AMENDED1 ORDER DENYING MOTION FOR PRELIMINARY APPROVAL OF CLASS ACTION SETTLEMENT

Re: Dkt. No. 330

LUCY H. KOH, District Judge.

Plaintiffs Kimberly Heines, Hashmatullah Essar, Paul Dugas, Matthew Ridolfo, Deana Ridolfo, Yaniv Rivlin, Mali Granot, Brian Neff, and Andrew Mortensen (collectively, "Plaintiffs") bring a putative class action against Defendant Yahoo! Inc. ("Yahoo"). Plaintiff Brian Neff also brings a putative class action against Defendant Aabaco Small Business, LLC ("Aabaco") (collectively with Yahoo, "Defendants"). Before the Court is Plaintiffs' motion for preliminary approval of class action settlement. ECF No. 330 ("Mot."). Having considered the parties' motion and supplemental filings, arguments of counsel at the November 29, 2018 hearing, the relevant law, and the record in this Case, the Court DENIES Plaintiffs' motion for preliminary approval of class action settlement.

I. BACKGROUND

A. Factual Background

The instant lawsuit involves three data breaches that occurred from 2013 to 2016. According to Plaintiffs, Defendants did not use appropriate safeguards to protect users' personal identification information ("PII"), and Plaintiffs' PII was thus exposed to hackers who infiltrated Defendants' systems. Specifically, Plaintiffs allege three separate data breaches: a breach that occurred in 2013, a breach that occurred in 2014, and a breach that occurred in 2015-2016 (collectively, "data breaches"). ECF No. 196, ¶ 161. Moreover, Plaintiffs allege that Yahoo "made a conscious and deliberate decision not to alert any of Yahoo's customers that their PII had been stolen." Id., ¶ 9.

On July 23, 2016, before Yahoo publicly disclosed the 2013, 2014, or 2015-2016 data breaches, Yahoo and Verizon entered a stock purchase agreement under which Verizon would acquire Yahoo's operating business. Yahoo, Stock Purchase Agreement (July 23, 2016).

Plaintiffs allege that in August 2016 a hacker posted information from 200 million Yahoo accounts for sale on the dark web. ECF No. 196, ¶ 122. Plaintiffs allege that at least as early as August 2016, multiple copies of Yahoo's database were sold by hackers in Eastern Europe. Id., ¶ 123. At the motion for preliminary approval hearing, Plaintiffs stated that the entire user database was available for sale on the dark web in 2016, and that Yahoo purchased it using bitcoin. ECF No. 351 at 46. Yahoo acknowledged that Yahoo had purchased the data from the dark web, but disputed that the data had been sold multiple times. Id. at 46-47.

On September 9, 2016, in an SEC filing regarding the Verizon purchase agreement, Yahoo represented that Yahoo knew of no incidents of unauthorized access of personal data that might adversely affect the acquisition. Yahoo, Preliminary Proxy Statement (Schedule 14(A), at Exhibit A-18 (Sept. 9, 2016), available at https://www.altaba.com/static-files/ad5f11da-0a78-4f3e-90f8-dd204c1978fb.

Thirteen days later, on September 22, 2016, Yahoo publicly disclosed the 2014 data breach. ECF No. 196, ¶ 126. In the announcement, Yahoo claimed that it learned of the 2014 data breach during a "recent investigation." Yahoo Security Notice September 22, 2016, available at https://help.yahoo.com/kb/%20SLN28092.html. However, Plaintiffs allege that Yahoo's information security team and legal department both knew about the 2014 data breach as it was happening. ECF No. 196, ¶ 104. As discussed below, Yahoo itself conceded in its 10-K filing with the SEC about six months later, on March 1, 2017, that Yahoo had contemporaneous knowledge of the 2014 data breach. ECF No. 196, ¶ 129; Yahoo, 2016 Annual Report (Form 10-K), at 47 (Mar. 1, 2017).

On November 9, 2016, Yahoo disclosed the 2015-2016 data breach in an SEC filing. ECF No. 196, ¶ 139. Plaintiffs do not appear to allege when Yahoo first knew about the 2015-2016 data breach, but Plaintiffs allege that hackers used data obtained during the 2014 data breach to gain access to specific user accounts in the 2015-2016 data breach. Id., ¶ 119.

On December 14, 2016, Yahoo publicly disclosed the 2013 data breach and stated that this data breach affected more than one billion user accounts. Id., ¶ 133. Plaintiffs' state that "Yahoo claims to have been totally unaware of this breach until being notified by law enforcement in 2016. . . . Yahoo [] claim[s] that in 2013 its data security measures and breach detection measures were so poor that hackers were able to access every single Yahoo account . . . and exfiltrate users' PII, and Yahoo never detected it." Id., ¶ 11, 12. Plaintiffs allege that Yahoo "had drafted notifications pertaining to the [2015-2016] breach as early as December 15, 2016, but delayed in sending them to affected users." Id., ¶ 141. Yahoo began publicly notifying users of the 2015-2016 data breach in February 2017. Id., ¶ 142.

As a result of the data breach disclosures, Plaintiffs allege that Verizon demanded a $925 million discount on the purchase price of Yahoo. Id., ¶ 138. On February 21, 2017, Verizon and Yahoo announced that they had agreed on a $350 million price reduction and an adjustment regarding the parties' respective shares of liability and litigation costs. Press Release, Verizon, Verizon and Yahoo Amend Terms of Definitive Agreement (Feb. 21, 2017), available at https://www.prnewswire.com/news-releases/verizon-and-yahoo-amend-terms-of-definitive-agreement-300410420.html.

Although Yahoo had publicly claimed on September 22, 2016 that Yahoo had learned of the 2014 data breach during a "recent investigation," Yahoo Security Notice September 22, 2016, available at https://help.yahoo.com/kb/%20SLN28092.html, Yahoo admitted on March 1, 2017 in its 10-K filing with the SEC that Yahoo had "contemporaneous knowledge" of the 2014 data breach. ECF No. 196, ¶ 129; Yahoo, 2016 Annual Report (Form 10-K), at 47 (Mar. 1, 2017).

Although on December 14, 2016, Yahoo had claimed that the 2013 data breach affected only one billion user accounts, Yahoo disclosed on October 3, 2017 that the 2013 data breach affected approximately 3 billion user accounts—every user account then existing. Id., ¶ 145-46.

In addition to the instant class action lawsuit, Yahoo has settled several other actions arising out of the data breaches. On April 24, 2018, Yahoo settled for $35 million claims brought against it by the SEC alleging that Yahoo misled investors by failing to disclose the data breaches. Press Release, SEC, Altaba, Formerly Known as Yahoo!, Charged with Failing to Disclose Massive Cybersecurity Breach; Agrees to Pay $35 Million (April 24, 2018), available at https://www.sec.gov/news/press-release/2018-71.

On September 7, 2018, this Court granted final approval to Yahoo's $80 million settlement of a federal securities class action related to Yahoo's failure to disclose the 2013, 2014, and 2015-2016 data breaches. In re Yahoo! Inc. Securities Litigation, No. 17-CV-00373-LHK, ECF No. 118.

On January 9, 2019, the Superior Court of California for the County of Santa Clara granted final approval to Yahoo's $29 million settlement of shareholder derivative class actions. See Case No. 17-CV-00787-LHK, ECF No. 41-2. This settlement pertained to pending shareholder derivative class actions in California State Court, Delaware State Court, and this Court. See Case No. 17-CV-00787-LHK, ECF No. 41.

In addition to the 2013, 2014, and 2015-2016 data breaches discussed above, Plaintiffs' expert Mary Frantz submitted a 92-page report regarding Yahoo's data security. ECF No. 252-17. The report shows repeated failures to follow industry-standard security practices, extensive knowledge of ongoing security breaches beginning in 2008 with failure to adequately respond, failure to provide adequate staffing and training, and failure to comply with industry standard regulations. The report shows that the first full-scale breach of the user database was the 2013 breach. However, several incidents prior to 2013 appear to have involved several million accounts, as well as access to Yahoo's internal systems. At a minimum, the report concludes that Yahoo was on notice of significant security lapses several years prior to the 2013 data breach. Nonetheless, at the motion for preliminary approval hearing Yahoo stated that it continues to deny any data breach prior to 2013, and it has not notified users of any data breach prior to 2013. ECF No. 351 at 18-19.

B. Procedural History

On September 22, 2016, the same day that Yahoo first disclosed any of the data breaches that are the subject of this case, the first federal actions that became part of the instant Multidistrict Litigation ("MDL") were filed. See, e.g., Complaint, Havron v. Yahoo, No. 16-cv-01075 (S.D. Ill. Sept. 22, 2016), assigned Case No. 16-cv-07031-LHK. On December 7, 2016, the Judicial Panel on Multidistrict Litigation ("JPML") issued a transfer order selecting the undersigned judge as the transferee court for the instant MDL action. See ECF No. 1 at 1. Additional lawsuits were filed and related or transferred to the undersigned judge. ECF Nos. 6, 7, 9, 30, 33, 40, 64.

On April 12, 2017, Plaintiffs filed a Consolidated Class Action Complaint ("CAC") covering all three data breaches. ECF No. 80. In the CAC, Plaintiffs asserted claims on behalf of Yahoo users in the United States, Israel, Venezuela, Australia, and Spain. ECF No. 80 at 5-12. On May 22, 2017, Defendants filed a motion to dismiss the CAC. ECF No. 94. On August 30, 2017, the Court granted in part and denied in part the motion to dismiss the CAC. ECF No. 132. The Court dismissed the Australia, Venezuela, and Spain Plaintiffs. ECF No. 132 at 88.

After this Court had issued its ruling on the motion to dismiss the CAC, Yahoo disclosed that the 2013 data breach affected an additional two billion Yahoo user accounts. As a result, the Court gave Plaintiffs time to conduct discovery and to amend their complaint. ECF No. 147. The Court ordered Yahoo to "expedite its production of discovery regarding the recent data breach disclosure." ECF No. 142. Plaintiffs filed the First Amended Complaint ("FAC") on December 15, 2017. ECF No. 174, 196.

On January 19, 2018, Defendants filed the motion to dismiss the FAC. ECF No. 205. On March 9, 2018, the Court granted in part and denied in part the motion to dismiss the FAC. ECF No. 215.

On July 13, 2018, Plaintiffs filed a motion for class certification. ECF No. 246. On September 1, 2018, Defendants filed their opposition to the motion for class certification. ECF No. 295.

Separate from the instant case, parallel proceedings against Yahoo have been ongoing in California state court. Yahoo! Inc. Private Information Disclosure Cases, Judicial Council Coordination Proceeding No. 4895 ("JCCP case"). Seven class action lawsuits were filed against Yahoo in California state court. ECF No. 330-17, ¶ 14. As with the federal cases in the instant MDL, the first of the California state lawsuits was filed on September 22, 2016. Id., ¶ 14. On February 15, 2017, Judge Thierry Patrick Colaw of the Superior Court of California, County of Orange, granted the petition for coordination of the seven California lawsuits and recommended such coordination to the Judicial Council of California. ECF No. 330-19 at 2. On February 28, 2017, the Judicial Council of California authorized the Presiding Judge of the Orange County Superior Court to assign the lawsuits to a coordination trial judge. ECF No. 330-17, ¶ 16. On March 14, 2017, the Presiding Judge of the Orange County Superior Court assigned the JCCP case to Judge Colaw. ECF No. 330-20 at 2.

On June 23, 2017, Judge Colaw denied Yahoo's motion to stay the JCCP case. ECF No. 330-22. On June 27, 2017, Plaintiffs in the JCCP case filed a consolidated class action complaint. ECF No. 330-23. On December 13, 2017, Judge Colaw granted in part and denied in part Yahoo's demurrer. ECF No. 330-24. On August 27, 2018, Plaintiffs in the JCCP case filed a motion for class certification. ECF No. 330-17, ¶ 30.

The complaint in the JCCP case alleged violations of California's Consumers Legal Remedies Act, Unfair Competition Law, Customer Records Act, common law claims for negligence and breach of contract, and an invasion of privacy claim under the California Constitution and common law. ECF No. 330-23. Except for the invasion of privacy claim, all the California state law claims in the JCCP complaint were also asserted in the instant MDL case.

Plaintiffs in the instant MDL case and in the JCCP case jointly engaged in settlement discussions with Yahoo. ECF No. 330-17, ¶ 31. As part of the settlement negotiations, the parties agreed to seek approval of the class action settlement before the undersigned judge. ECF No. 330-17, ¶ 32. On September 14, 2018, the parties informed this Court that they had reached a settlement in principle and requested that the Court stay the matter in its entirety. ECF No. 322. On September 17, 2018, the Court denied the requested stay. ECF No. 323. On September 18, 2018, the Court stayed the deadlines related to the motion for class certification. ECF No. 325.

Following Judge Colaw's retirement in January 2018, the JCCP case was re-assigned to Judge Glenda Sanders. ECF No. 330-17 at 5, n.1. On September 19, 2018, during a status conference in the JCCP case, Judge Sanders approved the parties' proposed settlement process and set a further status conference on March 19, 2019, in order to give the Court in the instant case time to reach a final disposition on the proposed settlement. ECF No. 330-25. The parties in the JCCP case agreed that they anticipated requesting dismissal of the JCCP case if this Court approved the settlement. Id. at 4.

On October 22, 2018, Plaintiffs in the instant case filed a motion for preliminary approval of class action settlement. ECF No. 330. The settlement agreement applies to this case as well as to the JCCP case. ECF No. 330-3 at 1. On November 2, 2018, this Court ordered the parties to provide supplemental information to support the Motion for Preliminary Approval. ECF No. 333. On November 5, 2018, the Court ordered the parties to provide additional supplemental information. ECF No. 335. The parties filed their supplemental statement on November 9, 2018. ECF No. 337. The Court held a hearing on November 29, 2018.

II. LEGAL STANDARD

Federal Rule of Civil Procedure 23(e) provides that "[t]he claims, issues, or defenses of a certified class may be settled . . . only with the court's approval." Fed. R. Civ. P. 23(e). "The purpose of Rule 23(e) is to protect the unnamed members of the class from unjust or unfair settlements affecting their rights." In re Syncor ERISA Litig., 516 F.3d 1095, 1100 (9th Cir. 2008). Accordingly, in order to approve a class action settlement under Rule 23, a district court must conclude that the settlement is "fundamentally fair, adequate, and reasonable." Hanlon v. Chrysler Corp., 150 F.3d 1011, 1026 (9th Cir. 1998). Preliminary approval of a settlement and notice to the class is appropriate if "[1] the proposed settlement appears to be the product of serious, informed, non-collusive negotiations, [2] has no obvious deficiencies, [3] does not improperly grant preferential treatment to class representatives or segments of the class, and [4] falls within the range of possible approval." In re Tableware Antitrust Litig., 484 F.Supp.2d 1078, 1079 (N.D. Cal. 2007) (citing Manual for Complex Litigation (Second) § 30.44); see also In re High-Tech Emp. Antitrust Litig., No. 11-CV-02509-LHK, 2014 WL 3917126, at *3 (N.D. Cal. Aug. 8, 2014).

III. DISCUSSION

The Court denies the motion for preliminary approval of class action settlement on several bases. First, the settlement inadequately discloses the release of claims related to any unauthorized access of data in 2012. Second, the release of the 2012 claims is improper. Third, the proposed notice inadequately discloses the size of the settlement fund. Fourth, the settlement appears likely to result in an improper reverter of attorneys' fees. Fifth, the settlement inadequately discloses the scope of non-monetary relief. Sixth, the settlement inadequately discloses the size of the settlement class. Any of these bases would be sufficient to deny the motion for preliminary approval. The Court concludes its discussion with a comparison of this settlement with the settlement in In re Anthem, Inc. Data Breach Litig., 327 F.R.D. 299 (2018).

A. Legal Standard Regarding Inadequate Disclosures

Due process requires adequate notice before the claims of absent class members are released. In re Online DVD-Rental Antitrust Litig., 779 F.3d 934, 946 (9th Cir. 2015). Notice must be written in "plain, easily understood language" and "generally describe[] the terms of the settlement in sufficient detail to alert those with adverse viewpoints to investigate and to come forward and be heard." Fed. R. Civ. P. 23(c)(2)(B); Churchill Vill., L.L.C. v. Gen. Elec., 361 F.3d 566, 575 (9th Cir. 2004) (citation omitted). Notice is inadequate if it misleads potential class members. Molski v. Gleich, 318 F.3d 937, 952 (9th Cir. 2003), overruled on other grounds by Dukes v. Wal-Mart Stores, Inc., 603 F.3d 571 (9th Cir.2010).

"A district court's approval of a class-action settlement must be accompanied by a finding that the settlement is `fair, reasonable, and adequate.'" Lane v. Facebook, Inc., 696 F.3d 811, 818 (9th Cir. 2012) (quoting Fed. R. Civ. P. 23(e)). The Ninth Circuit has listed several factors that the district court should consider in determining whether a settlement is fair, reasonable, and adequate, including inter alia, the strength of the plaintiffs' case; the risk of maintaining class action status throughout the trial; and the amount offered in settlement. See Hanlon, 150 F.3d at 1026. "[S]ettlement approval that takes place prior to formal class certification requires a higher standard of fairness." Id.

B. Release of 2012 Claims

1. Inadequate Disclosures

The proposed notice to class members fails to provide reasonable notice that the settlement agreement releases claims arising from any unauthorized access of data in 2012. See In re Online DVD-Rental Antitrust Litig., 779 F.3d at 946 (explaining that due process requires notice prior to releasing claims of absent class members). The proposed notice explains that the settlement relates only to the 2013, 2014, and 2015-2016 data breaches. ECF No. 330-8 at 8. The proposed notice fails to disclose that claims related to any unauthorized access of data in 2012 are also being released even though the settlement releases such claims. Id.

The Court appreciates the parties' effort to provide relief for any unauthorized access of data in 2012. Such relief appears appropriate in light of Plaintiffs' expert Mary Frantz's 92-page report regarding Yahoo's data security. ECF No. 252-17. The report shows repeated failures to follow industry-standard security practices, extensive knowledge of ongoing security breaches beginning in 2008 with failure to adequately respond, failure to provide adequate staffing and training, and failure to comply with industry standard regulations. Id. The report shows that the first full-scale breach of the user database was the 2013 breach. Id. However, several incidents prior to 2013 appear to have involved several million accounts, as well as access to Yahoo's internal systems. Id.

Although providing relief is appropriate, it must be done correctly. The parties must provide sufficient information for the Court to review the settlement and for class members to make informed decisions as to their participation in the settlement based on any unauthorized access of data in 2012. The current record is devoid of such information. Plaintiffs did not allege any claims related to any data breaches prior to 2013 in the Consolidated Class Action Complaint ("CAC") and the First Amended Complaint ("FAC") even though the CAC and FAC alleged inadequate security measures as early as 2008. ECF No. 252-17, CAC or the FAC. Moreover, the CAC, FAC, settlement agreement, and motion for preliminary approval do not state what happened with Yahoo users' data in 2012 or identify any harm to any group of 2012 Yahoo users. Yahoo has never disclosed any such harm to its users and continues to deny any data breach prior to 2013. ECF No. 351 at 18-19. Accordingly, the Court and class members have no basis to evaluate the 2012 claims and their release.

Moreover, the parties do not identify how many additional users would be included in the settlement class that were not part of the class as defined in the FAC. The class as defined in the FAC included all persons in the U.S. and Israel who registered for free Yahoo accounts, paid Yahoo accounts, or Yahoo Small Business or Abacao accounts "whose PII was accessed, compromised, or stolen from Yahoo in the 2013 Breach, the 2014 Breach, or the [2015-2016] Breach." ECF No. 338-1 at 1.

On the other hand, the settlement class includes "[a]ll U.S. and Israel residents and small businesses with Yahoo accounts at any time during the period of January 1, 2012 through December 31, 2016, inclusive." ECF No. 338-1 at 2. The parties acknowledge that the settlement class is broader than the proposed class in the FAC. ECF No. 338-1 at 4. Thus, the settlement releases the claims of 2012 users who may not have been part of the later data breaches. The viability of any claims on behalf of 2012 users goes to both the strength of plaintiffs' case and the risk of maintaining class action status throughout the trial. See Hanlon, 150 F.3d at 1026. Accordingly, the parties' inadequate disclosures prevent the class members and the Court from assessing the reasonableness of the settlement. See In re Online DVD-Rental Antitrust Litig., 779 F.3d at 946; Lane, 696 F.3d at 818.

2. Improper Release of Claims

Independent of the inadequate disclosures regarding the release of 2012 claims, the Court concludes that the release of these claims conflicts with Ninth Circuit precedent, which only allows release of claims "where the released claim[s] [are] based on the identical factual predicate as that underlying the claims in the settled class action." Hesse v. Sprint Corp., 598 F.3d 581, 590 (9th Cir. 2010) (internal quotation marks omitted); see also Chavez v. PVH Corp., 2015 WL 581382, at *6 (N.D. Cal. Feb. 11, 2015) (where "the release in a settlement agreement does not limit released claims to those arising out of the facts alleged in the complaint, denial of . . . approval [to] the settlement is appropriate."). Specifically, the settlement releases claims on behalf of all users in 2012, but the FAC does not assert claims based on any incidents prior to the 2013 data breach. Accordingly, the Court concludes that the settlement releases claims that are not "based on the identical factual predicate as that underlying the claims in the settled class action." Hesse, 598 F.3d at 590. Accordingly, any future settlement must amend the FAC.

C. Failure to Adequately Disclose Size of the Settlement Fund

The proposed notice fails to disclose the total size of the settlement fund. As a result, class members cannot assess the reasonableness of the settlement. See Churchill Vill., L.L.C., 361 F.3d at 575 (notice must "describe[] the terms of the settlement in sufficient detail to alert those with adverse viewpoints to investigate and to come forward and be heard." (citation omitted)).

The proposed notice discloses $50 million to cover out-of-pocket costs, alternative compensation, paid user costs, and small business user costs. ECF No. 330-8 at 17. In addition, the proposed notice discloses that class counsel may apply for attorneys' fees of up to $35 million, costs and expenses of up to $2.5 million, and service awards of up to $7,500 each for settlement class representatives, to be paid separately from the settlement fund. The proposed notice does not disclose the costs of credit monitoring services or costs for class notice and settlement administration, and does not disclose the total size of the settlement fund. Without knowing the total size of the settlement fund, class members cannot assess the reasonableness of the settlement.

Similarly, the settlement agreement and the motion for preliminary approval also fail to identify the size of the settlement fund. The parties' lack of disclosure also inhibits the Court's ability to assess the reasonableness of the settlement.

In addition, without knowing the total size of the settlement fund, class members cannot reasonably assess the attorneys' fees award. The proposed notice provides inadequate information for class members to assess the potential attorneys' fees award as a percent of the total settlement fund. Moreover, as discussed in more detail below, the Court finds that the settlement may result in an improper reverter of attorneys' fees. This potential reverter would reduce the total amount that Yahoo would have to pay as a result of the settlement. As a result, the potential reverter further prevents class members and the Court from assessing the reasonableness of the settlement. See Churchill Vill., L.L.C., 361 F.3d at 575; Lane, 696 F.3d at 818.

D. Reverter of Attorneys' Fees

The Court concludes that the settlement may allow for unreasonably high attorneys' fees, and therefore any unawarded attorneys' fees may improperly revert to Defendants. See In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d 935, 949 (9th Cir. 2011). The proposed settlement authorizes up to $35 million in attorneys' fees, "to be paid separate and apart from the Settlement Fund." ECF No. 330-3 at 23. Because attorneys' fees do not come from the Settlement Fund, any amount not awarded by the Court would effectively revert to Defendants rather than to the benefit of the class.

In In re Bluetooth Headset Prods. Liab. Litig., the Ninth Circuit held that a similar attorneys' fees arrangement was not prohibited, but that the district court "was required to examine the negotiation process with even greater scrutiny than is ordinarily demanded, and approval of the settlement had to be supported by a clear explanation of why the disproportionate fee is justified and does not betray the class's interests." 654 F.3d at 949. The court explained that unless the attorneys' fees arrangement was "in the class' best interest as part of the settlement package," the "funds should have been negotiated to revert to the class rather than to the putative wrongdoer[s]." 654 F.3d at 949 (internal quotation and citation omitted). The Ninth Circuit explained that this sort of fee arrangement is particularly problematic where the district court concludes that "the settlement authorizes unreasonably high attorneys' fees." Id.

As discussed below, the Court relies on both the lodestar method and the percentage-of-recovery method to conclude that the settlement may authorize "unreasonably high attorneys' fees." Id.; see also In re Online DVD-Rental Antitrust Litig., 779 F.3d at 944-45 ("One way that a court may demonstrate that its use of a particular method or the amount awarded is reasonable is by conducting a cross-check using the other method.").

1. Lodestar Method

Class counsel provided a lodestar figure of $22 million for 38,278.81 hours of work. ECF No. 337-6. By the Court's count, Plaintiffs' lodestar covers 143 attorneys from 32 firms. Id. This Court authorized only five law firms to work on the instant MDL case. ECF No. 58. On February 1, 2018, the Court ordered "[o]ther that the Plaintiffs' Executive Committee, no other law firms shall work on this MDL without prior approval of the Court. Motions for approval of additional Plaintiffs' counsel shall identify the additional Plaintiffs' counsel and their background, the specific proposed tasks, and why Plaintiffs' Executive Committee cannot perform these tasks." ECF No. 208. On May 7, 2018, the Court granted class counsel's request to allow one attorney from four law firms to attend and help prepare their respective clients, who are Plaintiffs in the instant MDL case, for these Plaintiffs' depositions by Defendants. ECF No. 233. On June 20, 2018, the Court granted class counsel's request to substitute an attorney who was approved on May 7, 2018 with that attorney's colleague. ECF No. 241. The Court also allowed one additional attorney to assist an attorney approved on May 7, 2018. Id. Class counsel has not made any further motions for approval of additional counsel in the instant MDL case. However, the parties seek attorneys' fees for 24 firms in the instant MDL case and 8 firms in the JCCP case.

The scope of class counsels' work in the instant case was substantially limited by the parties' agreement that California law governed and by the small number of counts in the complaint, many with overlapping elements. See ECF No. 196, ¶ 175 (stipulation agreeing that "California common law and statutory law applies to all claims by members of the United States and Paid Users Classes"), ¶ 179 (Plaintiffs' contention that California law should apply to all claims for all classes). The Consolidated Class Action Complaint ("CAC") only included thirteen counts. ECF No. 80. On behalf of the United States and Israel Classes, the CAC alleged two California statutory violations (Unfair Competition Law and Consumer Legal Remedies Act). On behalf of the United States, Israel, and Small Business Classes, the CAC alleged two California statutory violations (Data Breach Notification Law and Online Privacy Protection Act), one federal statutory violation (Stored Communications Act), and three common law causes of action (Breach of Contract, Breach of Implied Contracts, and Breach of Implied Covenant of Good Faith and Fair Dealing). On behalf of the Small Business Class, the CAC alleged one California Statutory violation (Unfair Competition Law) and two common law causes of action (Fraudulent Inducement and Negligent Misrepresentation). On behalf of users from Australia, Venezuela, and Spain, the CAC alleged one common law cause of action (Negligence). On behalf of all classes, the CAC alleged one claim for Declaratory Relief under 28 U.S.C. § 2201.

The First Amended Complaint ("FAC") also included thirteen counts, which substantially overlapped with the CAC. ECF No. 196. The FAC added one count for Deceit by Concealment under Cal. Civil Code §§ 1709, 1710; two counts for violations of the California Customer Records Act; and alleged separate violations of the California Unfair Competition Law for Unlawful Business Practice and Unfair Business Practice. Unlike the CAC, the FAC did not allege violations of the Data Breach Notification Law, Online Privacy Protection Act, Stored Communications Act, or allege Fraudulent Inducement.

In addition to filing two complaints, class counsel filed oppositions to two motions to dismiss. ECF Nos. 117, 211. Finally, class counsel filed a motion for class certification, ECF No. 325, and a motion for preliminary approval of class action settlement. ECF No. 330. Class counsel also prepared four expert reports, took seven Yahoo depositions, and reviewed 9 million pages of discovery.

As Plaintiffs argued, their opposition to the second motion to dismiss substantially overlapped with their opposition to the first motion to dismiss. ECF No. 211 at 2 ("Now that Defendants can no longer challenge the sufficiency of [Plaintiffs'] claims head on, they argue in their Motion to Dismiss that the Court should grant them backdoor exits."). For example, Plaintiffs explained that Defendants "argue again that their services are not really services" under the Consumers Legal Remedies Act and California common law; "argue[] that money users paid to Yahoo for supposedly secure services does not grant standing under the Unfair Competition Law when this Court already held that it does;" and argue again "that Plaintiffs' PII is not covered under the California Customer Records Act—this time under Cal. Civ. Code § 1798.81.5." Id. at 3 (citations omitted). The Court agrees with Plaintiffs' description of the opposition to the second motion to dismiss. Specifically, the Court finds that both motions to dismiss addressed questions related to standing under the Unfair Competition Law, adequacy of claims under the Consumer Legal Remedies Act, adequacy of claims under the Customer Records Act, adequacy of California breach of contract claims, adequacy of negligence claims, application of the economic loss rule to negligence claims, and adequacy of declaratory judgment claims.

Moreover, the Court concludes that the legal theories involved were not particularly novel. In Adobe and Anthem, this Court previously addressed several of the legal theories that Plaintiffs relied on in the instant case. See In re Adobe Sys., Inc. Privacy Litig. ("Adobe"), 66 F.Supp.3d 1197 (N.D. Cal. 2014); In re Anthem, Inc. Data Breach Litig. ("Anthem I"), 162 F.Supp.3d 953 (N.D. Cal. 2016); In re Anthem, Inc. Data Breach Litig., ("Anthem II"), 2016 WL 3029783 (N.D. Cal. May 17, 2016). Specifically, the Court finds that these precedents addressed key issues in the current case, including: (1) whether risk of future identify theft and loss of value of PII constitute injury in fact to support Article III standing, see Adobe, 66 F. Supp. 3d at 1214-15; Anthem II, 2016 WL 3029783, at *14; (2) whether Plaintiffs can adequately state a causal connection between a data breach and alleged harms, see Anthem I, 162 F. Supp. 3d at 987; (3) whether benefit-of-the-bargain losses related to the value of reasonable data security are sufficient to support standing under the California Unfair Competition Law ("UCL"), Adobe, 66 F. Supp. 3d at 1224; Anthem II, 2016 WL 3029783, at *30; (4) whether Plaintiffs can adequately allege a claim under the UCL for unlawful conduct where Plaintiffs allege an underlying California Customer Records Act ("CRA") violation, Adobe, 66 F. Supp. 3d at 1226; (5) whether Plaintiffs can adequately allege a claim under the UCL for unfair conduct under the balancing test where Plaintiffs allege that Defendant failed to adequately protect customer data, Adobe, 66 F. Supp. 3d at 1227; Anthem I, 162 F. Supp. 3d at 990; and (6) whether contract language regarding "inherent limitations" in privacy safeguards "relieve[s] [Defendant] of the responsibility . . . to provide `reasonable' security," see Adobe, 66 F. Supp. 3d at 1221. In addition, other courts in this district had addressed whether software could constitute a "service" under the California Consumers Legal Remedy Act when the software provider "continually updates" and supports the software. See Haskins v. Symantec Corp., No. 13-CV-01834-JST, 2013 WL 6234610, at *9 n.9 (N.D. Cal. Dec. 2, 2013).

As to the discovery conducted by Plaintiffs, Plaintiffs took only 7 percipient witness and Rule 30(b)(6) depositions. Mot. at 1; ECF No. 351 at 29. Defendants took more depositions than Plaintiffs, including depositions of Plaintiffs' four experts. Plaintiffs never deposed Defendants' experts. Id. During Case Management Conferences, the Court had to encourage class counsel to actively litigate the case and take discovery. See, e.g., ECF No. 155 at 26 (instructing class counsel "You need to get this discovery so that you can amend this complaint in time.").

Class counsel in the JCCP case filed a complaint, opposed a motion to stay, opposed a demurrer, and filed a motion for class certification. ECF No. 330-17 at 3. However, as previously discussed, except for the invasion of privacy claim, the JCCP case asserted the same California state law claims as the instant MDL case. During case management conferences in the instant case, the parties explained to this Court that they were coordinating discovery with plaintiffs in the JCCP case. ECF No. 91 at 28; No. 108 at 6; No. 111 at 41. In their declarations supporting the motion for preliminary approval, class counsel in the instant MDL and JCCP cases explained that discovery was coordinated, including shared depositions and document production. ECF No. 330-2 at 3; 330-17 at 6-7. Accordingly, the Court concludes that the lodestar claimed is unreasonably high even taking account of the work done in the JCCP case. Specifically, the Court finds that class counsel prepared limited legal filings with numerous overlapping issues, and that class counsel completed limited discovery relative to the scope of the alleged claims. Moreover, class counsel fails to explain why it took 32 law firms to do the work in this case.

Comparing the lodestar claimed in the instant case to the lodestar claimed in a novel and highly complex case, In re High-Tech, further suggests that the hours and lodestar figure here are unreasonably high. In re High-Tech involved complex antitrust issues of first impression in an action against seven large technology companies—including Google, Apple, and Intel—regarding an alleged conspiracy to fix and suppress employee compensation. 2015 WL 5158730, at *10. Class counsel in In re High-Tech engaged in many more rounds of motions practice and settled on the eve of trial. Most prominently, in the four years that the case was pending, class counsel in In re High-Tech took 93 depositions, served 28 subpoenas to third parties, survived two motions to dismiss, litigated two rounds of class certification, opposed an appeal to the Ninth Circuit under Federal Rule of Civil 23(f), survived five summary judgment motions, survived multiple rounds of Daubert challenges, filed and opposed motions in limine, prepared for the pretrial conference and trial, negotiated multiple settlements, and opposed mandamus in the Ninth Circuit. Id. More precisely, class counsel in In re High-Tech:

Id. Class counsel's efforts in In re High-Tech produced a non-reversionary settlement fund of $415 million with no claim form and a recovery of about $5,770 per class member. Id. at *12. Class counsel in In re High-Tech, which consisted of four law firms, claimed to have spent 36,215.00 hours on the litigation, yielding a lodestar figure of $18,201,787.50. Id. at *10.

In the instant case, MDL and JCCP class counsel claim to have spent 38,278.81 hours on the litigation, with a lodestar figure of $22,069,890.30, effectively for work in the instant MDL case and in the JCCP case. ECF No. 337-6 at 3. In other words, class counsel claim to have spent more time litigating this case and request a higher lodestar figure than class counsel did in In re High-Tech. This claim is surprising. Moreover, class counsel in In re High-Tech secured a significantly larger settlement of $415 million with more direct payments to class members than the $50 million settlement fund disclosed in the proposed notice here. The Court recognizes that there are differences between the instant case and In re High-Tech that may bear on the lodestar. However, the Court finds that the comparison between the instant case and In re High-Tech provides an additional indication that the hours and lodestar figure here may be unreasonably high. Moreover, the $35 million authorized in the settlement agreement is more than 50% greater than the lodestar figure. As a result, the Court concludes that the settlement may allow for improper reverter of funds to Defendants. See In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 949.

2. Percentage of Recovery Method

In their supplemental filing, class counsel explains that attorneys' fees were only calculated based on the lodestar. See ECF No. 338-1 at 18. This is inconsistent with the Ninth Circuit's guidance that the district court should employ both the lodestar and percentage-of-recovery methods to cross-check their calculations of attorneys' fees and ensure that the fees are reasonable. See In re Online DVD-Rental Antitrust Litig., 779 F.3d at 949. The Ninth Circuit specifically instructs that the percentage-of-recovery method is still applicable as a cross-check, even when the parties relied solely on the lodestar method. See In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 945 ("If the lodestar amount overcompensates the attorneys according to the 25% benchmark standard, then a second look to evaluate the reasonableness of the hours worked and rates claimed is appropriate." (quoting In re Coordinated Pretrial Proceedings in Petroleum Prods. Antitrust Litig., 109 F.3d 602, 607 (9th Cir. 1997))). Accordingly, the Court calculates attorneys' fees based on the percentage of recovery method despite the parties' failure to do so.

The only numbers to which the parties commit in the settlement agreement, motion for preliminary approval, and proposed notice are $50 million for the settlement fund, up to $35 million in attorneys' fees, and up to $2.5 million in attorneys' costs and expenses, for a total of $87.5 million. ECF No. 330-1 at 9, 19. Based on these numbers, attorneys' fees would be 40% of the settlement fund. Taking account of the additional funds the parties disclosed under seal in their supplemental filing, the Court finds that the attorneys' fees request remains much greater than the 25% benchmark standard used in this Circuit. See In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 945. Even using the lower percentage that takes into account the additional funds disclosed in the parties' supplemental filing, the Court finds that the attorneys' fees award may be unreasonably high as calculated using the percentage-of-recovery method.

"Although a percentage award in a megafund case can be 25% or even as high as 30-40%, typically the percentage award in such a case is substantially less than the 25% benchmark applicable to typical class settlements in this Circuit." Alexander v. FedEx Ground Package Sys., Inc., No. 05-CV-00038-EMC, 2016 WL 3351017, at *2 (N.D. Cal. June 15, 2016). This rule reflects the basic reality that, at some point, the increasing amount of a settlement may be a function of class size, not counsel's efforts. See In re NASDAQ Mkt.-Makers Antitrust Litig., 187 F.R.D. 465, 486 (S.D.N.Y. 1998).

This Court has previously relied on a leading study conducted by Theodore Eisenberg and Geoffrey Miller, in which the authors reviewed large common-fund settlements over a 16-year period, between 1993 and 2008. See No. 11-CV-02509-LHK, 2015 WL 5158730, at *13 (N.D. Cal. Sept. 2, 2015) (citing Theodore Eisenberg & Geoffrey P. Miller, Attorney Fees and Expenses in Class Action Settlements: 1993-2008, 7 J. Empirical Legal Stud. 248 (2010)). Previously, the Court looked to the authors' analysis of common funds exceeding $175.5 million and concluded that a percentage recovery of 9.8% or 10.5% was appropriate for a fund of $415 million. Id. Relevant here, the authors analyzed fee awards based on a sample of 69 settlements ranging from $69.6-175.5 million and found that the median percentage was 19.9% and the mean percentage was 19.4% with a standard deviation of 8.4%. Eisenberg & Miller, supra, at 265 tbl.7.

In this case, the settlement falls within the $69.6-175 million range. However, the Court finds that the attorneys' fees award substantially exceeds the 19.4% mean and 19.9% median figures in the authors' study. The maximum requested fees award appears to reflect the size of the class rather than the work of class counsel. See In re NASDAQ Mkt.-Makers Antitrust Litig., 187 F.R.D. at 486. As discussed above, the proposed settlement was submitted before the parties even completed briefing on the motion for class certification, class counsel took only 7 depositions, and the causes of action involved a limited number of California law claims.

3. Conclusion

Under both the lodestar and percentage of recovery methods, the Court finds that the maximum attorneys' fee award may be unreasonably high. As a result, the Court finds that the settlement creates a potential reverter to Defendants rather than to the benefit of the class. The Court finds that the potential reverter is not "in the class' best interest as part of the settlement package." See In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 949.

E. Failure to Adequately Disclose the Scope of Non-Monetary Relief

The Court finds that the parties fail to adequately disclose the scope of non-monetary relief. The settlement agreement does not commit to any specific increases in budget or number of employees to improve information security. ECF No. 330-3 at 10. With respect to business practices to enhance security, the settlement agreement states that Defendants shall "maintain the business practice commitments related to information security to safeguard current users' and Settlement Class Members' Personal Information as set forth in Exhibit 2 . . . for a period of no less than three years." ECF No. 330-3 at 10. The Court finds that the referenced exhibit, filed under seal, is vague as to Yahoo's specific commitments to enhance its security. As a result of the lack of specific increases in budget or number of employees and the vague commitments as to changed business practices, the Court cannot adequately consider the benefits offered to the class in settlement. See Hanlon, 150 F.3d at 1026.

F. Misleading Estimate as to the Size of the Settlement Class

The Court finds that the parties' supplemental filings have disclosed a misleading estimate of the size of the settlement class. As a result, the Court is unable to adequately assess whether the settlement is fair, reasonable, and adequate. See Lane, 696 F.3d at 818.

The parties represent that the data breaches affected 3 billion user accounts worldwide, with approximately 1 billion user accounts in the U.S. and Israel, and that the class size is approximately 200 million U.S. and Israeli individuals. Mot. at 3, 8.

In their supplemental filings, the parties explain the estimated class size as follows:

The parties assert that "Yahoo did not track or maintain data regarding the number of actual individuals that used Yahoo Mail as distinguished from the number of active accounts." ECF No. 338-1 at 5.

However, Yahoo cites the number of active users publicly when disclosure suits Yahoo's interests. For example, Yahoo CEO Marissa Mayer's July 25, 2016 press release, announcing Verizon's acquisition of Yahoo, claims that Yahoo reaches "a global audience of more than 1 billion monthly active users." Press Release, Verizon, Verizon to Acquire Yahoo's Operating Business (July 25, 2016), available at https://www.verizon.com/about/news/verizon-acquire-yahoos-operating-business. In its 2016 Annual Report to the SEC, filed in March 2017, Yahoo stated that it had "[m]ore than 650 million [] monthly users." Yahoo, 2016 Annual Report (Form 10-K), at 15 (Mar. 1, 2017).

In addition, Yahoo provided under seal the number of active user accounts in the U.S. during the relevant period. ECF No. 337-5 at 8. At the motion for preliminary approval hearing Yahoo acknowledged that the number of active user accounts would provide a more accurate estimate of the number of affected individuals than the "population study" it provided to the court. ECF No. 351 at 16. Yahoo explained that when they ran simulations to match up numbers of accounts to numbers of actual users, "the number of users was far lower than what was estimated here." Id. The Court's own review of the parties' public and sealed supplemental filings shows that the number of active user accounts in the U.S. during the relevant period was far lower than Yahoo's public calculation of 200 million affected class members.

Based on Yahoo's own representations, the Court finds that Yahoo's public estimate of the class size is inaccurate and that there are more accurate means of estimating the class size. The parties' inadequate disclosure of the size of the affected class prevents the Court from adequately assessing the strength of the plaintiffs' case and the risk of maintaining class action status throughout the trial. See Hanlon, 150 F.3d at 1026.

G. Comparison to Anthem settlement

During the hearing, the parties compared the preliminary settlement agreement favorably to the settlement approved in In re Anthem, Inc. Data Breach Litig., 327 F.R.D. 299 (2018). The Court disagrees.

Anthem involved about 79 million victims and a $115 million settlement. Defendants there timely disclosed the data breaches to affected users and provided two years of free credit monitoring to users prior to any settlement of litigation. Moreover, in addition to the clear and specific terms of the settlement fund in that case, Defendants in Anthem committed to tripling their budget for data security for three years and very specific business practice changes to improve data security.

In contrast, this case involves allegedly 200 million users according to Yahoo's public estimates, which Yahoo admits are not accurate. Yahoo's user database was breached multiple times over a period of many years, and Yahoo denied any knowledge of unauthorized access of personal data in its filings with the SEC and delayed notification to users even when it had contemporaneous knowledge of the breaches. As a result, users were unaware of the need to take any steps to protect themselves against potential misuse of their data, and Yahoo has not provided any credit monitoring on its own up to this point.

Yahoo's misrepresentations are particularly concerning. For example, on September 9, 2016, in an SEC filing regarding Verizon's purchase of Yahoo, Yahoo represented that Yahoo knew of no incidents of unauthorized access of personal data that might adversely affect the Verizon acquisition of Yahoo. Yahoo, Preliminary Proxy Statement (Schedule 14(A), at Exhibit A-18 (Sept. 9, 2016), available at https://www.altaba.com/static-files/ad5f11da-0a78-4f3e-90f8-dd204c1978fb. However, thirteen days later, on September 22, 2016, Yahoo publicly disclosed the 2014 data breach. ECF No. 196, ¶ 126. In the announcement, Yahoo claimed that it learned of the 2014 data breach during a "recent investigation." Yahoo Security Notice September 22, 2016, available at https://help.yahoo.com/kb/%20SLN28092.html. Six months later, Yahoo admitted on March 1, 2017 in its 10-K filing with the SEC that Yahoo had "contemporaneous knowledge" of the 2014 data breach. ECF No. 196, ¶ 129; Yahoo, 2016 Annual Report (Form 10-K), at 47 (Mar. 1, 2017).

Moreover, Yahoo continues to deny any data breaches in 2012 but wants a release of any claims arising from that time period. Unlike Anthem, Yahoo's data was sold on the dark web, and Yahoo bought back the data on the dark web. Plaintiffs allege that others have also purchased Yahoo's data on the dark web. In the settlement agreement, Yahoo has only committed to the $50 million settlement fund and hides the total settlement fund amount. Yahoo misrepresents the number of affected Yahoo users by publicly filing an inflated, inaccurate calculation of users and simultaneously filing under seal a more accurate, much smaller number. Yahoo has not committed to any specific increases in budget for data security and has made only vague commitments as to specific business practices to improve data security. Yahoo's history of nondisclosure and lack of transparency related to the data breaches are egregious. Unfortunately, the settlement agreement, proposed notice, motion for preliminary approval, and public and sealed supplemental filings continue this pattern of lack of transparency.

IV. CONCLUSION

In sum, based on the parties' current filings, the Court cannot conclude that the settlement is "fundamentally fair, adequate, and reasonable." Hanlon, 150 F.3d at 1026. For the foregoing reasons, Plaintiffs' motion for preliminary approval of class action settlement is DENIED.2

IT IS SO ORDERED.