Dissent by Judge Nguyen
OPINION
IKUTA, Circuit Judge:This appeal involves a nationwide class action settlement arising out of misstatements by defendants Hyundai Motor America, Inc. (Hyundai) and its affiliate, Kia Motors America, Inc. (Kia)1 regarding the fuel efficiency of their vehicles. The district court had jurisdiction under the Class Action Fairness Act (CAFA), 28 U.S.C. § 1332(d), because the matter in controversy exceeded $5,000,000, the putative class comprised at least 100 plaintiffs, and at least one plaintiff class member was a citizen of a state different from that of at least one defendant. We have jurisdiction pursuant to 28 U.S.C. § 1291. We hold that the district court abused its discretion in concluding that common questions predominate and certifying this settlement class under Rule 23(b)(3) of the Federal Rules of Civil Procedure, and we remand to the district court for further proceedings consistent with this opinion. Because the district court may still certify a class on remand, we briefly clarify some principles of attorneys’ fees awards in the class action context for the district court on remand.
,1
“The class action is an exception to the usual rule that litigation is conducted by and on behalf of the individual named parties only.” Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 348, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011) (internal quotation marks omitted). “To come within the exception, a party seeking to maintain a class action must affirmatively demonstrate his compliance with Rule 23.” Comcast Corp. v. Behrend, 569 U.S. 27, 33, 133 S.Ct. 1426, 185 L.Ed.2d 515 (2013) (internal quotation marks omitted). “Before certifying a class, the trial court must , conduct a rigorous analysis to determine whether the party seeking certification has met the prerequisites of Rule 23.” Zinser v. Accufix Research Inst., Inc., 253 F.3d 1180, 1186 (9th Cir. 2001) (internal quotation marks omitted). A district court’s certification “must be supported by sufficient findings to be afforded the traditional deference given to such a determination.” Molski v. Gleich, 318 F.3d 937, 946-47 (9th Cir. 2003) (internal quotation marks omitted). “When a district court, as here, certifies for class action settlement only, the moment of certification requires heightened attention[.]” Ortiz v. Fibreboard Corp., 527 U.S. 815, 848-49, 119 S.Ct. 2295, 144 L.Ed.2d 715 (1999) (internal quotation marks and citation omitted).
We review the district court’s decision to certify a class for an abuse of discretion. Parra v. Bashas’, Inc., 536 F.3d 975, 977 (9th Cir. 2008). A district court abuses its discretion when it makes an error of law or when its “application of the correct legal standard was (1) illogical, (2) implausible, or (3) without support in inferences that may be drawn from the facts in the record.” United States v. Hinkson, 585 F.3d 1247, 1262 (9th Cir. 2009) (en banc) (internal quotations omitted). “We reverse if the district court’s certification is premised on legal error.” Molski, 318 F.3d at 947.
Rule 23 “does not set forth a mere pleading standard.” Comcast, 569 U.S. at 33, 133 S.Ct. 1426. The plaintiff seeking class certification bears, the burden of demonstrating that all the requirements of Rule 23 have been met. See Zinser, 253 F.3d at 1188. This- requirement means that the plaintiff must first demonstrate through evidentiary proof that the class meets the prerequisites of Rule 23(a), which provides that class certification is proper only if: “(1) the class is so numerous that joinder of all members is'impracticable; (2) there are questions of law or fact common to the class; (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class; and (4) the representative parties will fairly and adequately protect the interests of. the class.” Fed. R. Civ. P. 23(a); see also Comcast, 569 U.S. at 33, 133 S.Ct. 1426. The Rule 23(a) prerequisites “effectively limit the class claims to those fairly encompassed by the named plaintiffs claims.” Dukes, 564 U.S. at 349, 131 S.Ct. 2541 (internal quotation marks omitted). To meet the commonality requirement of Rule 23(a)(2), the plaintiffs’ claims “must depend upon a common contention” that is “of such a nature that it is capable of classwide resolution—which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke.” Id. at 350, 131 S.Ct. 2541.
After carrying its burden of satisfying Rule 23(a)’s prerequisites, the plaintiff must establish that the class meets the prerequisites of at least one of the three types of class actions set forth in Rule 23(b). Fed. R. Civ. P. 23(b); Comcast, 569 U.S. at 33, 133 S.Ct. 1426. Here, the district court certified the class under Rule 23(b,)(3),'which provides that a class action may be maintained'only if “the court finds that the questions of law or fact common to class members predominate over any questions affecting only individual mem-' bers, and that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy,” and which lists a number of matters “pertinent to these findings.” Fed. R. Civ. P. 23(b)(3).2
The Rule 23(b)(3) predominance inquiry is “far more demanding” than Rule 23(a)’s commonality requirement. Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 624, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997). The “presence of commonality alone is not sufficient to fulfill Rule 23(b)(3).” Hanlon v. Chrysler Corp., 150 F.3d 1011, 1022 (9th Cir. 1998). Rather, a court has a “duty to take a close look at whether common questions predominate over individual ones,” and ensure that individual questions do not “overwhelm questions common to the class.” Comcast, 569 U.S. at 34, 133 S.Ct. 1426 (internal quotation marks omitted). In short, “[t]he main concern of .the predominance inquiry under Rule 23(b)(3) is the balance between individual and common issues.” Wang v. Chinese Daily News, Inc., 737 F.3d 538, 545-46 (9th Cir. 2013) (internal quotation marks omitted).
Where plaintiffs bring a nationwide class action under CAFA and invoke Rule 23(b)(3), a court must consider the impact of potentially, varying state laws, because “[i]n a multi-state, class action, variations in state law may swamp any common issues and defeat predominance.” Castano v. Am. Tobacco Co., 84 F.3d 734, 741 (5th Cir. 1996). “Variations in state law do not necessarily preclude a 23(b)(3) action.” Hanlon, 150 F.3d at 1022. For instance, even when some class members “possess slightly differing remedies based on state statute or common law,” there may still be “sufficient common issues to warrant a class action,” Id. ■ at 1022-23; see also Sullivan v. DB Investments, Inc., 667 F.3d 273, 301-02 (3d Cir. 2011) (discussing the “pragmatic response to certifications of common claims arising under varying state laws,” and ■ citing a case that affirmed “the district court’s decision to subsume the relatively minor differences in state law within a single class” as illustrative) (citing In re Prudential Ins. Co. of Am. Sales Practice Litig. Agent Actions, 148 F.3d 283, 315 (3d Cir. 1998)); In re Mex. Money Transfer Litig., 267 F.3d 743, 747 (7th Cir. 2001) (noting that even though “state laws may differ in ways that could prevent class treatment if they supplied-the principal theories of recovery,” class representatives in that case met the predominance requirement in part by limiting “their theories to federal law plus aspects of state law that are uniform”). On the other hand, where “the consumer-protection laws of the affected States vary in material ways, no common legal issues favor a class-action approach to resolving [a] dispute.” Pilgrim v. Universal Health Card, LUC, 660 F.3d 943, 947 (6th Cir. 2011).
In determining whether predominance is defeated by variations in state law, we proceed through several' steps.- See Mazza v. Am. Honda Motor Co., 666 F.3d 581, 590 (9th Cir. 2012). First, the class action proponent must establish that the forum state’s substantive law may be constitutionally applied to the claims of a nationwide class. Id. at 589-90.3 If the forum state’s law meets this requirement, the district court must use the forum state’s choice of law rules to determine whether the forum state’s law or the law of multiple states apply to the claims. Id. at 590. “[I]f the forum state’s choice-of-law rules require the application of only one state’s laws to the entire class, then the representation of multiple states within the class does not pose a barrier to class certification.” Johnson v. Nextel Commc’ns Inc., 780 F.3d 128, 141 (2d Cir. 2015). But if class claims “will require adjudication under the laws of multiple states,” Wash. Mut. Bank, FA v. Superior Court, 24 Cal. 4th 906, 922, 103 Cal.Rptr.2d 320, 15 P.3d 1071 (2001), then the court must determine whether common questions will predominate over individual issues and whether litigation of a nationwide class may be managed fairly and efficiently. Id. As with any other requirement of Rule 23, plaintiffs seeking class certification bear the burden of demonstrating through eviden-tiary proof that the laws of the affected states do not vary in material ways that preclude a finding that common legal issues predominate. See Castaño, 84 F.3d at 741 (indicating that class action proponents must show that variations in state laws will not affect predominance; '“[a] court cannot accept such an assertion on faith.”) (quoting Walsh v. Ford Motor Co., 807 F.2d 1000, 1016 (D.C. Cir. 1986) (Ruth Bader Ginsburg, J.)).
We undertook this predominance inquiry in Mazza v. American Honda Motor Co., which is closely analogous to our case. Mazza considered a car manufacturer’s challenge to a district court’s decision to certify a nationwide class of consumers claiming that Honda had misrepresented material information regarding Acura RLs. Honda contended that the district court erred in certifying this class under Rule 23(b)(3), because “California’s consumer protection statutes may not be applied to a nationwide class "with members in 44 jurisdictions,” Mazza, 666 F.3d at 589, and therefore plaintiffs had not demonstrated “that the questions of law or fact common to class members predominate over any questions affecting only individual members.” Fed. R. Civ. Pro. 23(b)(3).
Mazza addressed this argument by undertaking the following analysis. The plaintiffs first established that defendants had adequate contacts to the forum state, and therefore the court should apply California’s choice of law rules. 666 F.3d at 590.4 Under these rules, the foreign law proponent had the burden of showing that the law of multiple states, rather than California law, applied to class claims. Id. Mazza therefore walked through the three parts of California’s governmental interest test. First, we determined that Honda showed there were material differences between the plaintiffs’ California misrepresentation claims and the laws of other states. Id. at 591. Second, we determined that each of the 44 different states where the car sales took place “has a strong interest in applying its own consumer protection laws to those transactions.” Id. at 592. Turning to the third step of the test, we determined that “if California law were applied to the entire class, foreign states would be impaired in their ability to calibrate liability to foster commerce.” Id. at 593. Therefore, we held that “each class member’s consumer protection claim should be governed by the consumer protection laws of the jurisdiction in which the transaction took place.” Id. at 594.
Our conclusion that the plaintiffs’ class claims “will require adjudication under the laws of multiple states,” Wash. Mut. Bank, 24 Cal. 4th at 922, 103 Cal. Rptr.2d 320, 15 P.3d 1071, led to the next question: whether this conclusion defeated predominance. Although Mazza did not expressly address the predominance question, its vacatur of the district court’s class certification order established that plaintiffs had failed to show that common questions would predominate over individual issues.5
Because the Rule 23(b)(3) predominance inquiry focuses on “questions that preexist any settlement,” namely, “the legal or factual questions that qualify each class member’s case as a genuine controversy,” Amchem, 521 U.S. at 623, 117 S.Ct. 2231, a district court may not relax its “rigorous” predominance inquiry when it considers certification of a settlement class, Zinser, 253 F.3d at 1186. To be sure, when “[confronted with a request for settlement-only class certification, a district court need not inquire whether the case, if tried, would present intractable management problems, for the proposal is that there be no trial.” Amchem, 521 U.S. at 620, 117 S.Ct. 2231 (citation omitted). But “other specifications of the Rule—those designed to protect absentees by blocking unwarranted or overbroad class definitions—demand undiluted, even heightened, attention in the settlement context.” Id. “Heightened” attention is necessary in part because a court asked to certify a settlement class “will lack the opportunity, present when a case is litigated, to adjust the class, informed by the proceedings as they unfold.” Id. Indeed, in Amchem itself, the court determined that both factual differences among class members and differences in the state laws applicable to class members’ claims defeated predominance for a single nationwide settlement class. Id. at 624, 117 S.Ct. 2231.
A court may not justify its decision to certify a settlement class on the ground that the proposed settlement is fair to all putative class members.6 Indeed, federal courts “lack authority to substitute for Rule 23’s certification criteria a standard never adopted—that if a settlement is fair, then certification is proper.” Id. at 622,117 5.Ct. 2231; see also Ortiz, 527 U.S. at 849, 119 S.Ct. 2295 (holding that “a fairness hearing under Rule 23(e) is no substitute for rigorous adherence to those provisions of the Rule designed to protect absentees[.]”) (internal quotation marks omitted). This prohibition makes sense: “[i]f a common interest in a fair compromise could satisfy the predominance requirement. of Rule 23(b)(3), that vital prescription would be stripped of any meaning in the settlement context,” and the safeguards provided by the Rule, which “serve to inhibit appraisals of the chancellor’s foot kind-class certifications dependent upon the court’s, gestalt judgment or overarching impression of the settlement’s fairness,” would be eviscerated. Amchem, 521 U.S. at 621, 623, 117 S.Ct. 2231.
II
We now turn to the facts of this case. Under the Clean Air Act, all new vehicles sold-in the United States must be covered by an Environmental Protection Agency (EPA)- certificate of conformity demonstrating compliance with fuel efficiency and greenhouse gas emission standards. See 42 U.S.C. § 7522(a)(1). To obtain such a certificate, a vehicle manufacturer must submit an application to the EPA with information about the fuel efficiency for each model year. Id. § 7525(a)(1). In November 2011, a consumer advocacy group sent a letter to the EPA regarding complaints that Hyundai and Kia had overstated the fuel efficiency of a number of their vehicles and asked the EPA to audit the manufacturers. In response, the EPA initiated an investigation into. .-Hyundai’s and Kia’s fuel efficiency test procedures. About a year later, in November 2012, the EPA investigation confirmed that Hyundai .and Kia used improper test procedures to develop the fuel efficiency information submitted for certain 2011, 2012, and 2013 models.7 These improper procedures resulted in overstated fuel efficiency, estimates.
At the same time as the EPA announced its findings, Hyundai and Kia announced that they'would lower their fuel efficiency estimates for approximately 900,000 Hyundai and Kia vehicles from model years 2011, 2012, and 2013. At the same time, Hyundai and Kia announced the institution of a voluntary Lifetime Reimbursement Program (LRP) to compensate affected vehicle owners and lessees for the additional fuel costs they had incurred and would incur in the future as a result of the overstated fuel efficiency estimates. Under the LRP, anyone who owned or leased an affected Hyundai or Kia vehicle on or before November 2, 2012 was entitled to periodic reimbursements based .on the number of miles driven, the difference between the original and revised fuel efficiency estimate, and the average fuel price in the area where the car was driven, plus an extra 15 percent to account for the inconvenience caused by the overstated fuel efficiency estimates. In order to receive these benefits, class members could enroll in the LRP and then periodically visit a Hyundai or Kia dealership to verify their odometer readings.- Car owners could register for the LRP until December 31, 2013, although the program would continue for those who registered for as long as they owned or leased their vehicles.8
After the EPA commenced its investigation, but before announcing its results, a number of plaintiffs filed suit against Hyundai and Kia. In January 2012, plaintiffs filed a putative nationwide class action in state court in Los Angeles County. See Espinosa v. Hyundai Motor Am., No. BC 476445 (Cal. Super. Ct. filed Jan. 6, 2012). The complaint raised- claims under California’s consumer protection laws- and common law, alleging that Hyundai had falsely advertised that its 2011 and 2012 Elantra and Sonata vehicles got 40 miles per gallon (MPG) on the highway, when in fact these vehicles got far lower MPG.-9 The plaintiffs sought damages, rescission, restitution, and injunctive relief in the form of corrective advertising on behalf of a putative nationwide class, of owners of specified ve-hides who .purchased.or leased- their vehicles in the United States.
After Hyundai' removed the Espinosa action to federah court, see No. 2:12-cv-800 (C.D. Cal. filed Jan. 30, 2012), the plaintiffs moved for certification of a nationwide class. In its opposition to ’ class certification, Hyundai argued, among other things, that differences in state consumer protection laws precluded the application of California law to consumers who are not Californians and defeated predominance. Hyundai supported this argument with a thirty-four page “Appendix of Variations in State Laws,” which detailed the numerous differences in the burden of proof; liability, damages, statutes of limitations, and attorneys’ fees awards' under: different state consumer protection laws and common law fraud actions. Hyundai also argued that there were individual questions regarding whether each class member was exposed to or relied' on Hyundai’s advertising, and that these questions prevented class certification.10 "
In November 2012, the district court issued a tentative ruling on the motion for class certification in Espinosa. Plaintiffs sought to.certify two.classes, an Elantra class (including purchasers and lessees of 2011-12 mode) year Elantras) and a Sonata class (including all purchasers and lessees of 2011-2012 model year Sonatas). The court stated it would likely find that the plaintiffs denionstrated both the Rule 23(a) commonality and Rule 23(b)(3) predominance requirements were met as to statutory, but not common law claims. With respect to the question whether plaintiffs could show individualized reliance on advertising, the court stated that it would likely find that class-wide reliance on the challenged advertising could be presumed due to the “extensive sweep” of Hyundai’s marketing efforts.11
Turning to the question whether plaintiffs, could certify a nationwide class, despite the fact that their cornplaint invoked only California law, the district court held it was required to perform a choice of law analysis. The court stated that California had sufficient, contacts to support the,extraterritorial application of California law to all claims, but “just as in Mazza, the three-part choice of law test ... comes out in Defendant’s favor.” In reaching this conclusion, the court relied on three factors. First, Hyundai’s submission of its appendix of variations in state law “unquestionably demonstrates that there are material differences as between the various states’ laws that would ‘make a difference in this litigation.’” (quoting Mazza, 666 F.3d at 590-91, specifically considering the scien-ter requirements and remedies). -Second, the court ruled that as in Mazza, each of the states “has an interest in balancing the range of products and prices offered to consumers with the legal protections afforded to them.” Mazza, 666 F.3d at 592. Third, the court determined that “the interests of the other states would be more impaired were California law imposed upon their citizens than California would be impaired were this action limited to a class of only California consumers.” In sum, the court found “that certification of a nationwide class where California law is applied to out-of state consumers is foreclosed by the Ninth Circuit’s decision- in Mazza, a case virtually on all fours with the instant matter.”
Because California law could not be applied to out-of-state class members, the court thought it was obvious that the class could not be certified: “were the laws of the other various states applied to out-of-state purchasers, class certification would be precluded because common questions of law and fact would no longer predominate.” The court held that it would consider certifying a class- of California consumers, defined to include only those California consumers who actually viewed one of the challenged advertisements or marketing materials. On November 29, 2012, the Court held a hearing on the class certification motion pending in Espinosa, but did not- make a final ruling, instead requesting supplemental briefing.
Immediately following Hyundai’s November 2, 2012 announcement of the LRP, and before the Espinosa court could make a final ruling on class certification, plaintiffs across the country filed a flurry of putative class actions alleging that Hyundai and Kia misrepresented the fuel efficiency of their vehicles through advertising and Monroney Stickers.12 Among other actions, plaintiffs filed Hunter v. Hyundai Motor America, No. 8:12-CV-01909 (C.D. Cal. filed Nov. 2, 2012), and Brady v. Hyundai Motor America, No. 8:12-cv-1930 (C.D. Cal. filed Nov. 6, 2012), in the Central District of California. Both actions claimed violations of California consumer protection laws and common law on behalf of putative nationwide classes of all persons who owned or leased a Hyundai or Kia vehicle that had been identified in the EPA investigation.13 In December 2012, the district court requested further supplemental briefing on the class certification motion in light of Hyundai’s November 2, 2012 announcement.
Plaintiffs in one putative nationwide class action, see Krauth v. Hyundai Motor Am., No. 8:12-cv-01935 (C.D. Cal. filed Nov. 6, 2012), initiated proceedings before the Multidistrict Litigation (MDL) judicial panel pursuant to 28 U.S.C. § 1407, requesting that twelve putative class actions against Hyundai and Kia (including Espinosa, Hunter, and Brady) relating to the marketing and advertising of the fuel efficiency estimates of Hyundai and Kia vehicles be transferred to a single district for coordinated pretrial proceedings. On February 6, 2013, the MDL judicial panel transferred those actions as MDL No. 2424 to the court that was already presiding over the Espinosa action. The MDL judicial panel noted that any other related actions were potential tag-along actions.14 In total, 56 actions were ultimately transferred to the MDL.
One week after the MDL judicial panel issued its transfer order, and approximately three months after the announcement of the EPA investigation and LRP, the district court held a status conference in the Espinosa matter. At that status conference, the Espinosa plaintiffs informed the district court that they (along with the plaintiffs in Hunter and Brady) had reached a settlement with Hyundai for a single nationwide class. Shortly thereafter, the parties informed the court that Kia had agreed to the same settlement terms as Hyundai.
The proposed settlement agreement had the following terms. The parties agreed that the district court should certify a nationwide class of all persons who were current and former owners and lessees of specified Hyundai and Kia vehicles on or before November 2, 2012.15 Hyundai and Kia would offer class members several alternative methods of compensation. First, class members could opt to receive the cash equivalent of the pre-existing LRP program. Specifically, class members could choose to receive a single lump sum payment rather than the periodic payments offered through the preexisting LRP. The lump sum payment for current owners was calculated based on an average 4.75-year term of ownership, 15,000 miles driven each year, and gas prices between $3.00 and $3.70.16 The predicted average total lump sum payment was .$353 for class members' owning or leasing Hyundais and $667 for class members owning or leasing Kias. A class member who had begun participating in the LRP before the settlement but elected to switch to the lump sum payment option would receive a smaller lump sum, reduced by any amount the' class member had already received through- the LRP, The class members would receive their lump sum payment in the form of a debit card that would expire one year after it was issued; any unused amount would revert to Hyundai or Kia unless the class member timely deposited the residual amount in a. bank account.
Two other compensation options offered consumers a credit that was nominally larger than the lump sum value of the existing LRP program, but which could be used only for purchasing more services or products from Hyundai or Kia. First, class members could choose to receive a Hyundai or Kia dealer service credit worth 150 percent of the value of the- lump sum payment. The credit expired after two years. Alternatively, class members could choose to receive a new car rebate certificate worth 200 percent of the lump sum payment, which could be used toward the purchase of a new Hyundai or Kia vehicle. The certificate would expire after three years.
Finally, class members who were already participating in the preexisting LRP could choose to forego any of the settlement options and simply remain in the preexisting LRP. The deadline for enrolling in the LRP was extended to July 6, 2015, giving class members who had not enrolled in the LRP by the original December 31, 2013 deadline an additional 18 months to do so. Class members who were current owners or lessees of certain Hyundai vehicles who elected to remain in the LRP could receive an additional $100 for current original owners and $50 for current lessees and fleet owners.17
Used car owners were included in the proposed settlement class, but received only half the amounts available to new car owners. The settling parties justified this settlement amount on the ground. that used car owners’ “reliance on the Monro-ney numbers is less clear and potentially individualized” because Monroney stickers are not required for sales of used cars. See 15 U.S.C. §§ 1232-1233.
The proposed settlement provided a process for class members to opt out of the settlement by mailing a request for exclusion. However, upon the district court’s final approval of the settlement agreement, the district court would dismiss “all other lawsuits centralized in the MDL in W'hich the named plaintiffs in such lawsuit(s) did not timely exclude themselves from the settlement.”
In addition to paying the requisite amounts for class members, Hyundai and Kia-agreed to pay class counsel reasonable attorneys’ fees. The-amount of attorneys’fees would be negotiated and awarded separately from -the relief provided to class members. •
Following the February 2013 announcement of this proposed settlement, the court ordered discovery in April 2013 to confirm the facts on which the settlement was based and to allow plaintiffs to evaluate the terms of the settlement. Hyundai and Kia produced several hundred thousand pages of- documents and allowed plaintiffs to interview 11 employees.
While this confirmatory discovery was. ongoing, a different group of plaintiffs filed another action- against Hyundai in the Western District of Virginia. See Gentry v. Hyundai Motor Am., No. 3:13-cv-0030 (W.D. Va. filed Oct. 14, 2013). The Gentry plaintiffs asserted claims under Virginia consumer protection, false advertising, and vehicle warranty laws on behalf of a putative class of all persons who had purchased a model year 2011, 2012, or 2013 Hyundai Elantra in Virginia.18 Claiming that. Hyundai’s false advertising was willful, the complaint demanded the greater of treble damages or $1000 for each class member under the Virginia Consumer Protection Act. See Va. Code Ann. § 59.1-204(A). On November 6, 2013, the MDL judicial panel identified the Gentry action as a tag-along action, and transferred it to the Central District of California as part of MDL No. 2424.
In December 2013, after approximately eight months of confirmatory discovery, the Hunter, Brady, and Espinosa plaintiffs moved for class certification and preliminary approval, of the nationwide class settlement. According to these plaintiffs, confirmatory discovery had failed to reveal any evidence that Hyundai and Kia had engaged in deceptive conduct,' knowing concealment, or other--bad acts. In their motion for certification of a settlement class, the settling parties contended that common questions of Tact or law predominated under Rule 23(b)(3) with respect to California causes of action. .
The Gentry plaintiffs opposed class certification and sought ’ remand 'of their' action to the Western District of Virginia. In their memorandum opposing class certification filed May 2014, the Gentry plaintiffs argued that California choice of law rules did not allow certification of the class. The memorandum discussed elements of both the governmental interest test and the contractual choice-of-law provision. First, with respect to their contractual claim, the plaintiffs stated that the Virginia plaintiffs had purchased their vehicles by means of a contract with a Virginia choice of law provision and under California law, “an otherwise enforceable choice-of-law agreement may not be disregarded merely because it may hinder- the prosecution of a multistate or nationwide class action or result in the exclusion of nonresident consumers from a California-based class action.” Wash. Mut. Bank, 24 Cal. 4th at 918, 103 Cal.Rptr.2d 320, 15 P.3d 1071. Second, under the elements of California’s governmental interest test, the Gentry plaintiffs noted that “[njumerous courts have recognized that conflicts exist among State substantive laws” applicable to analogous' consumer fraud claims, and argued that there were “material and significant conflicts in the law of Virginia as compared to the law and remedy sought to be applied by Espinosa, Hunter, and Brady.”19 Moreover, the memorandum contended, not only were the Virginia causes of action “materially different from those asserted by the Settling Plaintiffs,” but Virginia also had a strong interest in having its law apply. Accordingly, even without the contractual choice-of-law provisions, the Gentry plaintiffs argued, California law would require courts to apply Virginia law.20 Three sets of plaintiffs, including the Gentry plaintiffs, also filed objections to the terms of the settlement.21
In June 2014, the district court circulated a tentative ruling granting the plaintiffs’ motion for certification of the settlement class. The court acknowledged that it “would need to engage in an extensive choice of law analysis” if the case were going to trial. Nevertheless, the court thought such an analysis was not warranted in the settlement context, because notwithstanding the Gentry plaintiffs’ objections to class certification “on the grounds that Virginia law provides a materially different remedy to Virginia consumers” for certain claims, state law variations were less of a concern and could be addressed as part of the final fairness hearing under Rule 23(e). Accordingly, the district court declined to apply California’s choice of law rules to determine whether California law was applicable to the class, or to make any choice of law ruling, and instead held that even if “substantial differences in state law are brought to light at the final fairness hearing, those issues do not prevent the Court from certifying the class for settlement purposes.” The court adhered to this position in its subsequent rulings.
In August 2014, the court granted class certification without ever addressing variations in state law.22 At the same time, the district court granted preliminary approval of the proposed settlement, finding it sufficiently fair, reasonable, and adequate to merit disseminating notice of the settlement to the class. The court noted the settling parties’ agreement that an aggregate amount of $210 million represented the total lump sum compensation that would be available to class members.
In December 2014, counsel for the Espi-nosa, Hunter, and Brady plaintiffs, as well as counsel for plaintiffs in other actions that had been transferred to the district court, filed applications for attorneys’ fees. Through a series of hearings beginning in March 2015, the district court approved $2,700,000 in attorneys’ fees to class co-counsel who represented the plaintiffs in the Hunter and Brady cases, $2,850,000 in attorneys’ fees to class co-counsel who represented the plaintiffs in the Espinosa case, and collectively over $3 million to counsel for other plaintiffs. In calculating attorneys’ fees, the district court began with the lodestar method (multiplying the number of hours the prevailing party reasonably expended on the litigation by a reasonable hourly rate). The court then determined that the Hunter and Brady counsel were entitled to a lodestar enhancement in light of the complexity and volume of work and the amount of the settlement, and multiplied the lodestar amount by a 1.22 multiplier. The district court also determined that the Espinosa counsel was entitled to a lodestar enhancement- due to the risk of filing a lawsuit before the November 2, 2012 EPA announcement, and multiplied the lodestar amount by a 1.5521 multiplier. In total, the district court awarded approximately $9 million in attorneys’ fees and costs.
In March 2015, the Hunter, Brady, and Espinosa plaintiffs, along with Hyundai and Kia, jointly moved for final approval of class settlement. In support of this motion, Hyundai and Kia submitted declarations reporting on response rates of class members. The reports established that approximately 21 percent of class members had filed claims for some $44,000,000 in total value. Of the class members filing claims, more than two-thirds began participating in the LRP before the settlement. Therefore, the portion of the class filing new claims accounted for only a small fraction of the $44 million in total value.
In June 2015, the district court gave its final approval of the class settlement. The court reaffirmed its prior conclusion that the certification of the class for settlement was proper under Rule 23(b)(3) and that the settlement was fair, relying in part on its August 2014 finding that the settlement would provide an estimated $210 million to the class. In rejecting objections that the proposed attorneys’ fees awards were excessive and not in proportion to the benefit conferred on the class, the district court noted that the attorneys’ fees did not impact class recovery because they were awarded separately, and so the issue of collusion did not arise. Further, the court stated that the fees were in most cases less than the amount requested by counsel. Finally, the court dismissed all lawsuits in MDL No. 2424 except for those in which the named plaintiffs had timely excluded themselves from the settlement.
Ill
Objectors now bring five consolidated appeals raising challenges to class certification, approval of the settlement as fair and adequate, and approval of attorneys’ fees as reasonable in proportion to the benefit conferred on the class.
A
We first address objectors’ arguments that the district court abused its discretion by failing to conduct a choice of law analysis or rigorously analyze potential differences in state consumer protection laws before certifying a single nationwide settlement class-under Rule 23(b)(3). As explained in Mazza, the district court was required to apply California’s choice of law rules to determine whether California law could apply to all plaintiffs in the nationwide class, or whether the court had to apply the law of each state, and if so, whether variations in state law defeated predominance. 666 F.3d at 588-89. Under California’s choice of law rules, this required the district court to apply the California governmental, interest test. Id. at 590. Theré is no dispute that the district court did not do so. The parties acknowledge that the district court did not conduct a choice of law analysis, and did not apply California law or the law of any particular state in deciding to certify the class.for settlement.
In failing to apply California choice of law rules, the district court committed a legal error. “A federal court sitting in diversity must look to the forum state’s choice of law rules to determine the controlling substantive law.” Id. (quoting Zinser, 253, F.3d at 1187). The district court made a further error by failing to acknowledge, as it had in its tentative ruling, that Hyundai and the Gentry plaintiffs submitted evidence that the laws in various states were materially different than those in California, and that these variations prevented thé court from applying only California law.' Finally, the court erred by failing to make a final ruling as to whether the material variations in state law defeated predominance under Rule 23(b)(3). Because “variations in state law may swamp any common issues and defeat predominance,” Castano, 84 F.3d at 741, a court must analyze whether “the consumer-protection laws of the affected States vary in material ways,” Pilgrim, 660 F.3d at 947, even if the court ultimately determines that “the common, aggregation-enabling, issues in the case are more prevalent or important than the non-common, aggregation-defeating, individual issues,” Tyson Foods, Inc. v. Bouaphakeo, — U.S. —, 136 S.Ct. 1036, 1045, 194 L.Ed.2d 124 (2016) (citation omitted).
The district court’s reasoning that the settlement context relieved it of its obligation to undertake a choice of law analysis and to ensure that a class meets all of the prerequisites of Rule 23, is wrong as a matter of law. While the district court was correct that it need not consider litigation management issues in determining whether to certify a class, the Rule 23(b)(3) predominance inquiry focuses on whether common questions outweigh individual questions, an issue that preexists any settlement. Amchem, 521 U.S. at 623, 117 S.Ct. 2231. Therefore, factors such as whether the named plaintiffs were in favor of the settlement or whether other class members had an opportunity to opt out are irrelevant to the determination whether a class can be certified.
If anything, this case highlights the reasons underlying Amchem’s warning that district courts must give “undiluted, even heightened, attention in the settlement context,” Amchem, 521 U.S. at 620, 117 S.Ct. 2231, to scrutinize proposed settlement classes.23 Because the district court made clear that it would be unlikely to certify the same class for litigation purposes, the class representatives were well aware that they would be unlikely to succeed in any efforts to certify a nationwide litigation class. Thus, by “permitting class designation despite the impossibility of litigation, both class counsel and court [were] disarmed.” Id. at 621, 117 S.Ct. 2231. Hyundai and Kia knew that there was little risk that they would face a nationwide litigation class action if they did not reach a settlément agreement. Accordingly, “[c]lass counsel confined to settlement negotiations could not use the threat of litigation to press for a better offer, and the court [faced] a bargain proffered for its approval without benefit of adversarial investigation.” Id. (citation omitted). ■
Finally, the district court erred in holding that it could avoid considering the potential applicability of the laws of multiple states on the ground that' the proposed settlement was fair. “[A] fairness hearing under Rule 23(e) is no substitute for rigorous adherence to those provisions of the Rule designed to protect absentees[.]” Ortiz, 627 U.S. at 849, 119 S.Ct. 2296.
Because the district court erred in certifying a settlement class, we must vacate the class certification. This does not mean that the court is foreclosed from certifying a class (or subclasses) on remand. We make no ruling on this issue, and merely note that Mazza determined that no such class was possible in a closely analogous case.
B
Even if the district court had restricted the class to California consumers (as the court indicated it would do in its tentative ruling in Espinosa), we would still have to consider the objectors’ argument that the ■ district court abused its discretion-.in certifying, a settlement class under Rule 23(b)(3) that includes used car owners- without analyzing whether these class members were exposed to, and therefore could have relied on Hyundai’s and Kia’s misleading statements. According to the objectors, individual questions of reliance preclude the inclusion of used car owners in this class.
In Mazza, we provided guidance on how a district court should determine whether a court can presume that class members relied on misleading advertising; On the one hand, we explained, “[a]n inference of classwide reliance cannot be made where there is no evidence that the allegedly false representations were uniformly made to all members of the proposed class.” Mazza, 666 F.3d at 695 (quoting Davis-Miller v. Automobile Club of Southern California, 201 Cal. App. 4th 106, 125, 134 Cal.Rptr.3d 551 (2011)). Rather, the class proponent must establish that the scope of the advertising makes it reasonable to assume that all class members were exposed to the allegedly misleading advertisements. Id. On the other hand, we noted the California Supreme Court’s exception to this general rule in In re Tobacco II Cases. Tobacco II presumed that class members had relied on a pervasive advertising campaign for cigarettes, extending over 40 years by 11 different defendants, which “misled the smoking public of the health risks and addictive nature of smoking and targeted the putative class uniformly in an alleged class-wide effort to seduce and induce people to smoke.” 46 Cal. 4th 298, 309, 327-28, 93 Cal.Rptr.3d 559, 207 P.3d 20 (2009) (Tobacco II). Distinguishing Tobacco II, Mazza explained that "in the context of a decades-long tobacco advertising campaign where there was little doubt that almost every class member had been exposed to defendants’ misleading statements,” class members did not need to demonstrate individualized reliance. 666 F.3d at 596. Harmonizing these rules, Mazza concluded that “[i]n the absence of the kind of massive advertising campaign at issue in Tobacco II, the relevant class must be defined in such a way as to include only members who were exposed to advertising that is alleged to be materially misleading.” /d24 We held that the defendant’s scope of advertising in Mazza did not rise to that level, and therefore an individualized case had to be made for each member showing reliance. Id. For this reason, we held that common questions of fact did not predominate where the class included members who were not exposed to the false advertising or who purchased products after learning of the misrepresentations, and therefore it was an error to certify the class. Id.
The district court addressed the question whether class members could have relied on Hyundai’s and Kia’s misleading statements in its June 14, 2014 ruling, and concluded that it could presume that all class members relied on the misleading statements because “misrepresentations were uniformly made to all consumers by virtue of Monroney stickers and nationwide advertising.” In reaching this conclusion, the district court failed to reference any evidence in the record regarding the extent of the advertising campaign for the 41 different Hyundai models and 35 different Kia models from 2011 to 2013; nor did it provide any reasoning regarding how this advertising reached the level of the cigarette advertising campaign (extending over 40 years by 11 defendants) discussed in Tobacco II.25 Furthermore, the district court’s ruling is based on a factual error, because there is no requirement that Mon-roney stickers be provided to purchasers of used cars, and there is no evidence in the record that used car owners were uniformly exposed to such stickers. In fact, the settlement itself relied on this difference in exposure to misleading information in awarding used ear owners only half the amounts awarded to new car owners. See supra, at p. 698. Nor can we conclude that this error was harmless because exposure to the defendant’s advertising can be presumed. The settling parties have not identified any evidence in the record of this sort of massive advertising campaign that could give rise to such a presumption with respect to used car owners.
The settling parties argue that even if there are individualized questions regarding exposure to the nationwide advertising, these questions do not predominate in the settlement context, where there is no manageability concern. This argument is contrary to Amchem, where the Court held that factual differences among class members, such as the ways that class members were exposed to asbestos and the length of those exposures, translated into significant legal differences, thereby defeating predominance for a settlement class. 521 U.S. at 624, 117 S.Ct. 2231. Similarly here, factual differences regarding used car owners’ exposure to the misleading statements translate into significant legal differences regarding the viability of these class members’ claims.
In sum, because the record does not support the presumption that used car owners were exposed to and relied on misleading advertising, the district court had an obligation to define the relevant class “in such a way as to include only members who were exposed to advertising that is alleged to be materially misleading.” Mazza, 666 F.3d at 596. The district court erred by failing to do so here.
C
Because a court’s obligations under Rule 23 are heightened in the settlement-class context, Amchem, 521 U.S. at 620, 117 S.Ct. 2231, a district court’s obligation to conduct a “rigorous analysis” to ensure that the prerequisites of Rule 23 have been met, Comcast, 569 U.S. at 33, 133 S.Ct. 1426, is heightened as well. Here, the district court failed to conduct a rigorous inquiry into whether the proposed class could meet the Rule 23 prerequisites on the mistaken assumption that the standard for certification was lessened in the settlement context. Because our precedent raises grave concerns about the viability of a nationwide class in this context, see Mazza, 666 F.3d at 596-97, this certification decision cannot stand.
IV
Because the district court may yet determine, after a rigorous Rule 23 analysis, that it may certify a settlement class and approve a settlement, we briefly clarify some principles of attorneys’ fee approval for the district court on remand. See, e.g., In re Gen, Motors Corp. Pick-Up Truck Fuel Tank Prods. Liab. Litig., 55 F.3d 768, 820-22 (3d Cir. 1995). When awarding attorneys’ fees in a class action, the district court has “an independent obligation to ensure that the award, like the settlement itself, is reasonable, even if the parties have already agreed to an amount.” In re Bluetooth Headset Prods. Liab. Litig., 654 F.3d 935, 941 (9th Cir. 2011). Therefore, we have “encouraged courts to guard against an unreasonable result by cross-checking their calculations against a second method.” Id. at 944. “In this circuit, there are two primary methods to calculate attorneys fees: the lodestar method and the percentage-of-reeovery method.” In re Online DVD-Rental Antitrust Litig., 779 F.3d 934, 949 (9th Cir. 2015). “Under the percentage-of-recovery method, the attorneys’ fees equal some percentage of the common settlement fund; in this circuit, the benchmark percentage is 25%.” Id. (citing Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 942); see also Hanlon, 150 F.3d at 1029. If the district court employs the lodestar method, but calculates an award that “overcompensates the attorneys according to the 25% benchmark standard, then a second look to evaluate the reasonableness of the hours worked and rates claimed is appropriate.” In re Coordinated Pretrial Proceedings in Petroleum Prod. Antitrust Litig., 109 F.3d 602, 607 (9th Cir. 1997). When a district court fails to conduct a “comparison -between the settlement’s attorneys’ fees award and the benefit to the class or degree of success in the litigation” or a “comparison between the lodestar amount and a reasonable percentage award,” we may remand the case to the district court for further consideration. Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 943; see also In re HP Inkjet Printer Litig., 716 F.3d 1173, 1190 (9th Cir. 2013). Indeed, in the absence of an adequate explanation of whether the award is proportionate to the benefit obtained for the class, “we have no choice but to remand the case , to the district court to permit it to make the necessary calculations and provide the necessary explanations.” McCown v. City of Fontana, 565 F.3d 1097, 1102 (9th Cir. 2009).
Here, the district court used the lodestar method to calculate attorneys’ fees, awarding approximately $9 million in attorneys’ fees and costs. However, the court failed to calculate the value of the settlement in order to ensure that the attorneys’ fees were not excessive in proportion to the settlement value. Although the court mentioned that the settling parties had earlier estimated the value of the proposed settlement at $210 million, it did not make a finding regarding the actual value of the settlement based on claims made, and the claims data in the record indicates that the amount of settlement funds claimed by class members was far lower.26 Moreover, the court failed to address objectors’ reasonable questions about the value of the settlement, for example, whether the value for class members who began participating in the LRP before the settlement, and who elected to remain in the LRP or who switched from the LRP to the lump sum option, could be attributed to the attorneys’ efforts in this litigation.27 Because the district court could not compare the fees award to the settlement value without considering these questions and determining the actual, settlement value, it failed “to assure itself— and us—that the amount awarded was not unreasonably, excessive in light of the results achieved.” Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 943.
A district court must also provide adequate justification for the use of a multiplier, which is appropriate in only “rare” or “exceptional” cases. See Perdue v. Kenny A. ex rel. Winn, 559 U.S. 542, 554, 130 S.Ct, 1662, 176 L.Ed.2d 494 (2010). Here, the district court’s reasoning for enhancing the lodestar amount. by a multiplier for class counsel, namely that the Hunter and Brady multiplier was warranted by the “complexity and volume of work that counsel engaged in,” and that the Espinosa multiplier was warranted by the risk that Espinosa counsel assumed by filing a lawsuit before the announcement of the LRP, is insufficient to -explain why an enhancement is warranted, particularly given objectors’ concerns that the settlement confers only modest benefits to the class, see Bluetooth Headset Prod. Liab. Litig., 654 F.3d at 942 (holding that district courts should “award only that amount of fees that is reasonable in relation to the results obtained,” even where counting all hours reasonably spent would produce a larger fees award) (quoting Hensley v. Eckerhart, 461 U.S. 424, 440, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983)).28
On remand, if the district court properly approves class certification and a settlement, the district court must determine what value was created by the settlement and take a closér look at the reasonableness of the attorneys’ fees in light of the results achieved.29
V
We conclude that the district court abused its discretion in certifying a nationwide settlement class without’conducting a rigorous predominance - analysis under Rule 23(b)(3) to determine whether variations in state consumer protection laws, or individual factual questions regarding exposure to the misleading statements, precluded certification.30 We vacate class, certification and remand to the district court for further proceedings consistent with this opinion. Each party will bear its own costs on appeal.
VACATED AND REMANDED.
. Rule 23(b) provides, in relevant part:
A class action may be maintained if Rule 23(a) is satisfied and if: ...
(3) the court finds that the questions. of law or fact common to class members predominate over any questions affecting only individual membe.rs, and, that a class action is superior .to other .available methods for fairly and efficiently adjudicating the controversy.. The matters pertinent to these findings include: . '
(A) the class members’ interests in individually controlling the prosecution or defense of separate actions;
(B) the extent and nature of any litigation concerning the controversy already begun by or against class members;
(C) the desirability .or undesirability of concentrating the litigation of the claims in the particular forum; and
(D) the likely difficulties in managing a class action.
. The Supreme Court has explained that in order to apply the forum state’s law to out-of-state defendants, the state must have a "significant contact or significant aggregation of contacts” to the claims asserted by each member of the plaintiff class. Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 822, 105 S.Ct. 2965, 86 L.Ed.2d 628 (1985). There is no dispute that California has significant contacts with the defendants in this case.
. Under California choice of law rules, there are "two different analyses for selecting which law should be applied in an action”: one considering a contractual choice-of-law provision, and the other requiring an application of the governmental interest test. See Wash. Mut. Bank, 24 Cal. 4th at 914-15, 103 Cal.Rptr.2d 320, 15 P.3d 1071. The governmental interest test has three steps. "Under the first step of the governmental interest approach, the foreign law proponent must identify the applicable rule of law in each potentially concerned state and must show it materially differs from the law of California.” Id. at 920, 103 Cal.Rptr.2d 320, 15 P.3d 1071. If "the trial court finds the laws are materially different, it must proceed to the second step and determine what interest, if any, each state has in having its own law applied to the case.” Id. If "each state has an interest in having its own law applied, thus reflecting an actual conflict” the court "must select the law of the state whose interests would be more impaired if its law were not applied.” Id..; see also Mazza, 666 F.3d at 589-90.
. California takes the same approach in applying a choice-of-law analysis to class claims. Under California law, if the court concludes "that class claims will require adjudication under the laws of multiple states,” then " 'the court must determine "whether common questions will predominate over individual issues and whether litigation of a nationwide class may be managed fairly and efficiently.’ ” Wash. Mut. Bank, 24 Cal. 4th at 922, 103 Cal.Rptr.2d 320, 15 P.3d 1071. In California, “ 'the class action proponent bears the burden of establishing the propriety of class certification.' ” Id.
. A court must make such a fairness finding under Rule 23(e) of the Federal, Rules of Civil Procedure, which prohibits a court from approving a settlement unless it concludes that "it is fair, reasonable, and adequate,” Fed. R. Civ. P. 23(e)(2).
. According to the EPA, the improper procedures included selecting results from test rims that were aided by a tailwind, selecting only favorable results from test runs rather than averaging a broader set of results, restricting testing times to periods when the temperature allowed vehicles to coast farther and faster, and preparing vehicle tires to improve the test results.
. In October 2014, Hyundai and Kia entered into a $100 million consent decree with the United States and the California Air Resources Board to settle claims arising from the EPA investigation.
. Specifically, the Espinosa plaintiffs asserted claims for violations of California Unfair Competition Law, Cal, Bus. & Prof. Code §§ 17200-17209; violations of California False Advertising Law, id, §§ 17500-17509; violations of California Consumer Legal Remedies Act, id. §§ 1750-1784; fraud;' negligent misrepresentation; and deceit, id. § 1.710.
. During the period from January 2012, when the Espinosa plaintiffs filed their complaint, until November 2012, the date the EPA announced tiie result of its investigation and Honda annoúnced its LRP program, the Espi-nosa plaintiffs focused their efforts on certifying a class. The plaintiffs otherwise limited their actions to filing two amended complaints (one to join additional class representatives) and responding to Hyundai's motion to dismiss, which was denied by the district court on April 24, 2012. (Hyundai's prior motion to dismiss had been vacated when Espi-nosa filed its amended complaint.) Hunter v. Hyundai Motor America, No, 8:12-CV-01909 (C.D. Cal. filed Nov. 2, 2012); and Brady v. Hyundai Motor America, No. 8:12-cv-1930 (C.D. Cal. filed Nov. 6, 2012) were not filed until after the LRP program -was announced.
. Although the court indicated that the marketing efforts related to “the fuel efficiency of the Elantra and Sonata vehicles,” the campaign identified by the court was limited to the 2011 Elantra. Specifically, the court noted that Hyundai had purchased advertising for the 2011 Elantra during the NFL playoffs, the Super Bowl, and the Academy Awards, placed Elantra ads on Amazon, Facebook, Yahoo, and other internet sites, used print advertising, and placed billboard ads for the 2011 Elantra in Times Square in New York and on certain California freeways.
. A Monroney Sticker is named after Senator A.S. Mike Monroney, sponsor of the Automobile Information Disclosure Act of 1958, ís U.S.C. §§ 1231-1233. The Act requires a car manufacturer to affix a label displaying information about the car’s fuel efficiency to the window of every new vehicle sold in the United States. See 15 U.S.C. §§ 1232-1233; see also 49 U.S.C. § 32908; 49 C.F.R. § 575.401 (2012). Monroney stickers are not required for sales of used cars. See 15 U.S.C. §§ 1232-1233.
.Specifically, the Hunter and Brady plaintiffs asserted claims under California Unfair Competition Law, Cal. Bus. & Prof. Code §§ 17200-17209; California False Advertising Law, id. §§ 17500-17509; California Consumer Legal Remedies Act, id. §§ 1750-1784; for fraud; for negligent misrepresentation; for unjust enrichment; and for breach of express warranty, Cal. Com. Code § 2313.
. See Rule 1.1(h), Rules of Procedure of the United States Judicial Panel on Multidistrict Litigation (" ‘Tag-along action’ refers to a civil action pending in a district court which involves common questions of fact with either (1) actions on a pending motion to transfer to create an MDL or (2) actions previously transferred to an existing MDL, and which the Panel would consider transferring under Section 1407.”).
. The settlement agreement covered 41 different Hyundai models and 35 different Kia models from 2011 to 2013.
. For lessees, the lump sum payment was based on a 2,75-year term.
. In January 2014, the settling parties filed an addendum to the proposed settlement that extended the additional $100 offer to former owners of these Hyundai models. In May 2014, tlie settling parties filed a second addendum to the settlement agreement allowing class members to submit claims through the settlement website and requiring defendants to follow certain procedures for distributing class members' payments.
. The Gentry plaintiffs alleged violations of the Virginia Motor Vehicle Warranty Enforcement Act, Va. Code Ann. §§ 59.1-207.9 to 207.16:1, the Virginia Consumer Protection Act, id. § 59.1-200(14), .and, Virginia’s false advertising statute, id. § 18.2-216.
. The Gentry plaintiffs argued that the Virginia Consumer Protection Act provides for a minimum of $500 in statutory damages for individuals who suffer damage as a result of a violation of the act, see Va. Code Ann. § 59.1-204(A), while California’s Consumer Legal Remedies Act sets no statutory minimum damages for individuals who suffer violations of the act, see Cal. Civ. Code § 1780(a). This statutory minimum of $500 is superior to the average maximum lump sum benefit of $353 that Hyundai class members are entitled to under the settlement. In addition, under the Virginia Consumer Protection Act, the trier of fact can award treble damages within its discretion if it finds that the violation was "willful,” see Va. Code Ann. § 59.1-204; Holmes v. LG Marion Corp., 258 Va. 473, 478, 521 S.E.2d 528 (1999), whereas, under California’s Legal Remedies Act, the trier of fact can only award punitive damages if it finds “clear and convincing evidence” of “oppression, fraud, or malice,” Cal. Civ. Code. § 3294(a).
.The dissent contends that we should disregard the Gentry plaintiffs’ argument regarding California choice of law rules because they alternatively argue on appeal that the failure to include a Virginia subclass would violate their due process rights. See Dissent at 710, n.3. This claim is based on the Gentry plaintiffs’ interpretation of a Virginia Supreme Court case as holding that the commencement of a class action in California that does not include a Virginia cause of action will not toll the statute of limitations in Virginia, and thus they would be time-barred from bringing their Virginia-specific claims. Recognizing this interpretation is in dispute, the Gentry plaintiffs alternatively urged us to certify this question to the Virginia Supreme Court. Because the Gentry plaintiffs raised their choice-of-law argument to the district court, we do not place any significance on the fact that they later also raised an alternative argument.
. In addition to the Gentry plaintiffs, the objectors included the named plaintiffs in two other actions transferred to the district court as part of MDL No. 2424, Krauth and Wilson v. Kia Motors America, Inc., No. 13-CV-1069 (D.N.J. filed Jan. 24, 2013), These plaintiffs are not objectors in this appeal.
. The class was defined as: ”[a]ll current and former owners and lessees of a Class Vehicle (i) who were the owner or lessee, on or before November 2, 2012, of such Class Vehicle that was registered in the District of Columbia or one of the fifty (50) states of the United States,” with several small exceptions.
. The dissent argues that we fail to apply the correct standard of review. See Dissent at 714-15. In making this argument, the dissent echoes the dissenting justices in Amchem, which likewise argued that the majority had erred in failing to give sufficient deference to the district court. Amchem, 521 U.S. at 630, 117 S.Ct. 2231 (Breyer, J., concurring in part and dissenting in part). But we are bound by the Amchem majority, which indicates that a district court makes a legal error, and thus abuses its discretion, -when it fails to scrutinize a settlement class to the same extent as a litigation class. Id. at-620, 117. S.Ct. 2231.
. California courts have likewise read Tobacco II narrowly, and have rejected the argument that class-wide reliance can be presumed "whenever there is a showing that a misrepresentation was material." Tucker v. Pacific Bell Mobile Services, 208 Cal. App. 4th 201, 226-27, 145 Cal.Rptr.3d 340 (2012) (citing Tobacco II, 46 Cal. 4th at 327, 93 Cal. Rptr.3d 559, 207 P.3d 20). As indicated in Mazza, reliance can be presumed only when there is the sort of massive decades-long advertising campaign at issue in Tobacco II. 666 F.3d at 596. Regardless whether the Hyundai and Kia advertising campaign here was more extensive than the campaign in Mazza, see Dissent at 716, it does not come close to the level of cigarette advertising from the 1960s to the 2000s.
. The district court's statement in its November 2012 ruling that class-wide reliance on the challenged advertising could be presumed due to the "extensive sweep” of Hyundai’s marketing efforts focused solely on the 2011 Elantra model; the Sonata model is merely mentioned in an aside. The court did not address either the 35 other Hyundai models or any Kia models. This is not surprising, given that the district court relied on a declaration that focused almost exclusively on the 2011 Elantras, with only limited mention made of the 2011 Sonata models or any 2012 models. Moreover, because the advertising was limited in time (under a year) and scope, it does not come close to the pervasive campaign (extending over 40 years by 11 separate companies) described in Tobacco II,
, Although the settling parties filed expert •reports, the district court did not discuss or address them in any way. An examination of. the reports, would have likely led the district court to probe some of expert’s questionable assumptions, such as the assumption that car owners who entered the LRP program before the settlement would own their cars for a shorter period of time than car owners who entered the LRP program after the settlement, and the assumption that all of the class members who entered the LRP program after the settlement would not have done so of their own accord regardless of the settlement. •
. The dissent contends that "we have rejected objectors' arguments that a federal investigation merits a reduction in class counsel’s fees,” citing Vizcaino v. Microsoft Corp., 290 F.3d 1043, 1048 n.3 (9th Cir. 2002). See Dissent at 717. This is incorrect. Vizcaino concluded that the federal investigation was irrelevant to the pivotal issue in the suit, and therefore concluded that it did not merit a reduction in fees. By contrast, the EPA investigation here established that Hyundai and Kia had misstated fuel efficiency estimates for certain models, which was the pivotal issue in this class action, and which directly led Hyundai and Kia to implement the LRP program.
. We also disagree with the district court’s conclusion that “the issue of collusion is not present in the attomey[s’] fees context” because “the attomey[s’] fees were awarded separately from the class recovery and did not impact class recovery.” The district court’s responsibility to conduct an independent inquiry into the reasonableness of- attorneys' fees is of equal, if not greater, importance when attorneys’ fees are awarded separately from the class award. See Bluetooth Headset Prods. Liab. Litig., 654 F.3d at 943. Indeed, we have identified this exact arrangement as one of the "subtle signs” of collusion in the settlement context. See id. at 947. Similar to the ■ "clear sailing agreement” examined in Bluetooth Headset Prods. Liab. Litig., the parties reached an agreement on the amount of attorneys’ fees to be paid in the Hunter, and Brady actions, and the defendants did not contest the fees before the district court.
. In light of our decision that the district court abused its discretion in certifying a settlement class under Rule 23(b)(3) without conducting a choice of-law analysis and considering differences in state consumer protection laws, we do not reach the objection raised by James Feinman, counsel for the Gentry plaintiffs, that the district court abused its discretion in not awarding him attorneys’ fees.
.Objectors raised a number of additional arguments, including claims that: the district court abused its discretion in certifying the ■ settlement class because named plaintiffs did not adequately represent the interests of the class, as required under Rule 23(a)(4); the district court’s failure to conduct a choice of law analysis violated absent class members’ due process rights; the district court's failure to certify a Virginia subclass violated class members’ due process rights; and the settlement was not fair and adequate under Rule 23(e), Because we conclude that the district court abused its discretion in certifying the class under Rule 23(b)(3), we do not consider these arguments. See Wang, 737 F.3d at 546.