] the value of such performance as was actually rendered. In principle, this requires a determination of the values of those performances to the injured party himself and not their values to some hypothetical reasonable person or on some market." Restatement (Second) of Contracts § 347 (emphasis added); see also Postal Instant Press, Inc. v. Sealy, 43 Cal. App. 4th 1704, 1709 (Ct. App. 1996) (citing Restatement § 347's definition of contract damages); Section III-A-1, supra (discussing how the Restatement "is entitled to great consideration" in California contract law).

Here, Swain estimates what Restatement § 347 demands. Swain estimates the difference in value between (1) iCloud storage if it were fully in-house and thus not in alleged breach of the iCloud Agreement; and (2) iCloud storage as it was actually provided (i.e., partly outsourced). Swain Rep. ¶¶ 92-93. Furthermore, Swain's estimate considers the value of in-house storage "to the injured [consumer] himself and not [the storage's] value[] . . . on some market." Restatement § 347. Thus, Plaintiffs' damages model estimates contract damages even though it does not hypothesize a but-for world. Accordingly, Plaintiffs satisfy Comcast's requirement that a damages model must "translat[e] the legal theory of the harmful event into an analysis of the economic impact of that event." Comcast, 569 U.S. at 38 (emphasis in original) (quoting Federal Judicial Center, Reference Manual on Scientific Evidence 432 (3d ed. 2011)).

Apple's citation to an unpublished Ninth Circuit opinion is not to the contrary. In Zakaria v. Gerber Prods. Co., 755 F. App'x 623 (9th Cir. 2018), the Ninth Circuit held that it had not been an abuse of discretion to decertify a class. That class sought restitution and actual damages under California's Unfair Competition Law ("UCL"), False Advertising Law ("FAL"), and Consumer Legal Remedies Act ("CLRA")—not contract damages for breach of contract. Id. at 624. Specifically, the Zakaria plaintiffs challenged Gerber's addition of a misleading label to infant formula. Zakaria v. Gerber Prod. Co., No. 15-CV-00200-JAK, 2017 WL 9512587, at *3 (C.D. Cal. Aug. 9, 2017), aff'd, 755 F. App'x 623. As their damages model, the Zakaria plaintiffs proffered a conjoint survey which failed to consider "what had occurred to the actual market price of [the infant formula] with or without the label." Zakaria, 755 F. App'x at 625. Thus, the Ninth Circuit reasoned, "regardless whether consumers were willing to pay a higher price for the labelled product, the expert's opinion did not contain any evidence that such higher price was actually paid; hence, no evidence of restitution or actual damages was proffered." Id.

The instant case differs from Zakaria in at least two important respects. For one, Plaintiffs bring a different claim with a different theory of damages. That is, Plaintiffs allege breach of contract and seek contract damages. Contract damages, as detailed above, measures the difference in value between in-house and partly outsource iCloud storage "to the injured party himself and not . . . on some market." Restatement § 347 (emphasis added). For another, the instant case is not a mislabeling case where Plaintiffs could have measured "what had occurred to the actual market price of [the product] with or without the label." Zakaria, 755 F. App'x at 625. As far as the Court can tell—and Apple does not argue otherwise—the iCloud Agreement did not change in relevant part before the start of the Damages Class Period. Thus, unlike in Zakaria, the parties cannot compare "actual market price[s]" "with or without" the challenged language. All told, the Court disagrees with the conceptual premise of Apple's supply-side argument.

The Court's other ground for disagreeing with Apple's supply-side argument is that, contrary to Apple's assertion, Swain's survey does include some supply-side considerations. Specifically, Swain presented respondents with prices that were tethered to the historical prices of Apple's iCloud subscriptions. Swain Rep. ¶ 32 n.9. For each subscription size (50 GB, 200 GB, and 2 TB), Swain presented three prices: "Apple iCloud's current market price for that storage size" and two other prices that were 30% lower or higher than that market price, respectively. Id. "As a result, the average price for each storage size is equal to the Apple iCloud market price for that storage size." Id. Then, it is undisputed that Mangum multiplied Swain's "storage location" partworth with iCloud historical revenue—which reflect historical prices and quantities by definition—to estimate damages. Mangum Rep. ¶¶ 52-54. Plaintiffs' damages model thus considered "[1] actual market prices that prevailed during the class period; and (2) . . . the actual quantities of products sold during the class period." Hadley, 324 F. Supp. 3d at 1105. As the Court detailed in Hadley, conjoint analyses that consider both factors "adequately account for supply-side factors." Id. (analyzing case law); accord, e.g., In re MyFord Touch Consumer Litig. ("MyFord Touch II"), 291 F. Supp. 3d 936, 969-71 (N.D. Cal. 2018) (holding that conjoint analysis accounted for supply-side factors "by assuming that the supply—the quantity—was fixed").

The Court acknowledges, however, that Swain's consideration of "actual market prices" was not perfect. Apple rightly notes that "[f]or 22 of the 39 months of the Damages Period, the 2TB iCloud Service plan cost $19.99—not the $9.99 price Swain showed respondents." Opp'n at 20. Yet this mistake, without more, does not defeat predominance. Swain still presented the correct historical price data for 44% of the class period as to the 2 TB plan, and 100% of the class period as to the other two iCloud plans. In addition, Mangum used correct historical revenue figures to estimate damages. Thus, on balance, Plaintiffs considered supply-side factors.

c. The results of Plaintiffs' damages model are sufficiently reliable.

Apple's last Comcast argument is that Plaintiffs' damages model yielded flawed results. Apple alleges two flaws. The first purported flaw is that "Swain's conjoint analysis is built on a faulty premise that every consumer had at least some of their iCloud data stored using third-party servers during the class period." Opp'n at 21 (emphasis in original). The other purported flaw is that "Swain's conjoint analysis generates consumer preferences that are inconsistent with rational consumer behavior." Id. at 17.

Neither purported flaw defeats predominance. The first argument assumes certification of a class spanning the entire proposed class period, September 16, 2015 until October 31, 2018. During this time, Apple is correct that some iCloud subscribers' data was stored solely on in-house Apple servers. See Section III-A-2-b, supra (discussing non-de minimis percent of uninjured class members). Yet for the portion of the class period spanning before February 1, 2016 (September 16, 2015 to January 31, 2016), every iCloud subscriber in fact had their iCloud data stored on third-party servers. See Section III-A-2-a, supra (discussing dual-writing on third-party servers). Thus, Apple's first argument does not apply to the class period the Court certifies: September 16, 2015 to January 31, 2016.

Apple's other argument is that 66% of Swain's respondents allegedly "showed at least one irrational preference." Opp'n at 17 (quoting Dr. Lorin Hitt Rep. ¶ 24, ECF No. 81-20). In Apple's view, Swain's "results indicate that many of [Swain]'s survey respondents would prefer to pay a higher price or would prefer the option to sync with fewer (rather than more) operating systems." Id. (citing Hitt. Rep. ¶ 109). This argument is unavailing. Although Swain's results may appear unusual, they do not render Swain's survey unreliable. Three points counsel against overreading these survey results.

First, the results may not be irrational at all. Counterintuitively, some consumers prefer higher prices in certain circumstances. See generally, e.g., Harvey Leibenstein, Bandwagon, Snob, and Veblen Effects in the Theory of Consumers' Demand, 64 Q. J. Econ., 183, 189 (1950) (analyzing the Veblen Effect, by which "the demand for a consumers good is increased because it bears a higher rather than a lower price"). Some consumers may also rationally prefer product simplicity. Where simplicity is valued, superfluous syncing options could have the negative utility found in some of Swain's survey responses.

Second, economically irrational decisions are occasionally human nature. See generally, e.g., Richard H. Thaler, From Homo Economicus to Homo Sapiens, 14 J. Econ. Perspectives 133 (2000) (summarizing various irrational behaviors not modeled by classical economics). For instance, having more choices (such as more syncing options) sometimes "makes us feel worse off than we did before." E.g., Barry Schwartz, More Isn't Always Better, Harv. Bus. Rev. (June 2006). Thus, even if some of Swain's survey responses were irrational, that is to be expected even in an accurate survey.

Lastly, Apple's cited cases underscore that Swain's survey results are relatively reliable. In MacDougall, for instance, the court criticized conjoint survey results that were not just "economically irrational," but absurd. McDougall, 2020 WL 5583534, at *8. There, the survey measured willingness to pay for a non-defective transmission in a Honda Odyssey minivan. Id. The survey calculated that one respondent would pay "three million dollars more for a vehicle with a defect." Id. (emphasis added). For another respondent, the survey calculated that the respondent would pay "ten million dollars more for a vehicle without the alleged transmission defect." Id. (emphasis added). In comparison, the retail price for a new Honda Odyssey is about $29,550. Id. n.6. Understandably, the McDougall Court thus found that "the irrationality exhibited in individual survey responses evidences a deeply flawed conjoint study that produced unreliable results unreflective of actual [willingness to pay]." Id.

Analogously, in In re Volkswagen, the conjoint survey results had two obvious flaws. To start, the survey "results suggest[ed] that certain consumers value a $2,000 navigation system in a $16,000 vehicle at $9,000, and that VW is a premium brand compared to Audi." In re Volkswagen "Clean Diesel" Mktg., Sales Pracs., & Prod. Liab. Litig., No. 3:17-CV-4372-CRB, 2020 WL 6688912, at *8 (N.D. Cal. Nov. 12, 2020). Moreover, the survey results stated that the magnitude of consumers' "overpayment for low emissions" turned "radically" on just the price of any given Volkswagen vehicle. Id. at *2, *8. Specifically, consumers' overpayment supposedly "rang[ed] from 8.5% to 60.5%" based on the price of the car. Id. The In re Volkswagen Court concluded that this damages range "that resembles 'somewhere between almost nothing and almost everything' is facially unrealistic." Id. at *8.

By contrast, Swain's results are more reliable. They do not ascribe millions of dollars in value to the product attribute at issue, as in MacDougall. Nor do Swain's results present false brand associations or damages ranges "between almost nothing and almost everything." Id. at *8. Rather, Swain's results present certain idiosyncrasies or irrationalities that track some economic theories. Thus, Apple's last attack on Plaintiffs' damages model also fails. Plaintiffs' damages model "establish[es] that damages are capable of measurement on a classwide basis." Comcast, 569 U.S. at 34.

In sum, after analyzing all of Apple's arguments against the Damages Class, the Court concludes that some of Apple's arguments have merit. Even so, a temporal subset of the Damages Class—with Stewart but not Williams as class representative—satisfies the requirements of Rule 23(a) and Rule 23(b)(3). Specifically, the Court certifies the Damages Class with a narrower class period of September 16, 2015 to January 31, 2016. Accordingly, the Court GRANTS IN PART and DENIES IN PART Plaintiffs' motion to certify the Damages Class.

F. Plaintiffs fail to explain why an injunctive relief class should be certified.

Finally, without citing a single case on injunctive relief or offering any legal analysis, Plaintiffs conclusorily seek to certify the Injunctive Class under Rule 23(b)(2). Mot. at 1. But see, e.g., Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 360, 365 (2011) (discussing requirements of Rule 23(b)(2) certification). Apple thus correctly notes that "Plaintiffs' Motion does not address the standards governing the elements required to certify a (b)(2) class nor apply the record of this case to show how such elements are met." Opp'n at 25. Given Plaintiffs' failure to meaningfully analyze the injunctive class in their motion for class certification ("Motion"), the injunctive class has been waived. E.g., Grace v. Apple, Inc., 328 F.R.D. 320, 350 (N.D. Cal. 2018) (holding same); Davidson v. Apple, Inc., No. 16-CV-04942-LHK, 2018 WL 2325426, at *25 (N.D. Cal. May 8, 2018) (holding same); Diacakis v. Comcast Corp., No. 11-CV-3002-SBA, 2013 WL 1878921, at *9 (N.D. Cal. May 3, 2013) (holding same).

Indeed, in their reply brief, Plaintiffs' identify only three fragments of their Motion that "address[ed] their injunctive relief class claims." Reply at 15 (citing Mot. at 1:15-17, 6:22-7:8, and 9:13-17). None addresses the law governing Rule 23(b)(2) certification. The first fragment is merely the class definition in the Notice of Motion. Mot. at 1. The second fragment is a block-quote from Williams's deposition. In that block-quote—which the Court reproduces in its entirety—Williams vaguely describes the proposed injunctive changes to iCloud's Terms and Conditions as "I don't know how -- you know, just -- or make it so that it stands out":

Q. So are you asking for the terms of the iCloud Terms and Conditions to be changed?

A. Probably putting things at the forefront, you know, more. I mean, I don't know how -- you know, just -- or make it so that it stands out.

Q. Are you asking the Court to have Apple change its practices with respect to iCloud in any way?

Q. So are you asking for the terms of the iCloud Terms and Conditions to be changed?

A. Probably putting things at the forefront, you know, more. I mean, I don't know how -- you know, just -- or make it so that it stands out.

Q. Are you asking the Court to have Apple change its practices with respect to iCloud in any way? [No response provided in Plaintiffs' Motion]
Mot. at 7-8 (quoting Williams Dep. Tr. 120:14-121:18). Lastly, Plaintiffs identify a fragment of their Motion that quotes Stewart's deposition. Mot. at 9.

None of the above "approach[es] a legal argument" that develops Plaintiffs' claim for injunctive relief. Grace, 328 F.R.D. at 350; see also, e.g., John-Charles v. California, 646 F.3d 1243, 1247 n.4 (9th Cir. 2011) (holding party "failed to develop any argument on this front, and thus has waived it"). Plaintiffs thus leave many key issues unaddressed. For instance, Plaintiffs fail to analyze whether their proposed changes to iCloud's Terms and Conditions—assuming arguendo that Plaintiffs have proposed sufficiently precise changes—are "appropriate respecting the class as a whole." Dukes, 564 U.S. at 365 (emphasis in original). Nor do Plaintiffs analyze whether their Named Plaintiffs are adequate and typical claimants of injunctive relief, even though adequacy and typicality "take[] on special resonance and ha[ve] generated significant precedent" "[b]ecause Rule 23(b)(2) classes are non-opt-out classes." 2 Newberg on Class Actions § 4:27 (5th ed. Dec. 2020 update); see also Section III-C-2 (holding that Williams in an inadequate class representative). By contrast, Apple argues at length that Stewart faces the atypical defense of laches. Compare Opp'n at 14-15, 24 (analyzing laches), with Reply at 6 (asserting, without citation, that "laches does not affect an injunctive class"). Laches is a defense to injunctive relief. See, e.g., Inst. of Cetacean Rsch. v. Sea Shepherd Conservation Soc., 725 F.3d 940, 947 (9th Cir. 2013) ("[T]raditional equitable considerations such as laches, duress and unclean hands may militate against issuing an injunction . . . .").

In short, Plaintiffs' request for injunctive relief is "not well taken" because it "is devoid of any meaningful analysis." Grace, 328 F.R.D. at 350 (quoting Diacakis, 2013 WL 1878921, at *9). Accordingly, the Court DENIES Plaintiffs' motion to certify an injunctive class under Rule 23(b)(2).

IV. CONCLUSION

For the foregoing reasons, the Court GRANTS IN PART and DENIES IN PART Plaintiffs' motion for class certification. The Court certifies the following class:

All persons in the United States who paid for a subscription to iCloud at any time during the period September 16, 2015 to January 31, 2016. Excluded from this Class definition are all employees, officers, or agents of Defendant Apple Inc. Also excluded from this Class definition are all judicial officers assigned to this case as well as their staff and immediate families.
The Court denies certification for the putative class period spanning February 1, 2016 to October 31, 2018 and denies Plaintiffs' motion to certify an injunctive class under Rule 23(b)(2). Lastly, the Court orders that Named Plaintiff James Stewart shall serve as sole class representative. Named Plaintiff Andrea Williams is an inadequate class representative.

IT IS SO ORDERED.

Dated: May 28, 2021

/s/_________

LUCY H. KOH

United States District Judge


Summaries of

Williams v. Apple, Inc.

UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION
May 28, 2021
Case No. 19-CV-04700-LHK (N.D. Cal. May. 28, 2021)
Case details for

Williams v. Apple, Inc.

Case Details

Full title:ANDREA M WILLIAMS, et al., Plaintiffs, v. APPLE, INC., Defendant.

Court:UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA SAN JOSE DIVISION

Date published: May 28, 2021

Citations

Case No. 19-CV-04700-LHK (N.D. Cal. May. 28, 2021)

Citing Cases

True Health Chiropractic Inc. v. McKesson Corp.

See Andrews v. Plains All American Pipeline, L.P., 777 Fed.Appx. 889, 892 (9th Cir. 2019) (“These…

Brown v. Google LLC

. Additionally, “the contract language must be assessed objectively, from the perspective of a reasonable…