In Re: Google Inc. Cookie Plac v.

Aug 6, 2019

17-1480

Versions:934 F.3d 316

OPINION OF THE COURT

AMBRO, Circuit Judge

Cases with many plaintiffs, few to none of whom will sue solely for themselves because the costs far outweigh the benefits, frequently result in class actions under Federal Rule of Civil Procedure 23. Prerequisites are having (1) so many class members that joinder is impractical, (2) questions of law or fact that are common to the class, (3) one or more representatives whose claims or defenses are typical of those in the class, and (4) representatives who will fairly and adequately protect the interests of the class. Fed. R. Civ. P. 23(a). Besides these prerequisites, a class action must satisfy one of the subsections (1) through (3) in Rule 23(b). As pertinent here, Rule 23(b)(2) provides that a class action may be maintained if the requirements of Rule 23(a) are met and “the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.”

For the first time we review an order approving the settlement of a class action certified under Rule 23(b)(2) where the only benefit received by the class was the defendant‘s payment of a cy pres award to organizations the defendant approved.¹

The defendant, Google Inc., created a web browser “cookie” that tracks an internet user‘s data (think following the trail of cookie crumbs). For some Safari or Internet Explorer browser users, the cookie may have operated even if the user configured privacy settings to prevent it from tracking data. The class plaintiffs claim Google invaded users’ privacy under the California constitution and the state tort of intrusion upon seclusion (meaning the intrusion into a private place, conversation, or matter in a highly offensive manner). In a disputed Rule 23(b)(2) class settlement, Google has agreed to stop using the cookies for Safari browsers and to pay $5.5 million to cover class counsel‘s fees and costs, incentive awards for the named class representatives, and cy pres distributions, without directly compensating any class members. The six cy pres recipients are primarily data privacy organizations, and all must agree to use the funds to research and promote browser privacy. In addition, Google would obtain, among other things, a class-wide release of all class-member claims, including for money damages that did or could stem from the subject matter of the litigation.

The lone objector, Theodore H. Frank, challenges the District Court‘s certification of a settlement class and the terms of the approved settlement. He argues that the cy pres money properly belongs to the class as compensation. He asks us to vacate the settlement as unfair and require direct distributions to class members before resorting to cy pres awards. In the alternative, if direct distributions are truly infeasible, he asserts the class should not have been certified due to inadequate representation. Frank also challenges the parties’ choice of cy pres recipients because of their pre-existing relationships with Google and class counsel.

We believe that, in some Rule 23(b)(2) class actions, a cy pres-only settlement may properly be approved. But here the District Court‘s cursory certification and fairness analysis were insufficient for us to review its order certifying the class and approving the settlement. The settlement agreement‘s broad release of claims for money damages and its designation of cy pres recipients are particularly concerning. We thus vacate the order approving the settlement and remand for further proceedings consistent with this opinion.

I. Background

A. Prior Litigation

We outlined the facts underlying the class‘s claims in In re Google Inc. Cookie Placement Consumer Privacy Litigation, 806 F.3d 125, 130–34 (3d Cir. 2015). In brief, news broke in early 2012 that a Stanford graduate student had discovered Google‘s Doubleclick.net cookies were bypassing Safari and Internet Explorer privacy settings and tracking internet-user information. Google settled the resulting Federal Trade Commission and state attorneys general lawsuits, agreeing to cease the practice and to pay a combined $39.5 million in fines, though admitting no past acts or wrongdoing. Plaintiff internet users also filed claims against Google that were later consolidated into a putative class action. The class complaint alleged violations of federal privacy and fraud statutes, California unfair competition and privacy statutes, the California constitution‘s right to privacy, and that state‘s privacy tort law. It sought injunctive and monetary relief. Google moved to dismiss, and the District Court granted the motion in full. On appeal we affirmed the dismissal of all but the California constitutional and tort claims. See id. at 153.

B. Proposed Settlement

On remand, the parties began discovery. They then sought to avoid further litigation and began mediation before a former federal judge. With the help of the mediator, the parties agreed to a settlement (the “Settlement Agreement“) and simultaneously moved for certification of a Federal Rule of Civil Procedure 23(b)(2) class and approval of the settlement under Federal Rule of Civil Procedure 23(e). The latter states that “claims, issues, or defenses of a certified class . . . may be settled, voluntarily dismissed, or compromised only with the court‘s approval.” Fed. R. Civ. P. 23(e). It also lays out the procedures that apply to any such settlement, dismissal, or compromise. Id.

The proposed settlement defines the class as all persons in America who used Safari or Internet Explorer web browsers and “who visited a website from which Doubleclick.net . . . cookies were placed by the means alleged in the Complaint,” excluding individuals who had already obtained relief from Google or submitted “a valid and timely Request for Exclusion.” Settlement Agreement §§ 2.3, 2.5 (defining “Class” and “Class Member“). In sweeping language, the settlement would release all class member claims, including for damages, that did or could stem from, or relate to, the subject matter of the litigation. Settlement Agreement § 2.25 (“Released Claims“).

In exchange, Google would be required to assure it had “implemented systems configured to” abate or delete all third-party Google cookies that exist in Safari browsers. Settlement Agreement § 5.1 (“Assurance of Remediation“). And as noted, it would also pay $5.5 million, to be divided among the settlement administrator, class counsel, the named class representatives, and cy pres recipients. Settlement Agreement § 5.2 (“Settlement Fund“).

The Settlement Agreement requires both parties to agree to the cy pres recipients. The class must propose up to ten options, and Google may strike any for a non-arbitrary reason and request a replacement. Settlement Agreement § 5.3 (“Cy Pres Recipients“). The chosen recipients must agree “to devote the funds to promote public awareness and education, and/or to support research, development, and initiatives, related to the security and/or privacy of Internet browsers.” Id. The parties ultimately agreed on six recipients: (1) the Berkeley Center for Law & Technology; (2) the Berkman Center for Internet & Society at Harvard University; (3) the Center for Democracy & Technology (Privacy & Data Project); (4) Public Counsel; (5) the Privacy Rights Clearinghouse; and (6) the Center for Internet & Society at Stanford University. Neither Frank nor the District Court were privy to the selection process.

C. Class Action Settlements

The settlement of a putative class action requires the approval of a district court, both for certifying the class and for determining whether the settlement is “fair, reasonable and adequate.” Halley v. Honeywell Int‘l, Inc., 861 F.3d 481, 488 (3d Cir. 2017). We have provided guidance on the factors a court should consider in deciding whether to approve class action settlements, most notably the Girsh factors² and the Prudential factors.³ Where, as here, the parties seek

simultaneous class certification and settlement approval, courts should be “even more scrupulous than usual when they examine the fairness of the proposed settlement.” Prudential, 148 F.3d at 317 (quotation omitted). This heightened standard is designed to protect the interests of all class members. Id.

Some class action settlements also require the sending of notice to members of the putative class. Before approving the settlement of a class action certified under Rule 23(b)(3), which is primarily used to pursue money-damages class actions,⁴ a court “must direct to class members the best notice that is practicable under the circumstances, including individual notice to all members who can be identified through reasonable effort.” Fed. R. Civ. P. 23(c)(2)(B). For class actions certified under Rule 23(b)(1), which is often used to pursue class actions involving a common fund for injured stakeholders,⁵ or Rule 23(b)(2), which is used to pursue

injunctions in class actions, the Rules do not impose the same mandate of notice; rather, they state the court “may direct appropriate notice to the class.” Fed. R. Civ. P. 23(c)(2)(A); see also Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 363 (2011). Of course, regardless what these rules say, the procedures for class action settlement—including the notice procedures—must also comply with due process requirements. See In re Nat‘l Football League Players Concussion Injury Litig., 821 F.3d 410, 435 (3d Cir. 2016); see also Dukes, 564 U.S. at 362.

C. District Court Proceedings

The District Court preliminarily certified the class for settlement under Rule 23(b)(2). Following a notice period, the Court received fifty exclusion requests and one timely objection. Frank, the sole objector, asserted that either the settlement was unfair because it improperly preferred cy pres over direct compensation to class members, or, in the alternative, the class could not be certified if direct compensation was impossible. He further argued that the cy pres recipients raised a conflict-of-interest concern because class counsel was on one recipient‘s board and Google was a regular donor to four others. Finally, Frank challenged the attorneys’ fees—an issue he does not raise on appeal.

To determine whether the settlement class could be certified, the District Court conducted a hearing as required by Fed. R. Civ. P. 23(e)(2) and heard arguments from the parties and Frank. Following the hearing, it issued a short opinion and final order approving class certification and the settlement. In its opinion the Court correctly noted it could certify a class only if the requirements of Rule 23(a) and one of the prongs of Rule 23(b) were satisfied. It stated that “no objections had been filed” challenging class certification, then in three sentences explained why all the requisite factors were met—without stating what type of Rule 23(b) class was certified.

The Court then reviewed the proposed settlement for fairness, reasonableness, and adequacy, as required by Rule 23(e)(2). That Rule provides that if a class action settlement “would bind class members, the court may approve it only after a hearing and only on finding that it is fair, reasonable, and adequate.” Fed. R. Civ. P. 23(e)(2). The Court here noted that a presumption of fairness applies to class settlements that meet certain criteria established in our case law, but it did not make factual findings relating to those factors or hold whether the presumption applied in this case. It then applied the Girsh factors in determining whether the settlement was fair, reasonable, and adequate. Though it spent only four sentences analyzing the entirety of Girsh‘s nine factors, the Court concluded that “the record adequately establishes” they had been satisfied. It acknowledged additional factors to consider under Prudential, 148 F.3d at 323, and In re Baby Prods. Antitrust Litig., 708 F.3d 163 (3d Cir. 2013), but did not expressly apply them.⁶

Ultimately, the Court rejected Frank‘s arguments and approved the settlement. It found that “the nature of the likely compensation to the class members has always been complicated by the substantial problems of identifying the millions of potential class members⁷ and then of translating their alleged loss of privacy into individual cash amounts.” It further found that “payments to absent class members would be logistically burdensome, impractical, and economically infeasible, resulting (at best) with direct compensation of a de minim[i]s amount.” Because the cy pres money would be used by “preeminent institutions for researching and advocating for online privacy” to promote Internet browser privacy, it held that the cy pres awards “bear a direct and substantial nexus to the interests of absent class members.” In one sentence, the Court noted Frank‘s objections to the pre-existing relationships between Google, class counsel, and the cy pres recipients, but held “no conflict of interest” had “undermine[d] the selected cy pres recipients.”

Frank timely appealed.

II. Discussion

A. Article III Standing

After oral argument, we held this appeal in abeyance pending the Supreme Court‘s resolution of Frank v. Gaos, which also involved a cy pres-only class action settlement involving alleged privacy violations by Google. 139 S. Ct. 1041, 1043–44 (2019). But the Gaos Court did not review the class action settlement in that case. Instead, in a per curiam opinion it vacated the settlement and remanded under its decision in Spokeo, Inc v. Robins, in which it clarified the “concreteness” requirement of the injury-in-fact prong of Article III standing, holding that a “bare procedural violation” of the Fair Credit Reporting Act was not sufficiently “concrete” to establish standing. 136 S. Ct. 1540, 1549–50 (2016); Frank, 139 S. Ct. at 1046.

After Gaos was issued, we requested supplemental briefing from the parties on whether plaintiffs adequately alleged a “concrete injury” for purposes of Article III standing. They here adequately did so, following our decision in In re Nickelodeon Consumer Privacy Litigation, 827 F.3d 262, 273–74 (3d Cir. 2016). There, in the wake of Spokeo, we considered whether a concrete injury occurred when a website operator stated it would not track the browser history and video-viewing of children who visited Nick.com but then did so anyway. Id. at 269. We held the website operator‘s intrusion on the children‘s personal browsing and viewing activities was sufficiently injurious for purposes of Article III standing. Id. at 273–74. The allegations of standing here are indistinguishable from those in Nickelodeon: in both cases, the plaintiffs alleged that a third party (Viacom there, Google here) tracked their personal internet browsing information in violation of the third party‘s own promises not to do so. Id. The Nickelodeon decision, which considered and distinguished the Supreme Court‘s analysis in Spokeo, dictates that we recognize standing here.

More than precedent supports our conclusion. History and tradition reinforce 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. Privacy torts have become “well-ensconced in the fabric of American law.” In re Horizon Healthcare Servs. Inc. Data Breach Litig., 846 F.3d 625, 638 (3d Cir. 2017). Indeed, as Justice Thomas has explained, private actions to remedy intrusions on the private sphere trace back to England, where a property owner needed only to show that another person placed a foot on his property to establish a traditional case or controversy. See Spokeo, 136 S. Ct. at 1551 (Thomas, J., concurring) (citing Entick v. Carrington, 2 Wils. K.B. 275, 291 (1765)). Likewise, “Congress has long provided plaintiffs with the right to seek redress for unauthorized disclosures of information that, in Congress‘s judgment, ought to remain private.” Nickelodeon, 827 F.3d at 274. In an era when millions of Americans conduct their affairs increasingly through electronic devices, the assertion Google makes—that federal courts are powerless to provide a remedy when an internet company surreptitiously collects private data—is untenable. Nothing in Spokeo or any other Supreme Court decision suggests otherwise.

B. Jurisdiction and Standards of Review

The District Court had jurisdiction under 28 U.S.C. § 1332(d), and we have appellate jurisdiction under 28 U.S.C. § 1291. We review a district court‘s decision to approve a settlement and to certify a class for abuse of discretion. See Baby Prods., 708 F.3d at 175. “An appellate court may find an abuse of discretion where the district court‘s decision rests upon a clearly erroneous finding of fact, an errant conclusion of law or an improper application of law to fact.” Id. (citing Prudential, 148 F.3d at 299). However, a district court must “make its reasoning and application of the [law] clear, so that we, as a reviewing court, have a sufficient basis to review for abuse of discretion.” Gunter v. Ridgewood Energy Corp., 223 F.3d 190, 196 (3d Cir. 2000); see also In re Cendant Corp. PRIDES Litig., 234 F.3d 166, 168 (3d Cir. 2000).

C. Class Settlement Fairness

Frank raises two challenges to the proposed settlement‘s fairness, reasonableness, and adequacy under Rule 23(e)(2). First, he argues the settlement, whose main monetary component is a cy pres award, provides no marginal benefit to the class and therefore cannot satisfy that Rule. Indeed, he asserts a cy pres award is always inappropriate if some individual class members could be compensated through a claims or lottery process. Second, Frank challenges the selection of the cy pres recipients due to their pre-existing associations with either class counsel or Google.

We have identified two opposing interests a district court must weigh when reviewing motions for settlement-only class certification and approval of the settlement. First, we favor the parties reaching an amicable agreement and avoiding protracted litigation. See, e.g., In re Warfarin Sodium Antitrust Litig., 391 F.3d 516, 535 (3d Cir. 2004). We do not wish to intrude overly on the parties’ hard-fought bargain. See Ehrheart v. Verizon Wireless, 609 F.3d 590, 593–95 (3d Cir. 2010). A district court thus is to presume a settlement is fair if “(1) the negotiations occurred at arms length; (2) there was sufficient discovery; (3) the proponents of the settlement are experienced in similar litigation; and (4) only a small fraction of the class objected.”⁸ NFL Concussion Litig., 821 F.3d at 436.

At the same time, a district court has an obligation as a fiduciary for absent class members to examine the proposed settlement with care. See, e.g., Warfarin Sodium, 391 F.3d at 535. This duty is heightened in settlement-only class cases: “[W]here settlement negotiations precede class certification, and approval for settlement and certification are sought simultaneously, we require district courts to be even ‘more scrupulous than usual’ when examining the fairness of the proposed settlement.” Id.; see also Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 619–20 (1997) (viewing as “altogether proper” the Third Circuit‘s careful review of a settlement class‘s certification to ensure absent class members were adequately represented). In settlement negotiations, the interests of class representatives or class counsel do not always align with the interests of absent class members, and prior to certification class members may not know their claims are in litigation or that settlement negotiations are taking place. See In re Gen. Motors Corp. Pick-Up Truck Fuel Tank Prods. Liab. Litig., 55 F.3d 768, 788 (3d Cir. 1995). Thus a district court‘s “scrupulous” review of the settlement terms is “designed to ensure that class counsel has demonstrated ‘sustained advocacy’ throughout the course of the proceedings and has

protected the interests of all class members.” Prudential, 148 F.3d at 317 (quoting GM Trucks, 55 F.3d at 805–06).

These principles, combined with the legal analyses further described below, lead us to disagree with Frank that cy pres-only settlements are unfair per se under Rule 23(e)(2). In some cases a cy pres-only settlement may be proper. But we are troubled by the District Court‘s cursory certification and fairness analysis in this case, and two features of the settlement, discussed below, compel us to vacate and remand.

1. Cy Pres-Only Class Settlements

We have never addressed whether a class action settlement‘s monetary award may be given solely to cy pres recipients. In the usual cy pres case, money from a class settlement fund remains after distributions to class members, perhaps because “class members cannot be located, decline to file claims, have died, or the parties have overestimated the amount projected for distribution.” Baby Prods., 708 F.3d at 169. In those cases, parties may seek to “distribute to a nonparty (or nonparties) the excess settlement funds for their next best use—a charitable purpose reasonably approximating the interests pursued by the class.” Id. Cy pres is generally preferable to escheating funds to the state or reverting them to the defendant because it “preserve[s] the [class action‘s] deterrent effect . . . [and] (at least theoretically) more closely tailor[s] the distribution to the interests of class members.” Id. at 172.

Still, many federal courts, the media, academia, and even the Chief Justice of the United States view cy pres awards with skepticism.⁹ Among other things, they “present a

potential conflict of interest between class counsel and their clients” because they “may increase a settlement fund, and with it attorneys’ fees, without increasing the direct benefit to the class.” Baby Prods., 708 F.3d at 173. A settlement whose only monetary distributions are to class counsel, class representatives, and cy pres recipients, as in this case, presents the risk of a still greater misalignment of interests: the settlement clearly benefits the defendant (who obtains peace at a potentially reduced cost), class counsel (who are guaranteed payment in the settlement), and the named representatives (who are given an incentive award in the settlement), while any benefit to other class members is indirect and inconsequential monetarily.

Despite cy pres’ flaws, we held in Baby Products that “a district court does not abuse its discretion by approving a class action settlement agreement that includes a cy pres component directing distribution of excess settlement funds to a third party to be used for a purpose related to the class injury.” Id. at 172. Favorably citing the American Law Institute‘s (“ALI“) guidelines, we held that cy pres was appropriate where some class members were compensated directly and “further individual distributions are economically infeasible.” Id. at 173 (citing ALI, Principles of the Law of Aggregate Litigation § 3.07(b) (2010)). We left the door open for other permissible forms of cy pres, id., and nowhere addressed it in the context of a Rule 23(b)(2) class.

With this background, we now consider whether a cy pres-only settlement in a Rule 23(b)(2) class can ever satisfy the fairness requirements of Rule 23(e)(2), and, if so, whether

the District Court abused its discretion in approving this particular settlement as fair, reasonable, and adequate.

Frank‘s central argument on appeal is that cy pres awards should never be preferred over direct distributions to class members because the settlement fund properly “belongs” to individual class members as monetary compensation for their injuries. He proposes that the $5.5 million settlement fund is “sufficient to fund either a claims process or a lottery distribution for the class,” even though he acknowledges it would be “impossible to pay every class member.” Obj. Br. at 9. He criticizes the District Court‘s weighing of the settlement fund against the size of the entire class because “[e]ven in billion-dollar securities settlements where class members have suffered substantial losses, the parties do not know who each and every class member is and must rely upon class members to identify themselves and the size of their loss in a claims process.” Id. at 27–28.

We know no ruling that requires district courts to approve only settlements that provide for direct class distributions on these facts. Indeed, Baby Products suggested the opposite. We reasoned that ”cy pres distributions are most appropriate where further individual distributions are economically infeasible” and declined to prohibit cy pres distributions in other situations, including where even an initial distribution to some class members would be infeasible. 708 F.3d at 173. We also reaffirmed that settlement approval should be a practical inquiry rooted in the particular case‘s facts and procedural posture—as the Girsh and Prudential factors reflect. Id. at 173–74. Ultimately “[t]he role of a district court is not to determine whether the settlement is the fairest possible resolution . . . [, but rather to] determine whether the compromises reflected in the settlement . . . are fair, reasonable, and adequate when considered from the perspective of the class as a whole.” Id.

Google further argues that this settlement fund was never intended to compensate class members monetarily. Rather, it claims the fund enhances the settlement‘s deterrent effect by funding data privacy institutions that will work to prevent similar potential privacy invasions from occurring in the future. The class representatives similarly point out that a claims-made process or lottery geared toward individual distributions would “not come close to addressing the Internet privacy concerns of all class members in this case.” Class Br. at 23.

This rationale accords with the purpose of the Rule 23(b)(2) class structure. The Supreme Court has described this as a more “traditional” collective action because it assumes a single, “indivisible” injunctive or declaratory remedy against the defendant will provide relief to all class members equally. Dukes, 564 U.S. at 361–63. A (b)(2) class therefore does not involve individualized determinations of liability or damages, id.,¹⁰ or even require that individual class members be ascertainable, Shelton v. Bledsoe, 775 F.3d 554, 563 (3d Cir. 2015). Nor does certification require the “greater procedural protections” of (b)(3) classes, such as the opportunity for individual class members to receive notice of the action or the opportunity to opt out. See Dukes, 564 U.S. at 362. To certify a (b)(2) class, a court must simply find that “the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” Fed. R. Civ. P. 23(b)(2). In other words, the class must be “cohesive” such that the members “have strong

commonality of interests.” Gates v. Rohm & Haas Co., 655 F.3d 255, 263–64 (3d Cir. 2011).

Contrary to Frank‘s arguments, then, we see no reason why a cy pres-only (b)(2) settlement that satisfies Rule 23‘s certification and fairness requirements could not “belong” to the class as a whole, and not to individual class members as monetary compensation. Direct monetary distributions typically would not accomplish the purpose of a (b)(2) class. It is not an inherent abuse of discretion for a district court to allow a cy pres-only settlement rather than random compensation, and possibly even overcompensation through a lottery, of some (b)(2) class members to the exclusion of others.

2. Review of the District Court‘s Settlement Approval

This brings us to the Settlement Agreement. District courts are to assess the fairness, reasonableness, and adequacy of a cy pres award under Rule 23(e)(2) using “the same framework developed for assessing other aspects of class action settlements“: namely, applying the Girsh factors; applying the Prudential factors where applicable; and also considering “the degree of direct benefit provided to the class,” which may include “the number of individual awards compared to both the number of claims and the estimated number of class members, the size of the individual awards compared to claimants’ estimated damages, and the claims process used to determine individual awards.” Baby Prods., 708 F.3d at 174. “[C]y pres awards should generally represent a small percentage of total settlement funds” unless a district court finds “sufficient justification.” Id.

In its opinion approving the Settlement Agreement, the District Court ran through the Girsh factors and held that all nine were satisfied, but in a more perfunctory fashion than we are accustomed to reviewing on appeal. It also acknowledged the additional factors we suggested district courts consider in Prudential, 148 F.3d at 323, and Baby Products, 708 F.3d at 174, but did not explicitly apply them.

In this context, we are not persuaded the Court sufficiently assessed the fairness, reasonableness, and adequacy of the settlement. In particular, two features of the settlement present concerns that were not adequately considered by it: the broad class-wide release of claims for money damages, and selection of the specific cy pres recipients.

a. Class-wide Release

The Settlement Agreement purports to release all class member claims, including for damages, that did or could stem from, or relate to, the subject matter of the litigation. This raises a red flag. Before the District Court, class counsel acknowledged that the dismissal of all claims providing for statutory damages made class certification more difficult. See District Court Docket, ECF No. 163 (Motion for Preliminary Approval of Class Action Settlement) at 15. That premise, in turn, shaped the structure of both class certification and class notice: the parties sought to certify an injunction class under Rule 23(b)(2), not a damages class under Rule 23(b)(3), see District Court Docket, ECF No. 163, and thus avoided the heightened certification and notice requirements that apply to the latter. See Fed. R. Civ. P. 23(b)(3) (requiring predominance and superiority); Fed. R. Civ. P. 23(c)(2)(B) (requiring “the best notice that is practicable under the circumstances” for 23(b)(3) settlements).

Yet having sidestepped these requirements, Google and class counsel nonetheless obtained—for themselves anyway—the precise benefits that a Rule 23(b)(3) class gives to the defendant and class counsel: namely, a broad class-wide release of claims for money damages for the defendant, and a percentage-of-fund calculation of attorneys’ fees for class counsel. The District Court‘s failure to scrutinize this troubling aspect of the Settlement Agreement prevents us from reviewing its fairness, reasonableness, and adequacy. See Gunter, 223 F.3d at 196 (district court‘s reasoning and application of the law must be clear enough to provide basis for review). We also question, and leave to the District Court on remand, whether a defendant can ever obtain a class-wide release of claims for money damages in a Rule 23(b)(2) settlement, and if so, whether a release of that kind requires a heightened form of notice either under Rule 23(c)(2)(B) or due process tenets. Cf. NFL Concussion Litig., 821 F.3d at 435 (noting that class action notice must comply with due process); see also In re Payment Card Interchange Fee & Merchant Discount Antitrust Litig., 827 F.3d 223, 236 (2d Cir. 2016) (reversing district court‘s approval of Rule 23(b)(2) settlement with broad release); Koby v. ARS Nat‘l Servs., 846 F.3d 1071, 1079 (9th Cir. 2017) (same).

b. Cy Pres Recipients

The District Court held that “the proposed cy pres distributions are appropriately tailored and focused.” It found the selected cy pres recipients to be “among the preeminent institutions for researching and advocating for online privacy,” and that they would properly be required to use the funds for internet browser privacy initiatives. Id. Because “this case is about Google‘s alleged circumvention of Internet browser privacy settings,” it concluded, the proposed cy pres distributions thus “bear a direct and substantial nexus to the interests of absent class members.” The Court made no finding that the class‘s nationwide nature was reflected in the geographical scope of cy pres recipients’ work. See, e.g., Schwartz v. Dallas Cowboys Football Club, Ltd., 362 F. Supp. 2d 574, 576 (E.D. Pa. 2005) (“In [reviewing cy pres awards], the court should consider . . . the geographic scope of the case.“). However, the record shows they overlap; the promotion of better online privacy inherently has nationwide effects.

In sum, on the District Court‘s review of the record, the proposed cy pres awards would be used for a purpose directly and substantially related to the class‘s interests. The record also reflects that the geographic scope of the class corresponds to that of the cy pres recipients. The Court held the proposed cy pres awards were fair, reasonable, and adequate when viewed “from the perspective of the Settlement Class as a whole.” Although other aspects of its analysis were unusually brief, the Court did not abuse its discretion by approving this (b)(2) class settlement‘s cy pres structure.

Yet we are troubled here by the selection of the specific cy pres recipients. Frank challenges the selection of those recipients as unfair under Rule 23(e)(2) due to pre-existing associations between them and class counsel or Google. He asserts that Google has long ties to Stanford and is a regular donor and cy pres payor to the Berkeley Center for Law & Technology, the Berkman Center for Internet & Society at Harvard University, the Center for Internet & Society at Stanford University, and the Center for Democracy & Technology. One of the lawyers representing the class is also a board member of Public Counsel.

We have not previously addressed when a prior relationship between a cy pres recipient and one of the litigants in a class action undermines the proposed settlement‘s fairness. Regardless of the relevant standard, however, the District Court‘s treatment of the question warrants remand. It conducted no fact finding, either through additional filings or an evidentiary hearing, to determine the nature of the relationships between the cy pres recipients and Google or class counsel. In its opinion, the Court misstated the nature of Frank‘s objections, stating he “takes the position that any relationship between a party (and its counsel) and a proposed cy pres recipient automatically disqualifies the proposed cy pres recipient.” It then dismissed the issue in a single sentence with no analysis, holding that “no conflict of interest” had “undermine[d] the selected cy pres recipients.” This is not the “scrupulous” examination required of a court acting as a fiduciary for absent class members. See Warfarin Sodium, 391 F.3d at 534; see also Gunter, 223 F.3d at 196.

We further hold that, if challenged by an objector, a district court must review the selected cy pres recipients to determine whether they have a significant prior affiliation with any party, counsel, or the court. A settlement should not be approved if such a prior affiliation “would raise substantial questions . . . whether the selection of the recipient was made on the merits.” ALI, Principles of the Law of Aggregate Litigation § 3.07 cmt. b. The parties seeking settlement approval bear the burden of explaining to a court why the cy pres selection was fair, which may include describing the nature of any prior affiliations; what role, if any, each affiliation played in the cy pres selection process; whether other recipients were sincerely considered; and why these recipients are the proper choice. As always, “[w]here a court fears counsel is conflicted, it should subject the settlement to increased scrutiny.” Baby Prods., 708 F.3d at 173.

On remand, the District Court should consider whether these cy pres recipients have significant prior affiliations with Google, class counsel, or the Court, and, if so, whether the selection process failed to satisfy Rule 23(e)(2) by raising substantial questions whether the recipients were chosen on the merits. Where, as here, the only benefit to class members is a cy pres award, parties may also want to involve class members or a neutral participant in the selection of recipients to ward off any appearance of impropriety.

D. Class Certification

Frank asserts that, even if his settlement fairness challenge fails, reversal is still appropriate on class-certification grounds. If a settlement will not confer a meaningful benefit to absent class members, he argues it must fail to satisfy Rule 23(a)(4) and (g)(4). Those Rules provide that a class‘s representative plaintiffs and counsel, respectively, must “fairly and adequately” represent the interests of the class. Frank contends that a cy pres-only settlement in these circumstances confers no benefit on the class, which inherently shows that the class representative and counsel failed to represent the class fairly and adequately. For this reason, he claims a cy pres-only settlement class cannot be certified due to inadequate representation.

Frank‘s class certification challenge is not really an alternative ground for reversal. It is predicated on his challenge to the settlement. In light of our decision to remand, we leave to the District Court whether to reconsider Frank‘s contentions concerning the propriety of class certification in light of its further consideration and potential factfinding.

* * * * *

The vista view of this case is not pretty. According to the complaint, an internet behemoth with unprecedented tools for monitoring private conduct told millions of Americans it would not track their personal browser history, and then it did so anyway to profit from the data. Through the proposed class-action settlement, the purported wrongdoer promises to pay a couple million dollars to class counsel and make a cy pres contribution to organizations it was already donating to otherwise (at least one of which has an affiliation with class counsel). By seeking certification under Rule 23(b)(2), the defendant and class counsel avoid the additional safeguards that apply to Rule 23(b)(3) actions. One might think this would leave room for class members to pursue damages individually; yet that relief is foreclosed as well, as the settlement contains a nationwide release of claims for money damages that arose or could arise were there unauthorized snooping, presumably covering tens if not hundreds of millions of Americans. In this context, we believe the District Court‘s factfinding and legal analysis were insufficient for us to review its order certifying the class and approving the fairness, reasonableness, and adequacy of the settlement. We thus vacate and remand for further proceedings in accord with this opinion.

Notes

“The term ‘cy pres’ is derived from the Norman French expression cy pres comme possible, which means ‘as near as possible.‘” (quoting ). The doctrine “originated in trusts-and-estates law as a rule of construction used to preserve testamentary charitable gifts that otherwise would fail.” . “When it becomes impossible to carry out the charitable gift as the testator intended, the doctrine allows the next best use of the funds to satisfy the testator‘s intent as near as possible.” (quotation omitted).

The Girsh factors, which a court must apply, are: “(1) the complexity, expense and likely duration of the litigation; (2) the reaction of the class to the settlement; (3) the stage of the proceedings and the amount of discovery completed; (4) the risks of establishing liability; (5) the risks of establishing damages; (6) the risks of maintaining the class action through trial; (7) the ability of the defendants to withstand a greater judgment; (8) the range of reasonableness of the settlement fund in light of the best possible recovery; [and] (9) the range of reasonableness of the settlement fund to a possible recovery in light of all the attendant risks of litigation.” (quoting ) (internal ellipses omitted).

The Prudential factors, which a court may apply if relevant, are: (1) “the maturity of the underlying substantive issues . . . the development of scientific knowledge, the extent of discovery on the merits, and other factors that bear on the ability to assess the probable outcome of a trial on the merits of liability and individual damages“; (2) the “existence and probable outcome of claims by other classes and subclasses“; (3) “the comparison between the results achieved by the settlement for individual class or subclass members and the results achieved—or likely to be achieved—for other claimants“; (4) “whether class or subclass members are accorded the right to opt out of the settlement“; (5) “whether any provisions for attorneys’ fees are reasonable“; and (6) “whether the procedure for processing individual claims under the settlement is fair and reasonable.” .

Rule 23(b)(3) provides that a class action may be maintained if the requirements of Rule 23(a) are met and “the court finds that the questions of law or fact common to class members predominate over any questions affecting only individual members, and that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy.”

Rule 23(b)(1) provides that a class action may be maintained if the requirements of Rule 23(a) are met and prosecuting separate actions would create a risk of (1) “inconsistent or varying adjudications with respect to individual class members that would establish incompatible standards of conduct for the party opposing the class,” or (2) “adjudications with respect to individual class members that, as a practical matter, would be dispositive of the interests of the other members not parties to the individual adjudications or would substantially impair or impede their ability to protect their interests.”

Baby Products instructed courts to consider further “the degree of direct benefit provided to the class” from the settlement, potentially including “the number of individual awards compared to both the number of claims and the estimated number of class members, the size of the individual awards compared to claimants’ estimated damages, and the claims process used to determine individual awards.” .

In its brief, Google asserts that because many people download Doubleclick.net cookies while browsing the internet and there is no meaningful way to distinguish cookies downloaded “by the means alleged in the Complaint” rather than by other, unobjectionable means, it would be exceptionally difficult, and perhaps impossible, for anyone to determine who is a class member. Google Br. at 19. At the fairness hearing, Judge Robinson also stated she “believe[d] this has always been a case in which the facts preclude direct individual compensation.”

In its opinion the District Court recognized the fairness presumption‘s existence in the case law, but did not tie any factual findings to the test‘s factors or expressly hold whether the presumption applied. Its preliminary order certifying the class similarly suggested, but did not expressly hold, that the presumption applies. Frank does not challenge the application or misapplication of the fairness presumption, however. Given the case‘s facts and procedural posture, we assume the presumption applies.

Chief Justice Roberts has voiced concerns regarding cy pres, particularly where it constitutes a class‘s only monetary relief. See (Roberts, C.J., statement respecting denial of certiorari).

In fact, a non-settlement (b)(2) class cannot be certified if it brings claims for monetary relief that are more than “incidental.” .

Read the detailed case summary