Campbell v. Facebook, Inc.

951 F.3d 1106

• Court: Court of Appeals for the Ninth Circuit
• Decided: March 3, 2020
• Docket: 17-16873
• Status: Published

Opinion

FOR PUBLICATION

UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT

MATTHEW CAMPBELL; MICHAEL No. 17-16873 HURLEY, on behalf of themselves and all others similarly situated, D.C. No. Plaintiffs-Appellees, 4:13-cv-05996- PJH v.

FACEBOOK, INC., OPINION Defendant-Appellee,

v.

ANNA W. ST. JOHN, Objector-Appellant.

Appeal from the United States District Court for the Northern District of California Phyllis J. Hamilton, Chief Judge, Presiding

Argued and Submitted January 18, 2019 Submission Vacated June 4, 2019 Resubmitted March 3, 2020 San Francisco, California

Filed March 3, 2020

Before: J. Clifford Wallace, Richard R. Clifton, and Michelle T. Friedland, Circuit Judges. 2 CAMPBELL V. ST. JOHN

Opinion by Judge Friedland

SUMMARY *

Objector / Class Action Settlement

In an appeal brought by an objecting class member, the panel affirmed the district court’s approval of a settlement between Facebook and a nationwide class of its users who alleged that Facebook routinely used website links in users’ private messages without their consent in violation of federal and California privacy laws.

Facebook acknowledged in the settlement agreement that it had already made several changes to the practices challenged in this action, and it agreed to add a disclosure to a Help Center page on its website for a year. The district court, over the objector’s challenge, found the settlement to be fair and approved it; and granted in full class counsel’s request for $3.89 million in fees and costs.

As a threshold matter, the panel held that the plaintiff class had Article III standing to bring the case. First, the panel held that the plaintiffs identified a concrete injury. Specifically, the panel concluded that the plaintiffs identified a concrete injury by claiming that Facebook violated the federal Electronic Communications Privacy Act and the California Invasion of Privacy Act when it intercepted, catalogued, and used without consent URLs that users had shared in private messages. Second, the panel held that the

* This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. CAMPBELL V. ST. JOHN 3

plaintiffs established standing to seek injunctive relief, and post-filing developments did not moot this case. The panel concluded that the district court had jurisdiction to approve the settlement, and the panel therefore had jurisdiction to review the merits of that decision.

The panel rejected the objector’s challenges to the substantive fairness of the settlement. First, the panel rejected the argument that the settlement was invalid because the class received only “worthless injunctive relief.” Koby v. ARS National Services, Inc., 846 F.3d 1071, 1081 (9th Cir. 2017). The panel held that the district court did not clearly err in finding that the settlement’s injunctive relief had value to absent class members. Moreover, the class did not need to receive much for the settlement to be fair because the class gave up very little. Second, the panel rejected the objector’s argument that the settlement was invalid under In re Bluetooth Headset Products Liability Litigation, 654 F.3d 935 (9th Cir. 2010), because it prioritized class counsel’s interests over those of their clients. The panel held that the district court looked at the Bluetooth warning signs of possible collusion between class counsel and Facebook, and the district court did not abuse its discretion in concluding that none of the warning signs weighed against approval of the settlement. The panel concluded that applying the Bluetooth framework did not demonstrate that the settlement in this case was unfair. 4 CAMPBELL V. ST. JOHN

COUNSEL

Adam Ezra Schulman (argued), Anna St. John, and Theodore H. Frank, Center for Class Action Fairness, Washington, D.C.; for Objector-Appellant.

Hank Bates (argued), Allen Carney, and David Slade, Carney Bates & Pulliam PLLC, Little Rock, Arkansas; Michael W. Sobol, David T. Rudolph, and Melissa Gardner, Lieff Cabraser Heimann & Bernstein LLP, San Francisco, California; Rachel Geman and Nicholas Diamand, Lieff Cabraser Heimann & Bernstein LLP, New York, New York; for Plaintiffs-Appellees.

Christopher Chorba (argued), Joshua A. Jessen, Ashley M. Rogers, and Ryan S. Appleby, Gibson Dunn & Crutcher LLP, Los Angeles, California, for Defendant-Appellee.

Marc Rotenberg, Alan Butler, and Sam Lester, Electronic Privacy Information Center, Washington, D.C., for Amicus Curiae Electronic Privacy Information Center (EPIC).

OPINION

FRIEDLAND, Circuit Judge:

Objecting class member Anna St. John (“Objector”) appeals from the district court’s approval of a settlement between Facebook and a nationwide class of its users who alleged that Facebook routinely captured, read, and used for several purposes the website links included in users’ private messages without their consent, and that these practices violated federal and California privacy laws. After years of litigation that included lengthy discovery, four mediation CAMPBELL V. ST. JOHN 5

sessions, and Facebook’s failed attempts to convince the district court to dismiss the case or deny class certification, the parties reached a settlement. Facebook acknowledged in the settlement agreement that it had already made several changes to the practices challenged in this action, and it agreed to add a disclosure to a Help Center page on its website for a year. The settlement agreement also provided that class counsel could apply for court approval of up to $3.89 million in attorney’s fees and costs, and that Facebook would not take any position on that application. The district court, over Objector’s challenge, found the settlement to be fair and approved it. The district court also granted in full class counsel’s request for $3.89 million in fees and costs.

Addressing Objector’s appeal from the district court’s approval of the settlement, we first consider whether Plaintiffs had standing to bring this action and whether it later became moot. We conclude that the district court had jurisdiction, and, accordingly, that we have jurisdiction to evaluate the fairness of the settlement. Second, we reject on the merits Objector’s contentions that the district court abused its discretion by approving the settlement.

I.

A.

“Facebook operates one of the largest social media platforms in the world, with over one billion active users. About seven in ten adults in the United States use Facebook.” Patel v. Facebook, Inc., 932 F.3d 1264, 1267 (9th Cir. 2019) (citations omitted), cert. denied, No. 19-706 (U.S. Jan. 21, 2020). Facebook has a messaging function on its platform that allows users to send electronic messages to one or more other users. Facebook explains on its website that these messages are “private” because their contents and 6 CAMPBELL V. ST. JOHN

history are viewable only by the sender and his or her chosen recipients—in contrast to, for example, posts shared with a broader audience, such as all of the user’s Facebook friends.

In December 2013, Matthew Campbell and Michael Hurley (“Plaintiffs”) filed a putative class action against Facebook. Plaintiffs alleged that Facebook scanned their private messages looking for links to web pages, also known as URLs, contained in those messages. They alleged that if a message contained a URL, Facebook would collect that information and use it in a variety of ways without the user’s consent.

The main allegations concerned how Facebook integrated these private message URL shares into a feature that enabled third parties to show on their own websites a count of how many Facebook users had “Liked” the pages on their sites—a proxy for those pages’ popularity. Plaintiffs alleged that Facebook would increase a page’s “Like” counter not only when a Facebook user affirmatively pressed a “Like” button, but also when the user sent a private message containing a URL corresponding to the page, regardless of what the message said about the URL. 1 Plaintiffs also alleged that Facebook used the private message URL data that it was collecting to help build profiles of individual users that could facilitate, among other things, targeted advertising on Facebook.

Plaintiffs contended that Facebook’s handling of their messages amounted to interception and use of electronic communications in violation of Title I of the Electronic

1 Plaintiffs did not allege that it was possible for users or third parties to tell anything from the “Like” counter other than the total number of users whose activity had resulted in a “Like” counter increase. CAMPBELL V. ST. JOHN 7

Communications Privacy Act (“ECPA”), 18 U.S.C. § 2510 et seq., 2 and the California Invasion of Privacy Act (“CIPA”), Cal. Penal Code § 630 et seq. They also alleged that it amounted to an unlawful, unfair, or fraudulent business practice under the California Unfair Competition Law (“UCL”), Cal. Bus. & Prof. Code § 17200 et seq. Plaintiffs sought damages as well as declaratory and injunctive relief.

Facebook filed a motion to dismiss, which the district court granted in part and denied in part. The district court dismissed the UCL claim and a portion of the CIPA claim, but it declined to dismiss the ECPA claim and the portion of the CIPA claim alleging interception and use of communications in violation of California Penal Code section 631. The district court also rejected Facebook’s sole jurisdictional argument: that the prayer for injunctive relief should be stricken for failure to allege ongoing or future injury. 3

The parties engaged in extensive discovery, which included the production of tens of thousands of pages of documents, depositions of eighteen fact and expert witnesses, hundreds of hours of analysis of Facebook’s

2 Title I of ECPA amended the federal Wiretap Act. For this reason, the provisions at issue in this case are often also referred to as part of the Wiretap Act.

3 Specifically, Facebook pointed out that Plaintiffs had not alleged that the “Like” counting practice was still in place. The district court held that Plaintiffs had sufficiently alleged that, even if Facebook had temporarily stopped that practice, Facebook was likely enough to inflict future injury in a similar manner to support standing to seek injunctive relief. 8 CAMPBELL V. ST. JOHN

source code, and significant briefing about discovery disputes.

Plaintiffs moved for certification of a damages class under Federal Rule of Civil Procedure 23(b)(3), and in the alternative an injunctive and declaratory relief class under Rule 23(b)(2). Facebook opposed the motion, arguing that Plaintiffs failed to satisfy Rule 23’s requirements—but not arguing that the court lacked jurisdiction. 4 In May 2016, the district court granted the motion in part and denied it in part.

The court denied certification of a damages class. It explained that Plaintiffs had proposed two methods for calculating damages but that neither was acceptable. One of Plaintiffs’ proposals was to measure Facebook’s profits from intercepting and using URL data from messages. The court explained, however, that Plaintiffs’ proposed methodology had unjustifiably assumed that all interceptions resulted in the same profit, which made the model too inaccurate. Plaintiffs’ other proposal was to award statutory damages, which fared no better. The court reasoned that it would be required to either award the full statutory sum or nothing, but “many individual damages awards [of that full sum] would be disproportionate” to the small amount of harm suffered by many class members. It held that class treatment was

4 Facebook included two statements in its opposition to Plaintiffs’ class certification motion that gestured toward standing: (1) that the testimony of one (but not both) of the named Plaintiffs “calls into question his standing under Article III to seek injunctive relief”; and (2) that “Facebook also reserves its rights pursuant to [the Supreme Court’s then-pending decision in the case that would ultimately be issued a few months later as Spokeo, Inc. v. Robins (Spokeo I), 136 S. Ct. 1540 (2016)].” Although Facebook has argued that these statements “raised [the] issue [of jurisdiction] at class certification,” neither was in fact an argument that jurisdiction was lacking. CAMPBELL V. ST. JOHN 9

inappropriate because “sorting out those disproportionate damages awards would require individualized analyses.”

The district court granted certification of an injunctive and declaratory relief class. The certified class encompassed:

All natural-person Facebook users [aside from those excluded through standard carveout provisions] located within the United States who have sent, or received from a Facebook user, private messages that included URLs in their content (and from which Facebook generated a URL attachment), from [December 2011] up through the date of the certification of the class.

In the class certification order, the district court observed that Plaintiffs had focused their claims on three specific uses of the URL data that had been collected from private messages: (1) Facebook’s counting URL shares as a “Like” of the relevant third-party web page; (2) Facebook’s sharing data regarding URLs in messages with third parties, enabling those third parties to generate customized content and targeted advertising on their own websites informed by this data; and (3) Facebook’s use of the URL data to generate recommendations for other Facebook users. The district court concluded that although Plaintiffs had made allegations in their complaint about the first of these uses and “arguably” about the third, they had not specifically alleged the predicate for the second use, “sharing of data with third parties.” But the district court further concluded that Plaintiffs’ decision to focus on all three of these uses of URL data was “based on a review of discovery that was not 10 CAMPBELL V. ST. JOHN

available at the time of the complaint’s filing.” The district court therefore permitted Plaintiffs to represent the class in challenging all three uses, and it directed them to file a conforming amended complaint (which they then did).

The parties expended significant effort to try to settle this case. Several months after the motion to dismiss ruling, they participated in a full-day mediation session. The parties returned to the negotiating table after the district court’s May 2016 order certifying an injunctive and declaratory relief class. During the last several months of 2016, the parties attended a total of three mediation sessions and continued to negotiate informally. About a week before the fact discovery period was scheduled to close, the district court approved the parties’ stipulation to stay discovery and vacate existing deadlines to facilitate their settlement efforts. Shortly thereafter, in December 2016—almost two years into the parties’ extensive discovery—the parties reached an agreement in principle during their fourth mediation session. A few months later, the parties executed a written settlement agreement, which they then submitted to the district court for approval.

In the settlement agreement, Facebook acknowledged that it had at one point used the URL data in the three ways the district court had described in its class certification order—but it also represented that it had since stopped each of them. Two of the data uses had ceased before this action was filed in late 2013: in December 2012, Facebook stopped using private message URLs to increase “Like” counts, and, in October 2012, Facebook stopped sharing with third party websites “information about URL shares in Facebook messages . . . and attendant statistics and demographic information.” The third use, which the settlement agreement specified was related to Facebook’s using the URL data in CAMPBELL V. ST. JOHN 11

its Recommendations Feed, ceased in July 2014, several months after this action was filed. Facebook further represented that these three uses had relied on “anonymous, aggregate” information curated from the URL shares. Facebook separately confirmed that, as of the date of the settlement, it was not using any private message URL data for targeted advertising (as Plaintiffs had initially alleged) and was not sharing with third parties any personally identifying user information associated with the URL data.

The settlement agreement further described “enhanced disclosures and practice changes” that Facebook had made after this case was filed. The agreement pointed out, for example, that Facebook had revised its Data Policy about a year after this action was filed to state that Facebook collects the “content and other information” that users provide when they “message or communicate with others,” and to further explain the ways in which Facebook may use that information.

Facebook agreed as part of the settlement to display for one year a new twenty-two-word disclosure in the Help Center portion of its site, stating: “We use tools to identify and store links shared in messages, including a count of the number of times links are shared.” 5 Although Facebook promised to display this Help Center disclosure for a year, it did not say that it would continue to refrain from any of the uses of URL data described above. Nor did Facebook promise to continue using the version of the Data Policy adopted after this action was filed. The agreement did not provide for monetary compensation to class members other

5 The agreement provided “that Facebook may update the disclosure[] to ensure accuracy with ongoing product changes.” 12 CAMPBELL V. ST. JOHN

than the two named Plaintiffs, each of whom was permitted to apply to the court for an award not to exceed $5,000.

In exchange, class members were required to release their declaratory and injunctive relief claims. Absent class members did not, however, release any “claims for monetary relief, damages, or statutory damages.” Only the named Plaintiffs released their damages claims.

The agreement provided that class counsel could request that the court award attorney’s fees and costs of up to $3.89 million—an amount that the parties represented had been negotiated after and independent of the other settlement terms and constituted a significant reduction from the more than $7 million that class counsel claimed would fully compensate them for their work on the case. Facebook agreed not to object to that request and to pay any amount the court approved up to this $3.89 million cap.

The district court granted preliminary approval of the settlement. Although the parties had agreed that it was unnecessary to provide notice to the class beyond what was already publicly available (primarily through news coverage), the court rejected that proposition. It held that class counsel would be required to post information about the settlement on their public websites during the period between preliminary approval and a final fairness hearing.

Objector Anna St. John, who is a member of the class and an attorney at the Center for Class Action Fairness, filed an objection to the settlement. Following a final fairness hearing, the district court approved the settlement. Evaluating the settlement using the factors outlined in Hanlon v. Chrysler Corp., 150 F.3d 1011, 1026 (9th Cir. 1998), the district court reasoned that “[t]he settlement offers immediate, tangible benefits directed to the three uses of CAMPBELL V. ST. JOHN 13

URLs challenged by Plaintiffs, without requiring class members to release any claims for monetary damages that they may have against Facebook.” The court also explained that “the relief to the class must be viewed against the likely rewards of litigation,” and observed that “proceeding with litigation would be very risky for the class.” In addition, the court emphasized that the settlement was “the result of four in-person, arms’-length mediations before two different mediators,” that both sides were able “to negotiate the settlement on a fully-informed basis,” and that “[c]lass counsel [are] highly experienced.”

The district court also granted the full $3.89 million in attorney’s fees and costs for which class counsel had applied. In addition to deeming the amount “reasonable” in light of the results obtained for the class and counsel’s substantial investment of time and resources on a contingency basis, the court determined, based on an assessment of the factors listed in In re Bluetooth Headset Products Liability Litigation, 654 F.3d 935, 947 (9th Cir. 2011), that this portion of the settlement agreement did not indicate collusion among the negotiating parties at the expense of absent class members.

Objector timely appealed the district court’s approval of the settlement.

B.

After oral argument in this appeal, the Supreme Court issued Frank v. Gaos, 139 S. Ct. 1041, 1045–46 (2019) (per curiam), which vacated the final approval of a class action settlement and remanded for an assessment of Article III standing under Spokeo, Inc. v. Robins (Spokeo I), 136 S. Ct. 1540 (2016). In Gaos, the plaintiffs had claimed that Google violated the Stored Communications Act by transmitting 14 CAMPBELL V. ST. JOHN

information to third-party websites about the search terms users had entered to arrive at those websites. See 139 S. Ct. at 1044. Although the Supreme Court had granted certiorari intending to resolve a question about use of cy pres awards in class action settlements, 6 the Court did not reach that issue “[b]ecause there remain[ed] substantial questions about whether any of the named plaintiffs ha[d] standing to sue in light of [Spokeo I].” Id. at 1043–44. The Court explained that federal courts’ “‘obligation to assure ourselves of litigants’ standing under Article III’ . . . extends to court approval of proposed class action settlements.” Id. at 1046 (quoting DaimlerChrysler Corp. v. Cuno, 547 U.S. 332, 340 (2006)).

Because of the possibility that this case would present “substantial questions” about standing like the ones the Court identified in Gaos, id. at 1043, we requested and received supplemental briefing from the parties about the effect, if any, of Spokeo I on jurisdiction in this case.

II.

To establish standing, plaintiffs must show that they have suffered an injury in fact that is fairly traceable to the challenged conduct of the defendant and is likely to be redressed by a favorable judicial decision. Lujan v. Defs. of Wildlife, 504 U.S. 555, 560–61 (1992). Injury in fact is “the

6 Cy pres refers to a method for distributing unclaimed settlement funds “to the ‘next best’ class of beneficiaries.” Nachshin v. AOL, LLC, 663 F.3d 1034, 1036 (9th Cir. 2011). “Under the cy pres approach, ‘class members receive an indirect benefit (usually through defendant donations to a third party) rather than a direct monetary payment.’” In re EasySaver Rewards Litig., 906 F.3d 747, 760 (9th Cir. 2018) (quoting Lane v. Facebook, Inc., 696 F.3d 811, 819 (9th Cir. 2012)), cert. denied, 139 S. Ct. 2744 (2019). CAMPBELL V. ST. JOHN 15

‘[f]irst and foremost’ of standing’s three elements.” Spokeo, Inc. v. Robins (Spokeo I), 136 S. Ct. 1540, 1547 (2016) (alteration in original) (quoting Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 103 (1998)). Among other things, “an injury in fact must be both concrete and particularized.” Id. at 1548.

“[A] plaintiff must demonstrate standing separately for each form of relief sought.” Friends of the Earth, Inc. v. Laidlaw Envtl. Servs. (TOC), Inc., 528 U.S. 167, 185 (2000). “[A]s the party invoking federal jurisdiction, [the plaintiff] bears the burden of establishing” standing. Spokeo I, 136 S. Ct. at 1547.

Our discussion of standing proceeds in two parts. We first conclude that Plaintiffs identified a concrete injury. 7 Next, we hold that Plaintiffs established standing to seek injunctive relief and that post-filing developments did not moot this case. These conclusions satisfy us that there was and is Article III jurisdiction over this case, and that we may therefore consider the merits of Objector’s challenges to the approval of the settlement.

An injury is concrete for purposes of standing if it “actually exist[s],” meaning it is “real, and not abstract”— but not necessarily “tangible.” Spokeo I, 136 S. Ct. at 1548– 49 (quotation marks omitted). Where, as here, we deal with an “intangible harm” that is linked to a statutory violation,

7 Once we conclude that this was a concrete injury, it is clear that it was also particularized, fairly traceable to Facebook, and likely to be redressed by a favorable judicial decision. The parties have not contested any of those other standing requirements in their supplemental briefs, and we do not discuss them in further detail. 16 CAMPBELL V. ST. JOHN

we are guided in determining concreteness by “both history and the judgment of Congress,” or the legislature that enacted the statute. See id. at 1549; Patel v. Facebook, Inc., 932 F.3d 1264, 1273 (9th Cir. 2019) (looking to “[t]he judgment of the Illinois General Assembly” to inform the Article III standing inquiry for a claim alleging a violation of an Illinois privacy statute), cert. denied, No. 19-706 (U.S. Jan. 21, 2020). Historical practice is “instructive” as to “whether an alleged intangible harm has a close relationship to a harm that has traditionally been regarded as providing a basis for a lawsuit in English or American courts.” Spokeo I, 136 S. Ct. at 1549. We also look to legislative judgment because legislatures may “elevat[e] to the status of legally cognizable injuries concrete, de facto injuries that were previously inadequate in law.” Id. (alteration in original) (quoting Lujan, 504 U.S. at 578). But the fact that a legislature has “grant[ed] a person a statutory right and purport[ed] to authorize that person to sue to vindicate that right” is not by itself sufficient for standing. Id. When a legislature has enacted a “bare procedural” protection, a plaintiff “cannot satisfy the demands of Article III” by pointing only to a violation of that provision, but also must link it to a concrete harm. Id. at 1550 (emphasis added). When, however, a statutory provision identifies a substantive right that is infringed any time it is violated, a plaintiff bringing a claim under that provision “need not allege any further harm to have standing.” Eichenberger v. ESPN, Inc., 876 F.3d 979, 983–84 (9th Cir. 2017).

For the reasons that follow, we conclude that the statutory provisions under which Plaintiffs sued protect concrete interests because, like the provisions examined in several of our recent decisions, they “codif[y] a context- specific extension of the substantive right to privacy.” See id. at 983. CAMPBELL V. ST. JOHN 17

ECPA includes a private right of action, 18 U.S.C. § 2520, against anyone who “intentionally intercepts, endeavors to intercept, or procures any other person to intercept or endeavor to intercept, any wire, oral, or electronic communication,” 18 U.S.C. § 2511(1)(a). Plaintiffs sued under that provision and also alleged a violation of ECPA’s prohibition on the intentional use of the contents of information knowingly obtained through such interception. See id. § 2511(1)(d). Plaintiffs also sued under CIPA, which likewise includes a private right of action, Cal. Penal Code § 637.2(a), and which similarly prohibits the unauthorized reading, or attempting to read, of “any message, report, or communication while the same is in transit or passing over any wire, line, or cable,” as well as the use of “any information so obtained.” Cal. Penal Code § 631(a).

The harms protected by these statutes bear a “close relationship” to ones that have “traditionally been regarded as providing a basis for a lawsuit.” See Spokeo I, 136 S. Ct. at 1549. “Violations of the right to privacy have long been actionable at common law.” Eichenberger, 876 F.3d at 983. And one of the several privacy torts historically recognized was “unreasonable intrusion upon the seclusion of another,” which traditionally extends to, among other things, “tapping . . . telephone wires” as well as “opening . . . private and personal mail.” Restatement (Second) of Torts § 652B cmt. b. There is a straightforward analogue between those traditional torts and the statutory protections codified in ECPA and CIPA against viewing or using private communications. Moreover, under the privacy torts that form the backdrop for these modern statutes, “[t]he intrusion itself makes the defendant subject to liability.” Restatement (Second) of Torts § 652B cmt. b. “In other words, ‘privacy torts do not always require additional consequences to be 18 CAMPBELL V. ST. JOHN

actionable.’” Patel, 932 F.3d at 1274 (quoting Eichenberger, 876 F.3d at 983). Thus, historical practice provides support not only for the conclusion that wiretapping is actionable, but also for the conclusion that a wiretapping plaintiff “need not allege any further harm to have standing.” See Eichenberger, 876 F.3d at 984.

The reasons articulated by the legislatures that enacted ECPA and CIPA further indicate that the provisions at issue in this case reflect statutory modernizations of the privacy protections available at common law. The purpose of ECPA includes creating “[f]ederal statutory standards,” analogous to the “protection against unauthorized opening” of mail, “to protect the privacy and security of communications” made using newer technology. S. Rep. No. 99-541, at 5 (1986), as reprinted in 1986 U.S.C.C.A.N. 3555, 3559. CIPA’s purpose similarly includes “protect[ing] the right of privacy of the people of [California],” because “the invasion of privacy” resulting from the use of new technology to “eavesdrop[] upon private communications” causes “a serious threat to the free exercise of personal liberties.” Cal. Penal Code § 630. Plaintiffs’ challenges here arise under the core provisions of those statutes regarding interception and use of private communications. We respect the legislatures’ judgment about the importance of the privacy interests violated when communications are intercepted, as reflected in their decisions to enact a private right of action that is available when these provisions are infringed. See Spokeo I, 136 S. Ct. at 1549.

Our precedent confirms that Plaintiffs have asserted a concrete harm. We have, in the years since Spokeo I, identified several statutory provisions that guard against invasions of concrete privacy interests. See, e.g., Patel, 932 F.3d at 1269, 1271–75 (concrete interests protected by CAMPBELL V. ST. JOHN 19

the Illinois Biometric Information Privacy Act’s requirements relating to private entities’ “collection, retention, disclosure, and destruction of biometric identifiers and biometric information,” such as face templates (quoting Rosenbach v. Six Flags Entm’t Corp., 129 N.E.3d 1197, 1203 (Ill. 2019))); Eichenberger, 876 F.3d at 981, 983–84 (concrete interests protected by the Video Privacy Protection Act’s requirement that “video tape service providers” not disclose “personally identifiable information concerning any consumer of such provider” (quoting 18 U.S.C. § 2710(b)(1))); Van Patten v. Vertical Fitness Grp., LLC, 847 F.3d 1037, 1041–43 (9th Cir. 2017) (concrete interests protected by the Telephone Consumer Protection Act, which “establishes the substantive right to be free from certain types of phone calls and texts absent consumer consent”). There is no meaningful distinction between the concrete, substantive privacy interests protected by the statutes at issue in Patel, Eichenberger, and Van Patten and the interests protected by the provisions of ECPA and CIPA at issue in this case. For example, just as the Video Privacy Protection Act’s prohibition on disclosure of information about an individual’s video rentals protects a substantive privacy interest, rather than merely codifying “a procedure that video service providers must follow,” Eichenberger, 876 F.3d at 983, ECPA and CIPA section 631 are targeted at the substantive intrusion that occurs when private communications are intercepted by someone who does not have the right to access them, rather than merely setting out a procedure for handling data.

For all of the foregoing reasons, every violation of the provisions of ECPA and CIPA at issue in this case “‘present[s] the precise harm and infringe[s] the same privacy interests Congress [and the California legislature] sought to protect’ by enacting” ECPA and CIPA section 631. 20 CAMPBELL V. ST. JOHN

See Eichenberger, 876 F.3d at 984 (first two alterations in original) (quoting Van Patten, 847 F.3d at 1043). The Third Circuit has indeed already held that violations of ECPA and CIPA of the type alleged here “involve[] a clear de facto injury.” In re Nickelodeon Consumer Privacy Litig., 827 F.3d 262, 274 (3d Cir. 2016); see also In re Google Inc. Cookie Placement Consumer Privacy Litig., 934 F.3d 316, 325 (3d Cir. 2019) (concluding, in accord with Nickelodeon and after Gaos was decided, that “a concrete injury for Article III standing purposes occurs when Google, or any other third party, tracks a person’s internet browser activity without authorization”). We agree. We thus conclude that Plaintiffs identified a concrete injury by claiming that Facebook violated ECPA and CIPA when it intercepted, catalogued, and used without consent URLs they had shared in private messages. 8

Facebook initially did not contest in this appeal that Plaintiffs had standing, and it argued that we should affirm the settlement approval. Facebook now argues in its supplemental brief, however, that the settlement should be vacated and the case dismissed because Plaintiffs lacked standing to bring this case. Specifically, Facebook argues

8 Robins v. Spokeo, Inc. (Spokeo II), 867 F.3d 1108 (9th Cir. 2017), further supports our conclusion. There, on remand from the Supreme Court, we held that the plaintiff had alleged sufficiently concrete injuries resulting from Spokeo’s failure to follow procedures of the Fair Credit Reporting Act that ensure accuracy of consumer report information. Id. at 1110–11, 1117. Standing is easier to prove here than it was in Spokeo because “although the [Fair Credit Reporting Act] outlines procedural obligations that sometimes protect individual interests,” ECPA and CIPA section 631, as discussed above, codify “a substantive right to privacy” the intrusion of which causes concrete harm “any time” there is a violation. See Eichenberger, 876 F.3d at 983–84. CAMPBELL V. ST. JOHN 21

that Plaintiffs suffered no concrete harm from the “use of anonymized and aggregated data from website links.”

But, ultimately, this new argument is beside the point in the context of this case. Plaintiffs alleged, and Facebook has confirmed (through, among other things, its revised Data Policy), that Facebook identifies and collects the contents of users’ individual private messages. Plaintiffs’ position that this was being done without consent meant that they claimed a violation of the concrete privacy interests that ECPA and CIPA protect, regardless of how the collected data was later used. No more is needed to support standing under Spokeo. 9

To have standing to seek to enjoin Facebook’s private message practices, Plaintiffs must show “either ‘continuing, present adverse effects’ due to [their] exposure to

9 Facebook’s supplemental brief also argues that, because Plaintiffs consented to the uses of URL data, they lack standing. Both this argument and Facebook’s contention about anonymized and aggregated use of data are better understood as arguments that ECPA and CIPA were not actually violated by the practices challenged here. Such merits arguments in disguise tell us nothing about whether Plaintiffs had standing to bring the case in the first place. See Kirola v. City & County of San Francisco, 860 F.3d 1164, 1175 (9th Cir. 2017) (rejecting an injury-in-fact argument that would have meant there was “no difference between [the plaintiff] succeeding on the merits and establishing standing to assert her claims in the first place”). As explained below, the merits of this case would have turned on several legal questions that Plaintiffs have since acknowledged are difficult and unresolved. Because the parties settled this case rather than continuing to litigate, we have no occasion to resolve any of those questions here. We emphasize that this case also does not present the question whether standing could be based entirely on injury from anonymized, aggregated uses of data because Plaintiffs also focused on unconsented-to collection and storage of information from private messages that enabled those uses. 22 CAMPBELL V. ST. JOHN

[Facebook’s] past illegal conduct or ‘a sufficient likelihood that [they] will again be wronged in a similar way.’” Villa v. Maricopa County, 865 F.3d 1224, 1229 (9th Cir. 2017) (first quoting O’Shea v. Littleton, 414 U.S. 488, 495–96 (1974); then quoting City of Los Angeles v. Lyons, 461 U.S. 95, 111 (1983)). Objector and Facebook both contend that Plaintiffs did not satisfy this requirement. We disagree.

This aspect of standing, like any other, must “focus[] on whether the party invoking jurisdiction had the requisite stake in the outcome when the suit was filed.” Davis v. FEC, 554 U.S. 724, 734 (2008); see also Slayman v. FedEx Ground Package Sys., Inc., 765 F.3d 1033, 1047–48 (9th Cir. 2014) (“When evaluating whether [the standing] elements are present, we must look at the facts as they exist at the time the complaint was filed.” (alteration in original) (quoting Am. Civil Liberties Union of Nev. v. Lomax, 471 F.3d 1010, 1015 (9th Cir. 2006))). It turns out that some (but not all) of Facebook’s challenged uses of private message URL data had ended before Plaintiffs sued, as Facebook ultimately acknowledged in the settlement agreement. But when Plaintiffs sued, Facebook was actively accessing private messages—conduct that Facebook has never claimed to have ceased. Facebook was also still using the data from these messages for its Recommendations Feed. And Facebook’s ongoing retention of the data collected from private messages meant that there was a risk that it would resume using the data for “Like” counters, resume sharing the data with third parties, or begin using the data for some other purpose. This combination of continuing harm plus likelihood of future harm was sufficient for Plaintiffs to have standing to seek injunctive relief.

To be sure, Facebook apparently did stop using data from private messages in its Recommendations Feed after this CAMPBELL V. ST. JOHN 23

case was filed, and did not resume either that use of URL data or the already-stopped uses of “Like” counting and third-party sharing at any time between the action’s filing and its settlement. But the question whether Facebook’s conduct since Plaintiffs filed this action made it unlikely that Plaintiffs would again be injured by the challenged practices presents a separate issue: whether their challenges had become moot. There are “important difference[s] between” standing and mootness, motivated in part by the “wasteful[ness]” of “abandon[ing]” cases as moot that “ha[ve] been brought and litigated, often (as here) for years.” Friends of the Earth, 528 U.S. at 191–92. Among the differences between standing and mootness is that, to show that Plaintiffs’ claims had become moot, Facebook would have needed to satisfy “the formidable burden of showing that it is absolutely clear the allegedly wrongful behavior could not reasonably be expected to recur.” See id. at 190. Neither Facebook nor Objector has suggested that Facebook could have carried that heavy burden, nor does the record demonstrate that it could have.

* * *

In sum, we conclude that the district court had jurisdiction to approve the settlement, and that we therefore have jurisdiction to review the merits of that decision.

III.

Under Federal Rule of Civil Procedure 23(e)(2), a district court may approve a class action settlement only after finding that the settlement is “fair, reasonable, and adequate.” Courts reviewing class action settlements must “ensure[] that unnamed class members are protected ‘from 24 CAMPBELL V. ST. JOHN

unjust or unfair settlements affecting their rights,’” while also accounting for “the ‘strong judicial policy that favors settlements, particularly where complex class action litigation is concerned.’” In re Hyundai & Kia Fuel Econ. Litig., 926 F.3d 539, 556, 568 (9th Cir. 2019) (en banc) (first quoting Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 623 (1997); then quoting Allen v. Bedolla, 787 F.3d 1218, 1223 (9th Cir. 2015)).

We have described several factors for district courts to consider when evaluating the fairness of a class action settlement:

[1] the strength of the plaintiffs’ case; [2] the risk, expense, complexity, and likely duration of further litigation; [3] the risk of maintaining class action status throughout the trial; [4] the amount offered in settlement; [5] the extent of discovery completed and the stage of the proceedings; [6] the experience and views of counsel; [7] the presence of a governmental participant; and [8] the reaction of the class members to the proposed settlement.

Hanlon v. Chrysler Corp., 150 F.3d 1011, 1026 (9th Cir. 1998). District courts may consider some or all of these factors. See Rodriguez v. West Publ’g Corp., 563 F.3d 948, 963 (9th Cir. 2009). 10

10 After the district court’s approval of the settlement in this case, Rule 23(e)(2) was amended. Whereas the rule previously did not expand upon what was necessary for a settlement to be “fair, reasonable, and adequate,” it now lists criteria that are relevant to that determination. We CAMPBELL V. ST. JOHN 25

Our review of a district court’s decision to approve a class action settlement is “extremely limited.” Hanlon, 150 F.3d at 1026. “Parties seeking to overturn the settlement approval must make a ‘strong showing’ that the district court clearly abused its discretion.” Hyundai, 926 F.3d at 556 (quoting Linney v. Cellular Alaska P’ship, 151 F.3d 1234, 1238 (9th Cir. 1998)). A district court clearly abuses its discretion by either failing to apply the correct legal standard or by making clearly erroneous factual determinations. See id. When the issue presented is the substantive fairness of the settlement, we must refrain from “substitut[ing] our notions of fairness for those of the district judge.” In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d 935, 950 (9th Cir. 2011) (alteration in original) (quoting Officers for Justice v. Civil Serv. Comm’n, 688 F.2d 615, 626 (9th Cir. 1982)); see also Lane v. Facebook, Inc., 696 F.3d 811, 818 (9th Cir. 2012) (explaining that the “district court should have broad discretion because it ‘is exposed to the litigants, and their strategies, positions and proof’” (quoting Hanlon, 150 F.3d at 1026)).

The issue Objector raises here is the substantive fairness of the settlement, so we approach our review with substantial deference. And this case does not implicate the “higher standard of fairness” that applies when parties settle a case

need not determine whether this amendment should be applied retroactively here because applying the amended version of the rule would not change our conclusions. See Fed. R. Civ. P. 23 advisory committee’s note to 2018 amendment (explaining that “[t]he goal of this amendment is not to displace” any of the factors historically considered in assessing settlement fairness, “but rather to focus the court and the lawyers on the core concerns of procedure and substance that should guide the decision whether to approve the proposal”). 26 CAMPBELL V. ST. JOHN

before the district court has formally certified a litigation class. See Hanlon, 150 F.3d at 1026 (explaining that when a case settles before class certification, “[t]he dangers of collusion between class counsel and the defendant, as well as the need for additional protections when the settlement is not negotiated by a court-designated class representative, weigh in favor of a more probing inquiry than may normally be required under Rule 23(e)”); see also Newberg on Class Actions § 13:13 (5th ed.) (explaining that earlier settlements can make it “more difficult to assess the strengths and weaknesses of the parties’ claims and defenses, to determine the appropriate definition of the class, and to consider how class members will actually benefit from the proposed settlement” (quoting Manual for Complex Litigation § 21.612 (4th ed.))). 11

As to the settlement’s substantive fairness, Objector does not seriously dispute the district court’s findings that several of the governing factors weigh in favor of approving this settlement—including that the case proceeded to nearly the close of discovery before settling, which was “much further than almost every other class action” for which the judge had overseen settlement “in the past 17 years”; that the settlement was “the result of four in-person, arms’-length mediations before two different mediators”; and that “highly experienced” class counsel advocated for the settlement. See Hanlon, 150 F.3d at 1026 (“the extent of discovery completed and the stage of the proceedings,” as well as “the experience and views of counsel,” are relevant to approval of settlement); Rodriguez, 563 F.3d at 965 (“We put a good

11 The district court did, as part of its approval of the settlement, certify a settlement-only class, but that class did not differ materially from the litigation class the court had already certified in its order granting in part the contested class certification motion. CAMPBELL V. ST. JOHN 27

deal of stock in the product of an arms-length, non-collusive, negotiated resolution.”).

Objector’s contentions focus instead on only a subset of the considerations that were relevant to the district court’s holistic assessment of the settlement’s fairness. First, relying primarily on our decision in Koby v. ARS National Services, Inc., 846 F.3d 1071 (9th Cir. 2017), Objector argues that the settlement should not have been approved because it provides absent class members with “worthless injunctive relief.” See id. at 1080–81. Second, relying primarily on our decision in Bluetooth, Objector argues that the settlement contains several “warning signs” that indicate “that class counsel have allowed pursuit of their own self- interests . . . to infect the negotiations.” See 654 F.3d at 947. For the reasons that follow, we reject each of these arguments.

1.

In Koby, we held that a magistrate judge had abused her discretion by approving a settlement, “for one primary reason: There [was] no evidence that the relief afforded by the settlement [had] any value to the class members, yet to obtain it they had to relinquish their right to seek damages in any other [similar] class action” against the same defendant. 846 F.3d at 1079. That “primary reason” had two components. First, the settlement at issue in Koby provided injunctive relief that was “of no real value” because the defendant did not have “to do anything it was not already doing,” plus, “[t]o make matters worse,” there was a clause in the settlement giving the defendant an option to “escape” the injunction in “the only scenario in which [it] might be tempted” to do something inconsistent with the injunction. Id. at 1080. Second, the lack of any value in what the class received meant that class members “could not fairly or 28 CAMPBELL V. ST. JOHN

reasonably be required to give up anything in return.” Id. Yet the settlement would have nevertheless required class members to “relinquish” something that “plainly” had at least “some value”: claims for damages in other class actions against the defendant, including in one such action that was already pending. Id. at 1080–81.

Objector’s argument that the settlement here is invalid because the class received only “worthless injunctive relief,” see id. at 1081, is premised on a view of the settlement’s value that we decline to adopt. The district court found that the settlement’s injunctive relief had value to absent class members. This finding was not clearly erroneous. The settlement requires, in relevant part, that Facebook make a plain English disclosure on its Help Center page that tells users that Facebook “use[s] tools to identify and store links shared in messages.” The settlement requires Facebook to display the relevant language on its Help Center page for a year. 12 In light of the nature of the claims here, a year-long requirement to make such a disclosure has value: it provides information to users about Facebook’s message monitoring practices, making it less likely that users will unwittingly

12 Objector argues that this disclosure is duplicative of the change Facebook had already made to the disclosure language in its Data Policy, and therefore that the Help Center disclosure has no marginal value. The district court did not clearly err by concluding otherwise. As the district court explained, the twenty-two-word Help Center disclosure provides “further relief to the class” beyond the Data Policy change because, in addition to being required to stay on display for a year, it “explain[s] Facebook’s policy regarding its use of data in messages in plain English, on a web page accessed by hundreds of thousands of Facebook users per year.” CAMPBELL V. ST. JOHN 29

divulge private information to Facebook or third parties in the course of using Facebook’s messaging platform. 13

Further, contrary to Objector’s apparent interpretation of Koby, the relief provided to the class cannot be assessed in a vacuum. Rather, the settlement’s benefits must be considered by comparison to what the class actually gave up by settling. See Protective Comm. for Indep. Stockholders of TMT Trailer Ferry, Inc. v. Anderson, 390 U.S. 414, 424– 25 (1968) (“Basic to [the] process [of evaluating settlements] . . . is the need to compare the terms of the compromise with the likely rewards of litigation.”); see also Hanlon, 150 F.3d at 1026 (relevant considerations include not only “the amount offered in settlement,” but also “the strength of the plaintiffs’ case,” i.e., what plaintiffs could expect from further litigation). Our holding in Koby was that the settlement was invalid because it gave the class nothing and yet required the class to give up something. 846 F.3d at 1080 (comparing what the settlement provided to class members—“nothing of value”—with what they were “required to give up . . . in return”—the “right to pursue damages claims against [the defendant] as part of a class action”).

Here, the class did not need to receive much for the settlement to be fair because the class gave up very little. The district court did not err to the extent it concluded that class members’ claims were weak enough that the class was fairly likely to end up receiving nothing at all had this

13 Although we hold that the district court did not clearly err in finding that the settlement had value for the class, this does not mean that, if we were reviewing the settlement ourselves in the first instance, we would necessarily conclude that the benefits to the class from this disclosure were particularly substantial. 30 CAMPBELL V. ST. JOHN

litigation proceeded further. As the district court stated, “any possible benefit to the class from continued litigation [was] both uncertain and insubstantial.” Damages claims had already been eliminated from the case, so the only forms of potentially available relief (absent a successful appeal from the damages portion of the class certification order) were declaratory or injunctive. To obtain such relief, Plaintiffs would have had to overcome many doctrinal hurdles. For example, they would have had to show that Facebook’s reading of messages sent on Facebook amounted to an unlawful “interception” or wiretap. This would include proving that Facebook had improperly read Plaintiffs’ messages while the messages were in “transit” as opposed to in “storage,” see Konop v. Hawaiian Airlines, Inc., 302 F.3d 868, 878 (9th Cir. 2002); Cal. Penal Code § 631(a), an issue on which even Plaintiffs acknowledged “uncertainty in the law.” Plaintiffs also would have had to contend with Facebook’s potential argument that users had consented to the challenged practices—a defense that similarly could have precluded liability under both ECPA and CIPA. See 18 U.S.C. § 2511(2)(d); Cal. Penal Code § 631(a). In addition, Plaintiffs would have had to overcome Facebook’s potential argument that the challenged activity occurred in “the ordinary course of its business,” 18 U.S.C. § 2510(5)(a), and therefore could not be the basis for liability under ECPA—an issue that Plaintiffs acknowledged turns on caselaw that is “not fully developed.”

We need not and do not resolve the merits of any of these issues. Collectively, however, the numerous challenges that Plaintiffs faced provided the parties and the district court ample support to conclude that Plaintiffs were ultimately likely to have lost this entire case. CAMPBELL V. ST. JOHN 31

In evaluating what class members relinquished in this settlement, we must also consider whether class members were required to release claims that were more meritorious than the theories Plaintiffs pursued in this litigation. As an initial matter, the settlement here expressly excludes any release of absent class members’ claims for damages— unlike the release in Koby. See 846 F.3d at 1080. The release here does include a release of claims for declaratory and injunctive relief, and it uses somewhat broad language in describing which declaratory and injunctive relief claims are barred. We do not, however, read the terms of the injunctive and declaratory release in isolation.

Under our precedent, the only claims that would be barred in a future case by this release are those that share an “identical factual predicate” with the claims advanced in this case. See Hesse v. Sprint Corp., 598 F.3d 581, 590–92 (9th Cir. 2010). Facebook’s counsel assured us at oral argument that the parties’ intent was to release only those claims that could be released under that precedent. See Oral Argument at 36:48–37:20 (asserting that, because of Hesse, “if there are claims in [a different] case that involve URL shares in messages, then yes, [those claims] would be covered by this [release]. But other items would not be.”). Facebook would be bound in any future litigation by this representation about the intended scope of the release, even if that litigation took place in a jurisdiction that lacked the “identical factual predicate” rule. See U.S. Cellular Inv. Co. of L.A., Inc. v. GTE Mobilnet, Inc., 281 F.3d 929, 934 (9th Cir. 2002) (“[T]he fundamental goal of contract interpretation is to give effect to the mutual intent of the parties as it existed at the time of contracting.”). Interpreting the release consistent with that rule, it does not extinguish claims lacking the weaknesses described above, and thus does not render the relief provided to the class inadequate. 32 CAMPBELL V. ST. JOHN

In sum, given how little the class could have expected to obtain if it had pursued claims further based on the facts alleged here (and, correspondingly, how little it gave up in the release), it was not unreasonable that the settlement gave the class something of modest value.

2.

Objector next argues that the settlement is invalid under our decision in Bluetooth because it prioritizes class counsel’s interests over those of their clients. In Bluetooth, we identified three “subtle” warning signs that may indicate that class counsel colluded with defense counsel to settle the case in a manner that elevated class counsel’s interests over those of the class, and that the settlement is therefore not “fair, reasonable, and adequate” under Rule 23(e)(2): (1) when counsel receive a disproportionate distribution of the settlement, or the class gets no monetary distribution but class counsel are well compensated; (2) when the parties negotiate “clear sailing” arrangements for the payment of attorney’s fees wherein the defendant agrees not to object to the fee application presented to the court; and (3) when the agreement includes a “reversion” or “kicker” provision under which any reduction in attorney’s fees reverts to the defendant rather than being added to the class fund. See 654 F.3d at 947. We explained in Bluetooth that “assessment of [a] settlement’s overall reasonableness must take into account the defendant’s overall willingness to pay,” id. at 949, keeping in mind that the defendant’s indifference as to how that payment is divvied up on the plaintiffs’ side could result in “a tradeoff between merits relief and CAMPBELL V. ST. JOHN 33

attorneys’ fees,” id. at 946 (quoting Evans v. Jeff D., 475 U.S. 717, 733 (1986)). 14

The district court looked for each of these warning signs and concluded that none weighed against approval of the settlement. As to disproportionality, the court explained that although “it is difficult to put a dollar figure on” the value of the non-monetary relief obtained by the class, “the privacy interests of the class vindicated by the settlement” were significant, and the non-monetary nature of the relief was “a function of [the district] court’s decision to certify only an injunctive relief class.” Meanwhile, class counsel, who had filed this action under provisions that permit a court to award attorney’s fees, had received a substantial “lodestar discount,” the result of which was a fee that the court separately deemed reasonable in its ruling on the fee application. As to the second and third warning signs, the district court reasoned that Bluetooth’s concerns with “clear sailing” and reversion of unawarded attorneys’ fees to Facebook were “inapplicable to this case because there is no common fund, ‘constructive’ or otherwise,” and the class was certified for injunctive relief only. The district court added that Bluetooth “only requires that the court carefully scrutinize the settlement for collusion,” and here there was none. Rather, “[t]he case was extremely hard-fought, and settled at an advanced procedural stage, after multiple mediations.”

The district court did not abuse its discretion in concluding that the settlement was not the result of collusion

14 The Bluetooth warning signs, which bear on a settlement’s fairness under Rule 23(e)(2), are distinct from the issue whether the district court awarded “reasonable” fees and costs under the standards of Rule 23(h)—which was not raised in this appeal. 34 CAMPBELL V. ST. JOHN

between class counsel and Facebook. As an initial matter, Bluetooth explained that the warning signs were a necessary addition to the Hanlon factors because the settlement at issue in Bluetooth had been “negotiated prior to formal class certification,” when “there is an even greater potential for a breach of fiduciary duty owed the class during settlement.” Bluetooth, 654 F.3d at 946. Bluetooth therefore left open a question no subsequent case has answered: whether district courts are required to look for these subtle warning signs in cases, like this one, that are settled after formal class certification. We need not resolve that threshold question because, assuming without deciding that courts must look for these warning signs in a post-certification settlement, we conclude that applying the Bluetooth framework does not demonstrate that the settlement in this case was unfair. Cf. In re Online DVD-Rental Antitrust Litig., 779 F.3d 934, 944 n.6 (9th Cir. 2015) (assuming, without deciding, that Bluetooth’s “heightened scrutiny” does not apply to post- certification settlement in a case where such scrutiny had not been applied by the district court and had not been raised by the parties on appeal).

Turning to the district court’s evaluation of the specific warning signs enumerated in Bluetooth, the district court’s analysis of the first Bluetooth factor, disproportionality, was reasonable. Objector’s contention distills to her position that any attorney’s fee award that exceeds “roughly 25% of the settlement value” is “disproportionate,” and that therefore the only way to “justify the nearly $4 million allocated for attorney’s fees” in this settlement is by appraising the overall settlement “at more than $15.5 million.” Although 25% of the anticipated settlement value is a useful benchmark to keep in mind in all cases, our caselaw affords district courts discretion to refrain from attempting to measure the unmeasurable. As the district court here explained, “it is CAMPBELL V. ST. JOHN 35

difficult to put a dollar figure on” the value of what the class obtained, but the court concluded the class did obtain relief that has meaningful value. In situations like this one, where “the benefit to the class” is not “easily quantified,” district courts have discretion to award fees based on how much time counsel spent and the value of that time (a lodestar calculation) without needing to “perform a ‘crosscheck’” in which they attempt to estimate how this compares to the recovery for the class. See Hyundai, 926 F.3d at 571 (quoting Bluetooth, 654 F.3d at 942). The district court reasonably exercised that discretion not to perform a crosscheck of the lodestar in this case, given the difficulty of measuring the value of the injunctive relief.

To be sure, in a case where the class primarily receives non-monetary relief, but class counsel obtain millions of dollars, it may be an abuse of discretion not to at least attempt to approximate the value of injunctive relief and use that valuation in an assessment of disproportionality. But in this case, the district court did not abuse its discretion by declining to do so, because of three key circumstances that are all present here. First, the district court had already declined to certify a damages class. When further litigation may result in substantial monetary relief to class members, the failure to include meaningful monetary relief in a settlement might be (but is not necessarily) a subtle sign that class counsel bargained away something valuable to benefit themselves. In such settlements, in order to show that nothing was unfairly bargained away by counsel, it may be necessary for settling parties to show why their non- monetary settlement is at least as good for the class as any monetary figure that would approximate what they could expect from further litigation. But where, as here, further litigation of the case being settled is extremely unlikely to result in a damages award, the assessment of whether 36 CAMPBELL V. ST. JOHN

counsel got more fees by bargaining away valuable relief for the class is more qualitative, and more context-specific. Second, damages were also not part of the class release, so to the extent further litigation might yield damages, absent class members were not prohibited from trying again to obtain such damages—further reducing the likelihood that class counsel bargained away any potentially valuable relief. Third, the district court was well-positioned to recognize, based on its years-long oversight of this litigation and its attendant understanding of how users interact with Facebook, the value of the injunctive relief that was made available to the class through the settlement. Relatedly, the district court was able to determine that class counsel had not, at the eleventh hour after litigating the case well beyond a hotly contested class certification motion and up to near the close of extensive discovery, abandoned the interests of the class in favor of their own. Taking these factors together, we conclude that the court did not abuse its discretion by concluding that the value of what the class received was reasonable not only in proportion to what the class gave up (as discussed above) but also in proportion to what class counsel received.

The second and third Bluetooth factors do not demonstrate that reversal is warranted on the basis that the settlement was unfair. As to the third factor, the district court did not err in its evaluation of Bluetooth’s concern with “reversion.” An injunctive-relief-only class settlement, by definition, has no fund into which any fees not awarded by the court could possibly revert. There is no blanket rule foreclosing parties from agreeing that the class will receive only injunctive relief.

As to the second factor, we disagree with the district court to the extent that it held that a defendant’s agreement CAMPBELL V. ST. JOHN 37

to allow “clear sailing” of class counsel’s fee application should never be considered in an injunctive-relief-only settlement. The concern with a “clear sailing” arrangement is that class counsel may have obtained too little for the class “in exchange for red-carpet treatment on fees.” Bluetooth, 654 F.3d at 947 (quoting Weinberger v. Great N. Nekoosa Corp., 925 F.2d 518, 524 (1st Cir. 1991)). There may be some cases in which courts should be concerned that class counsel gave up valuable injunctive relief in exchange for a defendant’s promise not to contest class counsel’s fee application. But we hold that any error in the district court’s discussion of this factor is harmless. No one factor is dispositive. We conclude that the evidence is insufficient to prove that the class would have gotten meaningfully more injunctive or declaratory relief if Facebook had merely been permitted to oppose class counsel’s fee application, which Facebook already knew would be requesting substantially less than what class counsel represented would fully compensate them.

Finally, we reject Objector’s argument that, in addition to the three warning signs in Bluetooth, this case presents a “fourth red flag” indicating a lawyer-driven deal: the fact that the parties did not want to provide any notice of the settlement to the class, beyond what was already publicly available. It does seem odd that the parties repeatedly emphasized the informational value of the settlement while simultaneously arguing that it was unnecessary to provide class members formal notice that this information exists and that, if they had been dissatisfied with the settlement terms, they could have objected. But this is insufficient to prove that the settlement was unfair. Indeed, the district court, 38 CAMPBELL V. ST. JOHN

before granting final approval, did require the parties to provide more notice than they had proposed giving. 15

IV.

For the foregoing reasons, we AFFIRM the district court’s approval of the settlement.

15 Although Objector argues that the district court should have done more to notify class members of the settlement, we do not understand her position to be that the case should be remanded solely for supplemental notice and a further opportunity for other class members to object. Whether the class notice was adequate in its own right—as opposed to whether the claimed failure to provide enough notice undermines the settlement’s fairness—is thus outside the scope of our review.