HAMILTON, Circuit Judge.

We agreed to hear this appeal under Federal Rule of Civil Procedure 23(f), which permits interlocutory review of orders granting or denying class action certification, to address whether Rule 23(b)(3) imposes a heightened "ascertainability" requirement as the Third Circuit and some district courts have held recently. See, e.g., Carrera v. Bayer Corp., 727 F.3d 300 (3d Cir.2013). In this case, the plaintiff alleges consumer fraud by the seller of a dietary supplement, and the district court certified a plaintiff class. The court found that the proposed class satisfies the explicit requirements of Rule 23(a) and (b)(3), and the court rejected's argument that Rule 23(b)(3) implies a heightened ascertainability requirement.

We affirm. We and other courts have long recognized an implicit requirement under Rule 23 that a class must be defined clearly and that membership be defined by objective criteria rather than by, for example, a class member's state of mind. In addressing this requirement, courts have sometimes used the term "ascertainability." They have applied this requirement to all class actions, regardless of whether certification was sought under Rule 23(b)(1), (2), or (3). Class definitions have failed this requirement when they were too vague or subjective, or when class membership was defined in terms of success on the merits (so-called "fail-safe" classes). This version of ascertainability is well-settled in our circuit, and this class satisfies it.

More recently, however, some courts have raised the bar for class actions under Rule 23(b)(3). Using the term "ascertainability," at times without recognizing the extension, these courts have imposed a new requirement that plaintiffs prove at the certification stage that there is a "reliable and administratively feasible" way to identify all who fall within the class definition. These courts have moved beyond examining the adequacy of the class definition itself to examine the potential difficulty of identifying particular members of the class and evaluating the validity of claims they might eventually submit. See Byrd v. Aaron's Inc., 784 F.3d 154, 168 (3d Cir.2015) (distinguishing between our circuit's standard and the Third Circuit's ascertainability requirement).

This heightened requirement has defeated certification, especially in consumer class actions. See, e.g., Karhu v. Vital Pharmaceuticals, Inc., ___ Fed.Appx. ___, ___-___, 2015 WL 3560722, at *2-4 (11th Cir. June 9, 2015) (purchasers of dietary supplements); Carrera, 727 F.3d at 307-12 (purchasers of dietary supplements); Xavier v. Philip Morris USA Inc., 787 F.Supp.2d 1075, 1089-90 (N.D.Cal.2011) (Marlboro smokers); Weiner v. Snapple Beverage Corp., No. 07 Civ. 8742(DLC), 2010 WL 3119452, at *12-13 (S.D.N.Y. Aug. 5, 2010) (purchasers of Snapple beverages). All of these classes would seem to have satisfied the established meaning of "ascertainability." See generally Myriam Gilles, Class Dismissed: Contemporary Judicial Hostility to Small-Claims Consumer Class Actions, 59 DePaul L.Rev. 305 (2010) (describing recent cases).

We decline to follow this path and will stick with our settled law. Nothing in Rule 23 mentions or implies this heightened requirement under Rule 23(b)(3), which has the effect of skewing the balance that district courts must strike when deciding whether to certify classes. The policy concerns motivating the heightened ascertainability requirement are better addressed by applying carefully the explicit requirements of Rule 23(a) and especially (b)(3). These existing requirements already address the balance of interests that Rule 23 is designed to protect. A court must consider "the likely difficulties in managing a class action," but in doing so it must balance countervailing interests to decide whether a class action "is superior to other available methods for fairly and efficiently adjudicating the controversy." See Fed.R.Civ.P. 23(b)(3).

The heightened ascertainability requirement upsets this balance. In effect, it gives one factor in the balance absolute priority, with the effect of barring class actions where class treatment is often most needed: in cases involving relatively low-cost goods or services, where consumers are unlikely to have documentary proof of purchase. These are cases where the class device is often essential "to overcome the problem that small recoveries do not provide the incentive for any individual to bring a solo action prosecuting his or her rights." Amchem Products, Inc. v. Windsor, 521 U.S. 591, 617, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997), quoting Mace v. Van Ru Credit Corp., 109 F.3d 338, 344 (7th Cir.1997); see also Suchanek v. Sturm Foods, Inc., 764 F.3d 750, 760 (7th Cir. 2014) (reversing denial of class certification: "a class action has to be unwieldy indeed before it can be pronounced an inferior alternative — no matter how massive the fraud or other wrongdoing that will go unpunished if class treatment is denied — to no litigation at all"), quoting Carnegie v. Household Int'l, Inc., 376 F.3d 656, 661 (7th Cir.2004) (affirming certification of class with millions of members).

I. Factual and Procedural Background

Plaintiff Vince Mullins sued defendant Direct Digital, LLC for fraudulently representing that its product, Instaflex Joint Support, relieves joint discomfort. He alleges that statements on the Instaflex labels and marketing materials — "relieve discomfort," "improve flexibility," "increase mobility," "support cartilage repair," "scientifically formulated," and "clinically tested for maximum effectiveness" — are fraudulent because the primary ingredient in the supplement (glucosamine sulfate) is nothing more than a sugar pill and there is no scientific support for these claims. Mullins asserts that Direct Digital is liable for consumer fraud under the Illinois Consumer Fraud and Deceptive Business Practices Act, 815 ILCS 505/1 et seq., and similar consumer protection laws in nine other states.

Mullins moved to certify a class of consumers "who purchased Instaflex within the applicable statute of limitations of the respective Class States for personal use until the date notice is disseminated." The district court certified the class under Rule 23(b)(3).

Direct Digital filed a petition for leave to appeal under Rule 23(f) arguing that the district court abused its discretion in certifying the class without first finding that the class was "ascertainable." Direct Digital also argued that the district court erred by concluding that the efficacy of a health product can qualify as a "common" question under Rule 23(a)(2). We granted the Rule 23(f) petition primarily to address the developing law of ascertainability, including among district courts within this circuit. See Blair v. Equifax Check Services, Inc., 181 F.3d 832, 835 (7th Cir.1999) (granting an appeal is appropriate to "facilitate the development of the law" governing class actions).1

We review the grant or denial of a motion for class certification for an abuse of discretion, e.g., Harper v. Sheriff of Cook County, 581 F.3d 511, 514 (7th Cir. 2009), but a decision based on an erroneous view of the law, such as imposing a new requirement under Rule 23(b)(3), is likely to be an abuse of discretion. E.g., Ervin v. OS Restaurant Services, Inc., 632 F.3d 971, 976 (7th Cir.2011) ("If, however, the district court applies an incorrect legal rule as part of its decision, then the framework within which it has applied its discretion is flawed, and the decision must be set aside as an abuse.").

II. Analysis

A. The Established Meaning of "Ascertainability"

We begin with the current state of the law in this circuit. Rule 23 requires that a class be defined, and experience has led courts to require that classes be defined clearly and based on objective criteria. See William B. Rubenstein, Newberg on Class Actions § 3:3 (5th ed.2015); Joseph M. McLaughlin, McLaughlin on Class Actions § 4:2 (11th ed.2014); see, e.g., Matamoros v. Starbucks Corp., 699 F.3d 129, 139 (1st Cir.2012); Bakalar v. Vavra, 237 F.R.D. 59, 64 (S.D.N.Y.2006); DeBremaecker v. Short, 433 F.2d 733, 734 (5th Cir.1970) (per curiam). When courts wrote of this implicit requirement of "ascertainability," they trained their attention on the adequacy of the class definition itself. They were not focused on whether, given an adequate class definition, it would be difficult to identify particular members of the class.

This "weak" version of ascertainability has long been the law in this circuit. See Jamie S. v. Milwaukee Public Schools, 668 F.3d 481, 495 (7th Cir.2012) ("It's not hard to see how this class lacks the definiteness required for class certification; there is no way to know or readily ascertain who is a member of the class."); Oshana v. Coca-Cola Co., 472 F.3d 506, 513 (7th Cir.2006) (class definition "must be definite enough that the class can be ascertained"); accord, Adashunas v. Negley, 626 F.2d 600, 604 (7th Cir.1980) ("In summary, the proposed class of plaintiffs is so highly diverse and so difficult to identify that it is not adequately defined or nearly ascertainable.").

The language of this well-settled requirement is susceptible to misinterpretation, though, which may explain some of the doctrinal drift described below. To understand its established meaning, it's better to focus on the three common problems that have caused plaintiffs to flunk this requirement.

First, classes that are defined too vaguely fail to satisfy the "clear definition" component. See, e.g., Young v. Nationwide Mutual Ins. Co., 693 F.3d 532, 538 (6th Cir.2012) ("There can be no class action if the proposed class is amorphous or imprecise." (citation and internal quotation marks omitted)); APB Associates, Inc. v. Bronco's Saloon, Inc., 297 F.R.D. 302, 316 (E.D.Mich.2013) (denying certification because proposed class definition was too "imprecise and amorphous"); DeBremaecker, 433 F.2d at 734 (affirming denial of certification for proposed class defined as residents "active in the `peace movement'"); 7A Charles Alan Wright et al., Federal Practice & Procedure § 1760 (3d ed.2005) (collecting cases). Vagueness is a problem because a court needs to be able to identify who will receive notice, who will share in any recovery, and who will be bound by a judgment. See Kent v. Sun-America Life Ins. Co., 190 F.R.D. 271, 278 (D.Mass.2000). To avoid vagueness, class definitions generally need to identify a particular group, harmed during a particular time frame, in a particular location, in a particular way. See McLaughlin on Class Actions § 4:2; see, e.g., Rodriguez v. Berrybrook Farms, Inc., 672 F.Supp. 1009, 1012 (W.D.Mich.1987) (granting certification and noting the class definition specified "a group of agricultural laborers during a specific time frame and at a specific location who were harmed in a specific way").

Second, classes that are defined by subjective criteria, such as by a person's state of mind, fail the objectivity requirement. E.g., Simer v. Rios, 661 F.2d 655, 669-70 (7th Cir.1981) (affirming denial of certification of class of people who felt discouraged from applying for government energy assistance); Alliance to End Repression v. Rochford, 565 F.2d 975, 977-78 (7th Cir.1977) (affirming certification of class defined by actions of defendants rather than class members' states of mind); Harris v. General Development Corp., 127 F.R.D. 655, 659 (N.D.Ill.1989) (denying class certification of proposed subclass defined by mental state: "The proposed class of persons who allegedly were discouraged from applying at GDC is too imprecise and speculative to be certified."); 7A Wright et al., Federal Practice & Procedure § 1760 (collecting cases). Plaintiffs can generally avoid the subjectivity problem by defining the class in terms of conduct (an objective fact) rather than a state of mind. See, e.g., National Org. for Women, Inc. v. Scheidler, 172 F.R.D. 351, 358-59 (N.D.Ill.1997) (accepting modified class definition so that "membership in the classes sought to be certified is based exclusively on the defendants' conduct with no particular state of mind required"); Newberg on Class Actions § 3:5.

Third, classes that are defined in terms of success on the merits — so-called "fail-safe classes" — also are not properly defined. See In re Nexium Antitrust Litig., 777 F.3d 9, 22 (1st Cir.2015); Young, 693 F.3d at 538; Messner v. Northshore Univ. HealthSystem, 669 F.3d 802, 825 (7th Cir.2012); Randleman v. Fidelity Nat'l Title Ins. Co., 646 F.3d 347, 352 (6th Cir.2011); but see In re Rodriguez, 695 F.3d 360, 369-70 (5th Cir.2012) (affirming fail-safe class certification). Defining the class in terms of success on the merits is a problem because "a class member either wins or, by virtue of losing, is defined out of the class and is therefore not bound by the judgment." Messner, 669 F.3d at 825. This raises an obvious fairness problem for the defendant: the defendant is forced to defend against the class, but if a plaintiff loses, she drops out and can subject the defendant to another round of litigation. See Erin L. Geller, Note, The Fail-Safe Class as an Independent Bar to Class Certification, 81 Fordham L.Rev. 2769 (2013). The key to avoiding this problem is to define the class so that membership does not depend on the liability of the defendant.

The class definition in this case complies with this settled law and avoids all of these problems. It is not vague. It identifies a particular group of individuals (purchasers of Instaflex) harmed in a particular way (defrauded by labels and marketing materials) during a specific period in particular areas. The class definition also is not based on subjective criteria. It focuses on the act of purchase and Direct Digital's conduct in labeling and advertising the product. It also does not create a fail-safe class. If Direct Digital prevails, res judicata will bar class members from re-litigating their claims.

Direct Digital argues, however, that we should demand more. It urges us to adopt a new component to the ascertain-ability requirement that goes beyond the adequacy of the class definition itself. Drawing on recent decisions by the Third Circuit, Direct Digital argues that class certification should be denied if the plaintiff fails to show a reliable and administratively feasible way to determine whether a particular person is a member of the class. And, Direct Digital continues, affidavits from putative class members are insufficient as a matter of law to satisfy this requirement.

In support of this argument, Direct Digital asserts that the only method of identifying class members here is by affidavit from the putative class members themselves. That remains to be seen. We do not know yet what sales and customer records Direct Digital has. We assume for purposes of this decision that Direct Digital will have no records for a large number of retail customers. We also assume that many consumers of Instaflex are unlikely to have kept their receipts since it's a relatively inexpensive consumer good.

B. The Recent Expansion of "Ascertainability"

To understand the genesis of Direct Digital's argument, we briefly summarize the law of the Third Circuit, which has adopted this more stringent version of ascertainability. The Third Circuit's innovation began with Marcus v. BMW of North America, LLC, 687 F.3d 583 (3d Cir.2012), where the court vacated certification of a poorly defined class. The decisive portion of the opinion, id. at 592-94, certainly seems sound, but the opinion went on to caution that on remand, if defendants' records would not identify class members, the district court should not approve a method relying on "potential class members' say so," and the opinion said that reliance on class members' affidavits might not be "proper or just," id. at 594 (internal quotation marks omitted). The opinion did not explain this new requirement other than to cite an easily distinguishable district court decision.

Since Marcus, the court has applied this heightened ascertainability requirement in several more cases: Hayes v. Wal-Mart Stores, Inc., 725 F.3d 349, 354-56 (3d Cir. 2013); Carrera v. Bayer Corp., 727 F.3d 300, 305-12 (3d Cir.2013); Grandalski v. Quest Diagnostics Inc., 767 F.3d 175, 184-85 (3d Cir.2014); Shelton v. Bledsoe, 775 F.3d 554, 559-63 (3d Cir.2015); Byrd v. Aaron's Inc., 784 F.3d 154, 161-71 (3d Cir.2015). As the requirement has evolved, several members of the court have expressed doubts about the expanding ascertainability doctrine. See Byrd, 784 F.3d at 172-77 (Rendell, J., concurring); Carrera v. Bayer Corp., No. 12-2621, 2014 WL 3887938, at *1-3 (3d Cir. May 2, 2014) (Ambro, J., dissenting from denial of rehearing en banc).2

As it stands now, the Third Circuit's test for ascertainability has two prongs: (1) the class must be "defined with reference to objective criteria" (consistent with long-established law discussed above), and (2) there must be "a reliable and administratively feasible mechanism for determining whether putative class members fall within the class definition." Byrd, 784 F.3d at 163, quoting Hayes, 725 F.3d at 355; see also Shelton, 775 F.3d at 560 (making clear that "the question of ascertainability" is separate from "the question of whether the class was properly defined").

This second requirement sounds sensible at first glance. Who could reasonably argue that a plaintiff should be allowed to certify a class whose members are impossible to identify? In practice, however, some courts have used this requirement to erect a nearly insurmountable hurdle at the class certification stage in situations where a class action is the only viable way to pursue valid but small individual claims.

The demands of this heightened requirement are most apparent from the Third Circuit's discussion of self-identification by affidavit. It has said that affidavits from putative class members cannot satisfy the stringent ascertainability requirement. See Carrera, 727 F.3d at 308-12 (remanding to give plaintiff "another opportunity to satisfy the ascertainability requirement" but rejecting plaintiff's attempt to use affidavits from class members to show their purchases of weight loss supplement); Hayes, 725 F.3d at 356 ("But the nature or thoroughness of a defendant's recordkeeping does not alter the plaintiff's burden to fulfill Rule 23's requirements."); Marcus, 687 F.3d at 594 ("We caution, however, against approving a method that would amount to no more than ascertaining by potential class members' say so."). Direct Digital urges us to adopt this rule and to reverse the certification order here because the only method for identifying class members proposed by Mullins in the district court was self-identification by affidavit.

We decline to do so. The Third Circuit's approach in Carrera, which is at this point the high-water mark of its developing ascertainability doctrine, goes much further than the established meaning of ascertainability and in our view misreads Rule 23. Carrera and cases like it have given four policy reasons for requiring more than affidavits from putative class members. We address each one below and find them unpersuasive.

In general, we think imposing this stringent version of ascertainability does not further any interest of Rule 23 that is not already adequately protected by the Rule's explicit requirements. On the other side of the balance, the costs of imposing the requirement are substantial. The stringent version of ascertainability effectively bars low-value consumer class actions, at least where plaintiffs do not have documentary proof of purchases, and sometimes even when they do. Accordingly, we conclude that the district court here did not abuse its discretion by deferring until later in the litigation decisions about more detailed aspects of ascertainability and the management of any claims process. At bottom, the district court was correct not to let a quest for perfect treatment of one issue become a reason to deny class certification and with it the hope of any effective relief at all.

We now turn to the policy concerns identified by the courts that have embraced this heightened ascertainability requirement. The policy concerns are substantial and legitimate, but we do not believe they justify the new requirement. As will become clear, we agree in essence with Judge Rendell's concurring opinion in Byrd, 784 F.3d at 172-77, which urged "retreat from [the] heightened ascertainability requirement in favor of following the historical meaning of ascertainability under Rule 23," id. at 177.

1. Administrative Convenience

Some courts have argued that imposing a stringent version of ascertainability "eliminates serious administrative burdens that are incongruous with the efficiencies expected in a class action by insisting on the easy identification of class members." Marcus, 687 F.3d at 593 (citation and internal quotation marks omitted). It does this by ensuring that the court will be able to identify class members without "extensive and individualized fact-finding or mini-trials." Carrera, 727 F.3d at 307 (citation and internal quotation marks omitted).

This concern about administrative inconvenience is better addressed by the explicit requirements of Rule 23(b)(3), which requires that the class device be "superior to other available methods for fairly and efficiently adjudicating the controversy." One relevant factor is "the likely difficulties in managing a class action." Fed. R.Civ.P. 23(b)(3)(D).

The superiority requirement of Rule 23(b)(3) is clarified by substantial case law. See 7AA Wright et al., Federal Practice & Procedure §§ 1779, 1780. Imposing a stringent version of ascertainability because of concerns about administrative inconvenience renders the manageability criterion of the superiority requirement superfluous. See Daniel Luks, Note, Ascertainability in the Third Circuit: Name That Class Member, 82 Fordham L.Rev. 2359, 2395 (2014). It also conflicts with the well-settled presumption that courts should not refuse to certify a class merely on the basis of manageability concerns. See, e.g., In re Visa Check/MasterMoney Antitrust Litig., 280 F.3d 124, 140 (2d Cir.2001) (Sotomayor, J.) (noting that failure to certify a class action under Rule 23(b)(3) solely on manageability grounds is generally disfavored), overruled on other grounds by In re IPO, 471 F.3d 24 (2d Cir.2006); accord, Byrd, 784 F.3d at 175 (Rendell, J., concurring) ("Imposing a proof-of-purchase requirement does nothing to ensure the manageability of a class or the `efficiencies' of the class action mechanism; rather, it obstructs certification by assuming that hypothetical roadblocks will exist at the claims administration stage of the proceedings.").

A reader might fairly ask whether there is any practical difference between addressing administrative inconvenience as a matter of ascertainability versus as a matter of superiority. In fact, there is. When administrative inconvenience is addressed as a matter of ascertainability, courts tend to look at the problem in a vacuum, considering only the administrative costs and headaches of proceeding as a class action. See, e.g., Sethavanish, 2014 WL 580696, at *6 (purchasers of "all natural" nutrition bars sold through retailers; denying class certification solely on the ground of ascertainability without addressing other available methods for adjudicating the controversy). But when courts approach the issue as part of a careful application of Rule 23(b)(3)'s superiority standard, they must recognize both the costs and benefits of the class device. See 7AA Wright et al., Federal Practice & Procedure § 1780 ("Viewing the potential administrative difficulties from a comparative perspective seems sound and a decision against class-action treatment should be rendered only when the ministerial efforts simply will not produce corresponding efficiencies. In no event should the court use the possibility of becoming involved with the administration of a complex lawsuit as a justification for evading the responsibilities imposed by Rule 23.").

Rule 23(b)(3)'s superiority requirement, unlike the freestanding ascertainability requirement, is comparative: the court must assess efficiency with an eye toward "other available methods." In many cases where the heightened ascertainability requirement will be hardest to satisfy, there realistically is no other alternative to class treatment. See id. ("If judicial management of a class action ... will reap the rewards of efficiency and economy for the entire system that the drafters of the federal rule envisioned, then the individual judge should undertake the task. Ironically, those Rule 23(b)(3) actions requiring the most management may yield the greatest pay-off in terms of effective dispute resolution."); cf. Schleicher v. Wendt, 618 F.3d 679, 686-87 (7th Cir.2010) (rejecting defendant's invitation to "tighten" Rule 23 requirements for class certification and noting that doing so would make certification impossible in many securities fraud cases).

This does not mean, of course, that district courts should automatically certify classes in these difficult cases. But it does mean that before refusing to certify a class that meets the requirements of Rule 23(a), the district court should consider the alternatives as Rule 23(b)(3) instructs rather than denying certification because it may be challenging to identify particular class members. District courts have considerable experience with and flexibility in engineering solutions to difficult problems of case management.

In addition, a district judge has discretion to (and we think normally should) wait and see how serious the problem may turn out to be after settlement or judgment, when much more may be known about available records, response rates, and other relevant factors. And if a problem is truly insoluble, the court may decertify the class at a later stage of the litigation. See Carnegie v. Household Int'l, Inc., 376 F.3d 656, 661 (7th Cir.2004).

If faced with what appear to be unusually difficult manageability problems at the certification stage, district courts have discretion to insist on details of the plaintiff's plan for notifying the class and managing the action. In conducting this inquiry, district courts should consider also whether the administrative burdens can be eased by the procedures set out in Rule 23(c) and (d). See, e.g., Bobbitt v. Academy of Court Reporting, Inc., 252 F.R.D. 327, 344-45 (E.D.Mich.2008) (granting class certification despite potential manageability problems and noting options "a special master, representative trials, or other means" to manage the problems).

Under this comparative framework, refusing to certify on manageability grounds alone should be the last resort. See Carnegie, 376 F.3d at 661 ("a class action has to be unwieldy indeed before it can be pronounced an inferior alternative — no matter how massive the fraud or other wrongdoing that will go unpunished if class treatment is denied — to no litigation at all"), quoted in Suchanek, 764 F.3d at 760. In all events, deciding whether and when to insist on details, and how many details, are matters for the sound discretion of district judges who have so much first-hand experience managing class actions.

On the other hand, if courts look only at the cost-side of the equation and fail to consider administrative solutions like those available under Rule 23(c) and (d), courts will err systematically against certification. See Geoffrey C. Shaw, Note, Class Ascertainability, 124 Yale L.J. 2354, 2396-99 (2015) (explaining why addressing issue of manageability under umbrella of superiority is preferable to addressing it as a matter of ascertainability). The stringent version of ascertainability invites precisely this type of systemic error.

2. Unfairness to Absent Class Members

Courts also have asserted that the heightened ascertainability requirement is needed to protect absent class members. If the identities of absent class members cannot be ascertained, the argument goes, it is unfair to bind them by the judicial proceeding. See Carrera, 727 F.3d at 307; Marcus, 687 F.3d at 593. A central premise of this argument is that class members must receive actual notice of the class action so that they do not lose their opt-out rights.

We believe that premise is mistaken. For Rule 23(b)(3) classes, Rule 23(c)(2)(B) requires the "best notice that is practicable under the circumstances, including individual notice to all members who can be identified through reasonable effort." The rule does not insist on actual notice to all class members in all cases. It recognizes it might be impossible to identify some class members for purposes of actual notice. See Shaw, 124 Yale L.J. at 2367-69. While actual individual notice may be the ideal, due process does not always require it. See Mirfasihi v. Fleet Mortgage Corp., 356 F.3d 781, 786 (7th Cir.2004) (rejecting requirement of individual notice); Juris v. Inamed Corp., 685 F.3d 1294, 1321 (11th Cir.2012) (noting that "even in Rule 23(b)(3) class actions, due process does not require that class members actually receive notice" and collecting cases); accord, Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 811-12, 105 S.Ct. 2965, 86 L.Ed.2d 628 (1985); Mullane v. Central Hanover Bank & Trust Co., 339 U.S. 306, 314-15, 70 S.Ct. 652, 94 L.Ed. 865 (1950).

When class members' names and addresses are known or knowable with reasonable effort, notice can be accomplished by first-class mail. See, e.g., Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 174-75, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974). When that is not possible, courts may use alternative means such as notice through third parties, paid advertising, and/or posting in places frequented by class members, all without offending due process. See Hughes v. Kore of Indiana Enterprise, Inc., 731 F.3d 672, 676-77 (7th Cir.2013). As long as the alternative means satisfy the standard of Rule 23(b)(3), there is no due process violation. See, e.g., Lilly v. Jamba Juice Co., No. 13-cv-02998-JST, 2014 WL 4652283, at *5 (N.D.Cal. Sept. 18, 2014) (rejecting notice argument for same reason); Boundas v. Abercrombie & Fitch Stores, Inc., 280 F.R.D. 408, 418 (N.D.Ill. 2012) (same). Due process simply does not require the ability to identify all members of the class at the certification stage.

More broadly, the stringent version of ascertainability loses sight of a critical feature of class actions for low-value claims like this one. In these cases, "only a lunatic or a fanatic" would litigate the claim individually, Carnegie v. Household Int'l, Inc., 376 F.3d 656, 661 (7th Cir.2004), so opt-out rights are not likely to be exercised by anyone planning a separate individual lawsuit. When this is true, it is particularly important that the types of notice that courts require correspond to the value of the absent class members' interests. Cf. Mathews v. Eldridge, 424 U.S. 319, 334-35, 96 S.Ct. 893, 47 L.Ed.2d 18 (1976). That is why in Hughes, for example, where each plaintiff's claim was valued at approximately $1,000 or less, we approved a notice plan consisting of sticker notices on the defendant's two ATMs, publication of a notice in the primary local newspaper, and notice on a website. Hughes, 731 F.3d at 676-77. We did not insist on first-class mail even though the notice plan likely would not reach everyone in the class. We approved the plan because the notice plan was "commensurate with the stakes." Id. at 676.

The heightened ascertainability approach upsets this balance. It comes close to insisting on actual notice to protect the interests of absent class members, yet overlooks the reality that without certification, putative class members with valid claims would not recover anything at all. See Amchem, 521 U.S. at 617, 117 S.Ct. 2231; Eisen, 417 U.S. at 161, 94 S.Ct. 2140; Hughes, 731 F.3d at 677; Butler v. Sears, Roebuck & Co., 727 F.3d 796, 798 (7th Cir.2013); see also, e.g., Ebin v. Kangadis Food Inc., 297 F.R.D. 561, 567 (S.D.N.Y.2014) ("Against this background, the ascertainability difficulties, while formidable, should not be made into a device for defeating the action."); Astiana v. Kashi Co., 291 F.R.D. 493, 500 (S.D.Cal.2013) ("If class actions could be defeated because membership was difficult to ascertain at the class certification stage, there would be no such thing as a consumer class action." (citation and internal quotation marks omitted)). When it comes to protecting the interests of absent class members, courts should not let the perfect become the enemy of the good.

3. Unfairness to Bona Fide Class Members

The third concern offered to justify the heightened ascertainability requirement is the interests of class members with valid claims. Courts have expressed concern that if class members are identified only by their own affidavits, individuals without a valid claim will submit erroneous or fraudulent claims and dilute the share of recovery for true class members. See Carrera, 727 F.3d at 310 ("It is unfair to absent class members if there is a significant likelihood their recovery will be diluted by fraudulent or inaccurate claims.").3

Again, this concern about the danger of fraudulent or mistaken claims is legitimate and understandable, especially when contemplating the prospect that money might seem available just for the asking. In the words of then-future President John Adams, "it is prudent not to put virtue to too serious a test." 2 John Adams, The Works of John Adams, Second President of the United States: Diary, with A Life of the Author, Notes & Illustrations 457 (Charles Francis Adams ed. 1850) (during 1775 debate on whether to open ports for trade and the need for customs officials to regulate the ports).

We see two problems with using these concerns to impose the heightened ascertainability standard. First, in practice, the risk of dilution based on fraudulent or mistaken claims seems low, perhaps to the point of being negligible. We are aware of no empirical evidence that the risk of dilution caused by inaccurate or fraudulent claims in the typical low-value consumer class action is significant. In most cases, the expected recovery is so small that we question whether many people would be willing to sign affidavits under penalty of perjury saying that they purchased the good or service. See Byrd, 784 F.3d at 175 (Rendell, J., concurring). In this case, for example, the value of each claim is approximately $70 (the retail price). Direct Digital has provided no evidence, and we have found none, that claims of this magnitude have provoked the widespread submission of inaccurate or fraudulent claims.

We could be wrong, of course, about this empirical prediction. Suppose people are more willing to file inaccurate or fraudulent claims for low-value recoveries than we suspect. Even then, the risk of dilution appears small because only a tiny fraction of eligible claimants ever submit claims for compensation in consumer class actions. See Christopher R. Leslie, The Significance of Silence: Collective Action Problems and Class Action Settlements, 59 Fla. L.Rev. 71, 119-20 (2007) (noting that it is not unusual to have participation rates of 10 to 15 percent and examining more recent examples of rates lower than 5 percent). Any participation rate less than 100 percent leaves unclaimed funds in the pot, whether it is a judgment award or a settlement fund. When there are unclaimed funds, the addition of a fraudulent or inaccurate claim typically does not detract from a bona fide class member's recovery because the non-deserving claimant merely takes from unclaimed funds, not the deserving class member. It is of course theoretically possible that the total sum claimed by non-deserving claimants exceeds the total amount of unclaimed funds, in which case there would be dilution, but given the low participation rates actually observed in the real world, this danger is not so great that it justifies denying class certification altogether, at least without empirical evidence supporting the fear. See Myriam Gilles, Class Dismissed: Contemporary Judicial Hostility to Small-Claims Consumer Class Actions, 59 DePaul L.Rev. 305, 315 (2010) (given actual claims rates in practice, "it is simply not true that compensation of uninjured parties affects the compensation interests of injured class members"). Carrera and cases like it have given no reason to think otherwise.

We recognize that the risk of mistaken or fraudulent claims is not zero. But courts are not without tools to combat this problem during the claims administration process. They can rely, as they have for decades, on claim administrators, various auditing processes, sampling for fraud detection, follow-up notices to explain the claims process, and other techniques tailored by the parties and the court to take into account the size of the claims, the cost of the techniques, and an empirical assessment of the likelihood of fraud or inaccuracy. See Manual for Complex Litigation §§ 21.66-.661 (4th ed.2004); Newberg on Class Actions § 12:20; see also, e.g., Boundas v. Abercrombie & Fitch Stores, Inc., 280 F.R.D. 408, 417 (N.D.Ill.2012) (affirming class certification where class included individuals who threw away promotional gift cards because they were told that the balances had been voided: "anybody claiming class membership on that basis will be required to submit an appropriate affidavit, which can be evaluated during the claims administration process"). Relying on concerns about what are essentially claim administration issues to deny certification and to prevent any recovery on valid claims upsets the balance a district judge must consider. In the face of such empirical uncertainty, a district judge has discretion to say let's wait until we know more and see how big a problem this turns out to be.

The second problem with this dilution argument is that class certification provides the only meaningful possibility for bona fide class members to recover anything at all. Keep in mind what's at stake. If the class is certified and fraudulent or inaccurate claims actually cause dilution, then deserving class members still receive something. But if class certification is denied, they will receive nothing, for they would not have brought suit individually in the first place. See Amchem, 521 U.S. at 617, 117 S.Ct. 2231; Eisen, 417 U.S. at 161, 94 S.Ct. 2140; Hughes, 731 F.3d at 677; Butler, 727 F.3d at 798. To deny class certification based on fear of dilution would in effect deprive bona fide class members of any recovery as a means to ensure they do not recover too little.

This stringent approach has far-reaching consequences, too. By "focusing on making absolutely certain that compensation is distributed only to those individuals who were actually harmed," the heightened ascertainability requirement "has ignored an equally important policy objective of class actions: deterring and punishing corporate wrongdoing." Byrd, 784 F.3d at 175-76 (Rendell, J., concurring), discussing Hughes, 731 F.3d at 677 ("A class action, like litigation in general, has a deterrent as well as a compensatory objective."). Even if the risk of dilution is not trivial, refusing to certify on this basis effectively immunizes defendants from liability because they chose not to maintain records of the relevant transactions. See Daniels v. Hollister Co., 440 N.J.Super. 359, 113 A.3d 796, 801 (N.J.App.2015) ("Ascertainability... is particularly misguided when applied to a case where any difficulties encountered in identifying class members are a consequence of a defendant's own acts or omissions.... Allowing a defendant to escape responsibility for its alleged wrongdoing by dint of its particular recordkeeping policies ... is not in harmony with the principles governing class actions."); Birchmeier v. Caribbean Cruise Line, Inc., 302 F.R.D. 240, 250 (N.D.Ill.2014) ("Doing this — or declining to certify a class altogether, as defendants propose — would create an incentive for a person to violate the TCPA on a mass scale and keep no records of its activity, knowing that it could avoid legal responsibility for the full scope of its illegal conduct."); Lilly v. Jamba Juice Co., No. 13-cv-02998-JST, 2014 WL 4652283, at *4 (N.D.Cal. Sept. 18, 2014) ("Adopting the Carrera approach would have significant negative ramifications for the ability to obtain redress for consumer injuries."); Carrera v. Bayer Corp., No. 12-2621, 2014 WL 3887938, at *3 (3d Cir. May 2, 2014) (Ambro, J., dissenting from denial of rehearing en banc) (explaining that Carrera may have gone too far where "a defendant's lack of records and business practices make it more difficult to ascertain the members of an otherwise objectively verifiable low-value class").

When faced with this counterargument, courts applying the heightened ascertainability approach have tended to emphasize that the plaintiff has the burden to satisfy Rule 23 and that the deterrence concern is therefore irrelevant. See, e.g., Hayes v. Wal-Mart Stores, Inc., 725 F.3d 349, 356 (3d Cir.2013) ("Rule 23's requirements that the class be administratively feasible to ascertain and sufficiently numerous to warrant class action treatment cannot be relaxed or adjusted on the basis of Hayes' assertion that Wal-Mart's records are of no help to him."). With respect, that response begs an important question. Why are affidavits from putative class members deemed insufficient as a matter of law to satisfy this burden? In other words, no one disputes that the plaintiff carries the burden; the decisive question is whether certain evidence is sufficient to meet it. Cf. Carrera, 2014 WL 3887938, at *1 (Ambro, J., dissenting from denial of rehearing en banc) ("Even if ... the ability to identify class members is a set piece for Rule 23 to work, how far we go in requiring plaintiffs to prove that ability at the outset is exceptionally important and requires a delicate balancing of interests.").

If not disputed, self-serving affidavits can support a defendant's motion for summary judgment, for example, and defendants surely will be entitled to a fair opportunity to challenge self-serving affidavits from plaintiffs. We are aware of only one type of case in American law where the testimony of one witness is legally insufficient to prove a fact. See U.S. Const., Art. III, § 3, cl. 1 ("No person shall be convicted of treason unless on the testimony of two witnesses to the same overt act, or on confession in open court."). There is no good reason to extend that rule to consumer class actions.

Given the significant harm caused by immunizing corporate misconduct, we believe a district judge has discretion to allow class members to identify themselves with their own testimony and to establish mechanisms to test those affidavits as needed.

4. Due Process Interest of the Defendant

Finally, courts have said the heightened ascertainability requirement is needed to protect a defendant's due process rights. Relying on cases about a defendant's right to "present every available defense," e.g., Lindsey v. Normet, 405 U.S. 56, 66, 92 S.Ct. 862, 31 L.Ed.2d 36 (1972), these courts have argued that the defendant must have a similar right to challenge the reliability of evidence submitted to prove class membership. See Carrera, 727 F.3d at 307 ("Ascertainability provides due process by requiring that a defendant be able to test the reliability of the evidence submitted to prove class membership."); Marcus, 687 F.3d at 594 ("Forcing BMW and Bridgestone to accept as true absent persons' declarations that they are members of the class, without further indicia of reliability, would have serious due process implications.").

We agree with the due process premise but not the conclusion. A defendant has a due process right to challenge the plaintiffs' evidence at any stage of the case, including the claims or damages stage. That does not mean a court cannot rely on self-identifying affidavits, subject as needed to audits and verification procedures and challenges, to identify class members. To see why, separate the two claims about a defendant's interest. It is certainly true that a defendant has a due process right not to pay in excess of its liability and to present individualized defenses if those defenses affect its liability. See Wal-Mart Stores, Inc. v. Dukes, 564 U.S. ___, 131 S.Ct. 2541, 2560-61, 180 L.Ed.2d 374 (2011). It does not follow that a defendant has a due process right to a cost-effective procedure for challenging every individual claim to class membership. Cf. American Express Co. v. Italian Colors Restaurant, 570 U.S. ___, 133 S.Ct. 2304, 2309, 186 L.Ed.2d 417 (2013) ("the antitrust laws do not guarantee an affordable procedural path to the vindication of every claim"). And we should not underestimate the ability of district courts to develop effective auditing and screening methods tailored to the individual case.

Whether a defendant's due process interest is violated depends on the nature of the class action, the plaintiff's theory of recovery, and the defendant's opportunity to contest liability and the amount of damages it owes. The due process question is not whether the identity of class members can be ascertained with perfect accuracy at the certification stage but whether the defendant will receive a fair opportunity to present its defenses when putative class members actually come forward. A district court can tailor fair verification procedures to the particular case, and a defendant may need to decide how much it wants to invest in litigating individual claims.

To see why this due process argument does not justify the heightened ascertainability requirement, consider three types of class actions. The first type is where the total amount of damages can be determined in the aggregate. A leading treatise provides an example:

Newberg on Class Actions § 12:2 (footnote omitted). In this situation, the identity of particular class members does not implicate the defendant's due process interest at all. The addition or subtraction of individual class members affects neither the defendant's liability nor the total amount of damages it owes to the class. See, e.g., In re Urethane Antitrust Litig., 768 F.3d 1245, 1269 (10th Cir.2014) (rejecting Seventh Amendment challenge to allocation of damages award among class members because defendant "has no interest in the method of distributing the aggregate damages award among the class members"); In re Pharmaceutical Industry Average Wholesale Price Litig., 582 F.3d 156, 197-98 (1st Cir.2009) (rejecting due process challenge to entry of class-wide judgment and award of aggregate damages); Allapattah Services, Inc. v. Exxon Corp., 333 F.3d 1248, 1258 (11th Cir.2003) ("[A] defendant has no interest in how the class members apportion and distribute a[n] [aggregate] damage [award] among themselves."), aff'd, 545 U.S. 546, 125 S.Ct. 2611, 162 L.Ed.2d 502 (2005); Hilao v. Estate of Marcos, 103 F.3d 767, 786 (9th Cir.1996) (noting that defendant's interest is "only in the total amount of damages for which it will be liable," not "the identities of those receiving damage awards"); Six (6) Mexican Workers v. Arizona Citrus Growers, 904 F.2d 1301, 1307 (9th Cir. 1990) ("Where the only question is how to distribute the damages, the interests affected are not the defendant's but rather those of the silent class members.").

The second type of class action is where the total amount of damages cannot be determined in the aggregate, but there is a common method of determining individual damages. (Most consumer fraud class actions fit this model.) The same treatise provides this example:

Newberg on Class Actions § 12:2 (footnote omitted). In this situation, the defendant's due process interest is implicated because the calculation of each class member's damages affects the total amount of damages it owes to the class. That's why the method of determining damages must match the plaintiff's theory of liability and be sufficiently reliable. See Comcast Corp. v. Behrend, 569 U.S. ___, 133 S.Ct. 1426, 1433, 185 L.Ed.2d 515 (2013). It's also why the defendant must be given the opportunity to raise individual defenses and to challenge the calculation of damages awards for particular class members. See Allapattah Services, 333 F.3d at 1259.

But neither of these requirements has any necessary connection to the heightened ascertainability requirement. Whether putative class members self-identify by affidavits simply does not matter. Suppose an employee files an affidavit falsely claiming that she worked 60 hours a week when in fact she worked only 50, or suppose a person files an affidavit falsely claiming to have been an employee. In either case, so long as the defendant is given a fair opportunity to challenge the claim to class membership and to contest the amount owed each claimant during the claims administration process, its due process rights have been protected.

The third type of class action is where the defendant's liability can be determined on a class-wide basis, but aggregate damages cannot be established and there is no common method for determining individual damages. In this situation, courts often bifurcate the case into a liability phase and a damages phase. See Butler v. Sears, Roebuck & Co., 727 F.3d 796, 800 (7th Cir.2013) ("a class action limited to determining liability on a class-wide basis, with separate hearings to determine — if liability is established — the damages of individual class members, or homogeneous groups of class members, is permitted by Rule 23(c)(4) and will often be the sensible way to proceed").

It has long been recognized that the need for individual damages determinations at this later stage of the litigation does not itself justify the denial of certification. See Schleicher v. Wendt, 618 F.3d 679, 685 (7th Cir.2010) ("The possibility that individual hearings will be required for some plaintiffs to establish damages does not preclude certification."); Pella Corp. v. Saltzman, 606 F.3d 391, 394 (7th Cir.2010) (per curiam); Arreola v. Godinez, 546 F.3d 788, 799-801 (7th Cir.2008); Carnegie v. Household Int'l, Inc., 376 F.3d 656, 661 (7th Cir.2004). Here again, using the heightened ascertainability requirement to deny class certification is not the only means, or even the best means, to protect the defendant's due process rights.

As long as the defendant is given the opportunity to challenge each class member's claim to recovery during the damages phase, the defendant's due process rights are protected. See Lilly v. Jamba Juice Co., No. 13-cv-02998-JST, 2014 WL 4652283, at *6 (N.D.Cal. Sept. 18, 2014) ("Defendants would certainly be entitled to object to a process through which a non-judicial administrator `ascertains' each applicant's class membership on the basis of the applicants' own self-identification, gives a defendant no opportunity to challenge that determination, and then racks up the defendant's bill every time an individual submits a form."); Johnson v. General Mills, Inc., 276 F.R.D. 519, 524 (C.D.Cal.2011) ("If Mr. Johnson establishes liability for the class, Defendants may challenge reliance and causation individually during a determination of damages, after the issues that are common have been litigated and resolved."); Godec v. Bayer Corp., No. 1:10-CV-224, 2011 WL 5513202, at *7 (N.D.Ohio Nov. 11, 2011) ("In any event, to the extent Bayer has individualized defenses, it is free to try those defenses against individual claimants.").4

In sum, the concern about protecting a defendant's due process rights does not justify the heightened ascertainability requirement. In all cases, the defendant has a right not to pay in excess of its liability and to present individual defenses, but both rights are protected by other features of the class device and ordinary civil procedure. Carrera itself appeared to recognize this rejoinder, but it pivoted to the argument discussed above about protecting absent class members. See 727 F.3d at 310 ("Because Bayer's total liability cannot be so affected by unreliable affidavits, Carrera argues Bayer lacks an interest in challenging class membership.... But ascertainability protects absent class members as well as defendants, so Carrera's focus on Bayer alone is misplaced." (citation omitted)). Carrera gave no other reason to think the heightened ascertainability requirement is needed to protect a defendant's due process rights. We can't think of one either.

Ultimately, we decline Direct Digital's invitation to adopt a heightened ascertainability requirement. Nothing in Rule 23 mentions or implies it, and we are not persuaded by the policy concerns identified by other courts. Those concerns are better addressed by a careful and balanced application of the Rule 23(a) and (b)(3) requirements, keeping in mind under Rule 23(b)(3) that the court must compare the available alternatives to class action litigation. District courts should continue to insist that the class definition satisfy the established meaning of ascertainability by defining classes clearly and with objective criteria. If a class is ascertainable in this sense, courts should not decline certification merely because the plaintiff's proposed method for identifying class members relies on affidavits. If the proposed class presents unusually difficult manageability problems, district courts have discretion to press the plaintiff for details about the plain-tiff's plan to identify class members. A plaintiff's failure to address the district court's concerns adequately may well cause the plaintiff to flunk the superiority requirement of Rule 23(b)(3). But in conducting this analysis, the district court should always keep in mind that the superiority standard is comparative and that Rule 23(c) and (d) permit creative solutions to the administrative burdens of the class device.

C. Commonality

Direct Digital's other primary challenge to the district court's certification order relates to the commonality requirement of Rule 23(a)(2). The district court found this requirement satisfied by a preponderance of the evidence, see Messner v. Northshore Univ. HealthSystem, 669 F.3d 802, 811 (7th Cir.2012), explaining that whether Instaflex has been clinically tested or scientifically formulated to relieve joint pain, improve flexibility, increase mobility, and support cartilage repair are questions common to the class. [See R. 89 at 2, 3-4]

Direct Digital argues that Mullins cannot satisfy the commonality requirement because his suit alleges that Instaflex is ineffective. The efficacy of a health product can never form the basis of a common question, Direct Digital argues, because efficacy depends on individual factors such as the severity of the consumer's pre-use medical condition, the consumer's pattern of use, and other potentially confounding variables such as the consumer's overall health, age, activity level, use of other drugs, and the like.

Direct Digital's objection fails because it has mischaracterized Mullins's theory of liability. Mullins does not claim that Instaflex was ineffective, ergo defendant is liable. He alleges that Direct Digital's statements representing that Instaflex has been "clinically tested" and "scientifically formulated" to relieve joint discomfort, improve flexibility, increase mobility, and repair cartilage are false or misleading because they imply there was scientific support for these claims but in fact no reasonable scientific expert would conclude that glucosamine sulfate (the primary ingredient in the supplement) has any positive effect on joint health. Mullins alleges that these statements would have misled a reasonable consumer. See Barbara's Sales, Inc. v. Intel Corp., 227 Ill.2d 45, 316 Ill.Dec. 522, 879 N.E.2d 910, 925-27 (2007) (reasonable consumer standard); accord, Suchanek v. Sturm Foods, Inc., 764 F.3d 750, 756-57 (7th Cir.2014) (discussing consumer fraud statutes in Illinois and other states). As the district court correctly concluded, this theory presents a common question: Were the statements false or misleading? This is a "common contention" that is "capable of classwide resolution" because the "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." Wal-Mart Stores, Inc. v. Dukes, 564 U.S. ___, 131 S.Ct. 2541, 2551, 180 L.Ed.2d 374 (2011). Nothing more is required to satisfy Rule 23(a)(2).

Of course the efficacy of the product can be relevant to that determination. If consumers experience the reduction or elimination of their symptoms, then that is evidence that the supplement does in fact relieve joint discomfort consistent with Direct Digital's representations. But that's not the focus of Mullins's theory of consumer fraud. What really matters under his theory is whether there is any scientific support for the assertions contained in the labels and advertising materials. In other words, Mullins's claims do not rise or fall on whether individual consumers experienced health benefits, due to the placebo effect or otherwise. They rise or fall on whether Direct Digital's representations were deceptive. See Suchanek, 764 F.3d at 756-57 (reversing district court's order denying class certification; commonality is satisfied where plaintiff's theory of liability turns on proving unfair or deceptive marketing and packaging of consumer product).

That's why even if Direct Digital were to prove that consumers experienced less joint pain because of a placebo effect (a theory Direct Digital appears to embrace on appeal), it could still be liable for consumer fraud. Consumers might have paid more than they otherwise would have because of the representations about clinical testing. Or they could have decided not to seek out better therapeutic alternatives because they believed Instaflex was addressing their underlying condition. See FTC v. QT, Inc., 512 F.3d 858, 862-63 (7th Cir.2008) (placebo effect is not a defense to consumer fraud where defendant has made specific claims about intended benefits; requiring truth in labeling leads to appropriate prices and ensures that consumers do not forgo better alternatives in reliance on the placebo). At any rate, we express no view on the merits of Mullins's allegations. The key point is that whether the representations were false or misleading is a common question suitable for class treatment, even if Instaflex relieved joint discomfort for some consumers.

III. Conclusion

Direct Digital raises a number of other, less developed objections to the district court's certification order. None of these issues would have justified granting an appeal under Rule 23(f), but we have considered them and find them without merit. Direct Digital has not demonstrated that the district court abused its discretion in certifying the class. The order of the district court granting class certification is AFFIRMED.