LEE, Circuit Judge

Theodore H. Frank (argued) and Melissa A. Holyoak, Center for Class Action Fairness, Hamilton Lincoln Law Institute, Washington, D.C., for Objector-Appellant.

Samuel Issacharoff (argued), New York, New York; Robert Klonoff, Portland, Oregon; Ariana J. Tadler, A.J. de Bartolomeo, and Brian R. Morrison, Tadler Law LLP, New York, New York; Adam J. Levitt and Amy E. Keller, DiCello Levitt Gutzler LLC, Chicago, Illinois; David Azar, Milberg Phillips Grossman LLP, Irvine, California; for Plaintiffs-Appellees.

Angela M. Spivey (argued), Alston & Bird LLP, Atlanta, Georgia, for Defendant-Appellee.

Mark Brnovich, Attorney General; Oramel H. Skinner, Solicitor General; Kate B. Sawyer, Assistant Solicitor General; Keena Patel, Assistant Attorney General; Office of the Attorney General, Phoenix, Arizona; Steve Marshall, Kevin G. Clarkson, Leslie Rutledge, Lawrence G. Wasden, Curtis T. Hill Jr., Daniel Cameron, Jeff Landry, Eric Schmitt, Dave Yost, Mike Hunter, Alan Wilson, and Ken Paxton, Attorneys General; as and for Amici Curiae Attorneys General of Arizona, Alabama, Alaska, Arkansas, Idaho, Indiana, Kentucky, Louisiana, Missouri, Ohio, Oklahoma, South Carolina, and Texas.

Before: John B. Owens and Kenneth K. Lee, Circuit Judges, and David A. Ezra, Senior District Judge.

LEE, Circuit Judge:

We can perhaps sum up this case as "How to Lose a Class Action Settlement in 10 Ways." The parties crammed into their settlement agreement a bevy of questionable provisions that reeks of collusion at the expense of the class members: Class counsel will receive seven times more money than the class members; an injunction touted by an expert as worth tens of millions of dollars appears worthless; the defendant agrees not to challenge the plaintiffs’ attorneys’ fees amount; any reduction in those fees by the court reverts to the defendant; and on and on.

While courts should not casually second-guess class settlements brokered by the parties, they should not greenlight them, either, just because the parties profess that their dubious deal is "all right, all right, all right." We reverse the district court's approval of the class settlement because the agreement raises a squadron of red flags billowing in the wind and begging for further review. We hold that under the newly revised Rule 23(e)(2) standard, courts must scrutinize settlement agreements — including post-class certification settlements — for potentially unfair collusion in the distribution of funds between the class and their counsel.

BACKGROUND

For many years, ConAgra, then-owner of Wesson Oil, labeled that product as "100% Natural." In 2011, Robert Briseño and others sued ConAgra, alleging that "100% Natural" was misleading because Wesson Oil contains ingredients made from genetically modified organisms ("GMOs").

Three years later, plaintiffs sought class certification. Relying on the expert report of Colin B. Weir, they argued that they overpaid for Wesson Oil based on the "100% Natural" label. ConAgra responded that its market research showed that less than 3 percent of consumers bought the product because of that label. Although Weir testified that hedonic regression could quantify the supposed price premium charged for that label, he did not try to calculate it at first.

Unsurprisingly, ConAgra then challenged the admissibility of Weir's report. Enlisting its own expert, ConAgra asserted that historical price data showed that the label did not affect the price of Wesson Oil. According to ConAgra, if the public had cared about the "100% Natural" claim, then the price of Wesson Oil should have declined after ConAgra removed that claim from the product's label. The district court agreed, denying plaintiffs’ first certification request.

Still hoping to strike oil, plaintiffs filed an amended motion for class certification. This time, however, they supplemented Weir's expert material with a supporting opinion by Dr. Elizabeth Howlett. Together, plaintiffs’ experts asserted that consumers paid a 2.28% price premium for the allegedly mislabeled products. Furthermore, Dr. Howlett suggested that a conjoint analysis could help determine how consumers value "GMO content." Plaintiffs, however, never submitted that conjoint analysis. ConAgra, again, sought to strike plaintiffs’ experts and opposed class certification.

This time, the court denied ConAgra's motions and certified a Rule 23(b)(3) damage class, though it refused to certify a 23(b)(2) injunctive class for lack of standing. ConAgra twice pursued Rule 23(f) interlocutory review of class certification. It lost both appeals and an attempt to seek certiorari. The parties began settlement negotiations shortly after that.

Meanwhile, ConAgra agreed to sell Wesson Oil to The J.M. Smucker Company in May 2017. About two months later, ConAgra voluntarily removed the disputed label, and stopped marketing Wesson products as "natural." ConAgra maintains that this litigation played no role in either decision. In early 2018, the Smucker deal hit an insurmountable regulatory jam. Undeterred, ConAgra sought a new suitor for Wesson. At the same time, it engaged in mediation with the certified class. The district court assigned Magistrate Judge McCormick to help the parties grease the wheels of justice, and they emerged with an agreement-in-principle in November 2018. A month later, ConAgra agreed to sell Wesson to Richardson International. The deal closed in February 2019. The next month, the parties proposed a settlement agreement.

ConAgra agreed to provide, in relevant part:

(a) $0.15 for each unit of Wesson Oils purchased to households submitting valid claim forms (to a maximum of thirty units without proof of purchase, and unlimited units with proof of purchase) (b) an additional fund of $575,000 to be allocated to New York and Oregon class members submitting valid claim forms, as compensation for statutory damages under those states’ consumer protection laws, and (c) an additional fund of $10,000 to compensate those in all classes submitting valid proof of purchase receipts more than thirty purchases, at $0.15 for each such purchase above thirty, with Class Counsel paying any non-funded claims (i.e., claims above the $10,000 provided by ConAgra) from any attorneys’ fees awarded in this case.

With a class of nearly 15 million consumers, ConAgra claimed that it theoretically exposed itself to nearly $67.5 million in claims if every consumer submitted a claim. (Spoiler alert: that never happens — not even close). The settlement agreement established a fund on a claims-made basis — i.e. , ConAgra would pay out for only those claims submitted by consumers. The settlement, however, did not require ConAgra to identify or provide direct notice to class members.

The settlement agreement also provided injunctive relief: Should ConAgra have seller's remorse and decide to reacquire the Wesson brand in the future, it agreed not to advertise or market Wesson Oil as "natural," unless the FDA permits the use of the term to describe oil derived from GMO seeds. Relying on Mr. Weir's analysis, the parties asserted that the "the value of the injunctive relief to the Classes" is $27 million.

Finally, the settlement stated that plaintiffs would request — and ConAgra would not contest — $6.85 million in attorneys’ fees and expenses. That amount would come directly from ConAgra and be separate from the class settlement fund. If the court, however, sliced the agreed-upon attorneys’ fees, that reduction would revert to ConAgra rather than the class.

The parties thus represented that their settlement could theoretically be worth over $100 million — around $95 million in value to the class ($67.5 million in potential payout and $27 million in injunctive relief value), along with another $6.85 million for the attorneys. Yet, when the dust settled, ConAgra shelled out less than $8 million, with a mere $1 million of that going to the class. Class counsel's fees swallowed $5.85 million, and expenses devoured another $978,671. Of the 15 million class members, barely more than one-half of one percent of them submitted a claim.

Only one class member opted out of the settlement. M. Todd Henderson, a law professor at the University of Chicago, objected to the settlement under Rule 23(e), arguing that attorneys hoarded 88% of the class's actual recovery. He asserted that our precedent required the court to treat the settlement as a constructive common fund (i.e. , the settlement effectively establishes one common fund to pay out both the class members and their counsel). Henderson also contended that the settlement's "clear sailing" provision (i.e. , ConAgra's refusal to challenge the agreed-upon attorney's fees) and "kicker" clause (i.e. , any reduction in fees reverting to ConAgra, not the class members) raised the specter of collusion. He also objected to the stipulated value of the injunctive relief, describing it as "illusory." Likewise, Henderson castigated Mr. Weir, stating that his failure to conduct a price comparison rendered his opinion unreliable. Plaintiffs sought final approval of the settlement in July 2019. Based on Mr. Weir's declaration, they valued ConAgra's label change at $19,080,000. They also contended that, if Wesson's new owner, Richardson, continued to refrain from labeling the product as "natural" for even a year , the value of the injunction would surge to $30.2 million. And for each year that Richardson did not label Wesson Oil as "natural," the class would obtain an annual benefit of over $11 million, according to Mr. Weir. Plaintiffs argued that the fee request "represent[ed] approximately 25.4% of the parties’ estimated value of the injunctive relief or 23% of Plaintiffs’ conservative estimate[ ]." They also calculated their own lodestar fee at around $11.499 million.

When the claims deadline passed, class members made 97,880 timely claims for $418,919, a shadow of the $67.5 million potential liability that ConAgra touted in seeking approval of the settlement. Even with separately funded pools for New York and Oregon, ConAgra would pay class members a maximum of $993,919. Out of a class of 15 million consumers, fewer than 100,000 would receive a single cent.

The district court held its final fairness hearing in October 2019. Henderson and class counsel remained loggerheads on almost every issue. The parties disputed whether Henderson as an objector, or plaintiffs as proponents, bore the burden of establishing that the settlement satisfied Rule 23(e). Henderson argued that the "kicker" demonstrated ConAgra's willingness to settle for roughly $8 million, and that class counsel bargained away absent class members’ rights in exchange for much of the settlement. The district court disagreed.

Rejecting Henderson's motion to strike Mr. Weir's expert report, the district court explained that "[h]aving one expert's opinion — however purportedly flawed — on the value of that injunction helps the Court develop its own view." Despite recognizing the parties’ "vigorous dispute over the precise valuation," it still found that "the injunction adds at least some value to the amount offered in settlement." It continued, "even if there were no injunctive relief, the [c]ourt would likely find that the amount offered in settlement was fair and reasonable given the likely obstacles to Plaintiffs recovering [at trial]."

The district court went on to evaluate the settlement for fairness under our decision in Staton v. Boeing Co. , 327 F.3d 938, 959 (9th Cir. 2003). It explained that the length and nature of the suit allowed both sides to evaluate the costs and benefits of protracted litigation, supplemented by the recommendation of a court-appointed mediator. Again, the court's concerns over the merits of the plaintiffs’ suit heavily influenced its analysis. The court, however, stopped short of conducting a Rule 23(e) inquiry. Instead, it merely held that "[t]here is substantial overlap between [ Rule 23(e)(2) ] factors and the Staton factors."

The court, "rel[ying] on the lodestar method," found class counsel's $6.85 million reasonable given the lodestar amount of "nearly $11.5 million." Indeed, the court appeared impressed that "Defendant [was] willing to pay anything at all given the many liability and damages issues this case has had from the beginning." The court also pointed to "the amount of hours reasonably spent on the litigation, counsel's efforts in litigating this years-long complex action, the results achieved, and the risks inherent in continued litigation."

It also emphasized the "substantial" nature of "the [$0.15] per-unit award," given that it had restricted relief to "only the portion of [the] premium attributable to consumers’ belief that ‘100% natural’ meant that the products were GMO-free." It thus concluded that "[t]he settlement amount offered provide[d] an immediate and tangible benefit to class members and eliminate[d] the risk that they could receive less than that amount, or nothing at all, if litigation continued."

And while the court "appreciate[d] Objector's high-level concerns regarding an apparent trend [sic] toward class action settlements disproportionately benefitting attorneys," it was "not persuaded" that "the disproportionate attorney fee award under the settlement render[ed] the entire settlement unfair." Rather, the court maintained that "the record in this case sufficiently ‘dispel[s] the possibility that class counsel bargained away a benefit to the class in exchange for their own interests.’ " Henderson timely appealed, and we have jurisdiction under 28 U.S.C. § 1291.

STANDARD OF REVIEW

We review for abuse of discretion a district court's decision to approve a class action settlement. Roes, 1–2 v. SFBSC Mgmt., LLC , 944 F.3d 1035, 1043 (9th Cir. 2019). "A [district] court abuses its discretion when it fails to apply the correct legal standard or bases its decision on unreasonable findings of fact." Nachshin v. AOL, LLC , 663 F.3d 1034, 1038 (9th Cir. 2011). Even so, "[a]ppellate review of a settlement agreement is generally ‘extremely limited.’ " Dennis v. Kellogg Co. , 697 F.3d 858, 864 (9th Cir. 2012) (citing Hanlon v. Chrysler Corp. , 150 F.3d 1011, 1026 (9th Cir. 1998) ). We, however, "hold district courts to a ‘higher procedural standard when making [a] determination of substantive fairness.’ " Roes, 1–2 , 944 F.3d at 1043 (quoting Allen v. Bedolla , 787 F.3d 1218, 1223 (9th Cir. 2015) ). See also Dennis , 697 F.3d at 864 (explaining that " ‘[t]o survive appellate review, the district court must show it has explored comprehensively all factors,’ and must give ‘a reasoned response’ to all non-frivolous objections" (citing Officers for Justice v. Civil Serv. Comm'n , 688 F.2d 615, 624 (9th Cir. 1982) )). Thus, this court "will rarely overturn an approval of a" compromised settlement "unless the terms of the agreement contain convincing indications that ... self-interest rather than the class's interest in fact influenced the outcome of the negotiations." Staton , 327 F.3d at 960 ; Allen , 787 F.3d at 1223.

"We also review for abuse of discretion a district court's award of fees and costs to class counsel, as well as its method of calculation." Lobatz v. U.S. Cellular of Cal., Inc. , 222 F.3d 1142, 1148–49 (9th Cir. 2000).

ANALYSIS

I. The district court erred by failing to apply the new Rule 23(e)(2), which requires courts to scrutinize attorneys’ fee arrangements.

A. Under the newly revised Rule 23(e)(2), courts should apply the Bluetooth factors even for post-class certification settlements.

Rule 23(e) imposes on district courts an independent obligation to ensure that any class settlement is "fair, reasonable, and adequate," accounting for the interests of absent class members. Fed. R. Civ. P. 23(e)(2). Likewise, we recognize "an independent obligation to ensure that [any attorneys’ fee] award, like the settlement itself, is reasonable, even if the parties have already agreed to an amount." In re Bluetooth Headset Products Liability Litigation , 654 F.3d 935, 941 (9th Cir. 2011) ; see also Staton , 327 F.3d at 960–64. Indeed, settlement agreements "warrant special attention when the record suggests that settlement is driven by fees; that is, when counsel receive a disproportionate distribution of the settlement." Hanlon , 150 F.3d at 1021. Regardless of "whether the attorneys’ fees come from a common fund or are otherwise paid, the district court must exercise its inherent authority to assure that the amount and mode of payment of attorneys’ fees are fair and proper." Zucker v. Occidental Petroleum Corp. , 192 F.3d 1323, 1328 (9th Cir. 1999).

Before the 2018 amendment, Rule 23 stated that class settlements should be "fair, reasonable, and adequate" but did not elaborate. Like our sister circuits, we filled in the gaps, instructing courts to consider the following factors (sometimes called " Hanlon factors" or " Staton factors") in assessing whether a settlement is "fair, reasonable, and adequate":

[T]he strength of the plaintiffs’ case; the risk, expense, complexity, and likely duration of further litigation; the risk of maintaining class action status throughout the trial; the amount offered in settlement; the extent of discovery completed and the stage of the proceedings; the experience and views of counsel; the presence of a governmental participant; and the reaction of the class members to the proposed settlement.

Hanlon , 150 F.3d at 1026 ; Staton , 327 F.3d at 959 (internal citations and quotations omitted). Admittedly, we never explicitly mandated consideration of the terms of attorneys’ fees in the Hanlon / Staton factors.

On the other hand, we have recognized the risks in allowing counsel to bargain on behalf of the entire class, especially pre-class certification when counsel may try to strike a quick settlement on behalf of the class. See Staton , 327 F.3d at 960. In Bluetooth , we explained that courts should scrutinize agreements for "subtle signs that class counsel have allowed pursuit of their own self-interests ... to infect the negotiations." 654 F.3d at 947. We identified three of those signs: (1) "when counsel receive[s] a disproportionate distribution of the settlement"; (2) "when the parties negotiate a ‘clear sailing arrangement,’ " under which the defendant agrees not to challenge a request for an agreed-upon attorney's fee; and (3) when the agreement contains a "kicker" or "reverter" clause that returns unawarded fees to the defendant, rather than the class. Id . In reviewing settlements struck before class certification, district courts must apply these so-called Bluetooth factors to smoke out potential collusion.

Last year, we noted that " Bluetooth therefore left open a question no subsequent case has answered," whether its heightened inquiry applies to post-class certification settlements. Campbell v. Facebook, Inc. , 951 F.3d 1106, 1125–26 (9th Cir. 2020). We now answer that question: indeed, it does.

That answer flows from the revised Rule 23(e). In December 2018, Congress and the Supreme Court amended Rule 23(e) to set forth specific factors to consider in determining whether a settlement is "fair, reasonable, and adequate," including:

23(e)(2)(C): [Considering whether] the relief provided for the class is adequate, taking into account:

(i) the costs, risks, and delay of trial and appeal;

(ii) the effectiveness of any proposed method of distributing relief to the class , including the method of processing class-member claims;

(iii) the terms of any proposed award of attorney's fees , including timing of payment; and

(iv) any agreement required to be identified under Rule 23(e)(3).

23(e)(2)(D): the proposal treats class members equitably relative to each other.

Fed. R. Civ. P 23(e)(2)(C)–(D) (emphasis added).

Under this revised text, district courts must now consider "the terms of any proposed award of attorney's fees" when determining whether "the relief provided for the class is adequate." Fed. R. Civ. P. 23(e)(2)(C)(iii). While none of our sister circuits has yet directly addressed what this provision specifically requires, the plain language indicates that a court must examine whether the attorneys’ fees arrangement shortchanges the class. In other words, the new Rule 23(e) makes clear that courts must balance the "proposed award of attorney's fees" vis-à-vis the "relief provided for the class" in determining whether the settlement is "adequate" for class members.

Nothing in the Rule's text suggests that this requirement applies only to pre-certification settlements. Congress required courts to scrutinize attorney's fees, even if the settlement occurred after class certification. And for good reason, too: The specter of collusion still casts a long shadow over post-class certification settlements when they involve divvying up funds between class members and class counsel.

We have observed that courts should scrutinize pre-class certification settlements because plaintiffs’ counsel may collude with the defendant to strike a quick settlement without devoting substantial resources to the case. See, e.g. , Hanlon , 150 F.3d at 1026 (adopting other circuits’ "more probing inquiry" for "settlement approval that takes place prior to formal class certification"). The potential for collusion reaches its apex pre-class certification because, among other things, (1) the court has not yet approved class counsel, who would owe a fiduciary duty to the class members; and (2) plaintiffs’ counsel has not yet devoted substantial time and money to the case, and may be willing to cut a quick deal at the expense of class members’ interests. See generally In re General Motors Corp. Pick-Up Truck Fuel Tank Prods. Liab. Litig. , 55 F.3d 768, 788–90 (3d Cir. 1995).

In contrast, by the time a court has certified a class — the theory goes — the parties have vigorously litigated the dispute, reducing the chance that class counsel will settle on the cheap for a quick buck. By devoting substantial time and resources to the case, class counsel has skin in the game, guaranteeing his or her interest in maximizing the size of the settlement fund. Likewise, because a district court has appointed class counsel who owes a fiduciary duty to the class members, class counsel would be ethically forbidden from sacrificing the class members’ interests. See Allen , 787 F.3d at 1223.

All of this is true — but also beside the point. Simply put, class certification does not cleanse all sins, especially when it involves potential collusion over divvying up funds between class counsel and the class (rather than the size of the settlement fund or relief). See Staton , 327 F.3d at 972 n.22 (recognizing "the inherent tensions among class representation, defendant's interests in minimizing the cost of the total settlement package, and class counsel's interest in fees").

Even after a court has certified a class, class counsel still has the incentive to conspire with the defendant to reduce compensation for class members in exchange for a larger fee. A defendant goes along with this collusion because it cares only about the total payout, not the division of funds between class and class counsel. After all, a defendant, no matter if a class has been certified, has "no reason to care about the allocation of its cost of settlement between class counsel and class members." Pearson v. NBTY, Inc. , 772 F.3d 778, 783 (7th Cir. 2014) (Posner, J.). Instead, "all it cares about as a rational maximizer of its net worth is the bottom line — how much the settlement is likely to cost it." Id.

Consider this example. What would any rational defendant do if faced with these two settlement options: (1) establish a $10 million fund for class members and pay $3 million in fees to class counsel for a total payout of $13 million, or (2) set up a $7 million fund and pay $4 million to class counsel for a total payout of $11 million. A defendant would choose the second option because it would save $2 million, even though it shortchanges class members. Nothing about class certification can make a defendant care more about its opponents than its own bottom line.

Put another way, a post-class certification settlement only ensures that the parties litigated aggressively to arrive at an adequate total fund size; it does not, however, address the inherent incentives that tempt class counsel to elevate his or her own interest over those of the class members. As one prominent academic who supports class actions put it, "the profit motive will give class action lawyers incentives to do sneaky things, just like it gives businesses incentives to do sneaky things." Brian T. Fitzpatrick, The Conservative Case for Class Actions 72 (2019). The potential for this type of collusion is no hoax — it is real, whether a class has been certified or not.

Congress sought to end this practice by changing the text of Rule 23(e)(2)(C). We thus now hold that courts must apply Bluetooth ’s heightened scrutiny to post-class certification settlements in assessing whether the division of funds between the class members and their counsel is fair and "adequate." Fed. R. Civ. P. 23(e)(2)(C).

B. The settlement had all the hallmarks of a potentially collusive settlement giving short shrift to the class.

Despite holding its final approval hearing after the new version of Rule 23(e)(2) took effect, the district court did not apply it and instead relied on this court's Staton factors. True, as the district court recognized, many of the Staton factors fall within the ambit of the revised Rule 23(e). But Congress provided district courts with new instructions — such as analyzing the "terms of the settlement" and "terms of any proposed award of attorney's fees" — that require them to go beyond our precedent. Although we need not decide whether a district court always abuses its discretion by applying the judicially manufactured factors in Staton and Hanlon , we must follow the law that Congress enacted. And that means scrutinizing the fee arrangement for potential collusion or unfairness to the class. Fed. R. Civ. P. 23(e)(2)(C)(iii).

Perhaps because our court had not clarified whether the Bluetooth factors apply to post-class certification settlements, the district court did not adequately scrutinize the fee arrangement for collusion. It instead noted that the fee request amounted to less than half of the lodestar amount. But the lodestar amount alone cannot tell us if the requested fees are reasonable. Counsel may have frittered away hours on pointless motions or unnecessary discovery, padding the lodestar. See Chambers v. Whirlpool Corp. , 980 F.3d 645, 665 (9th Cir. 2020) ("asymmetrical nature of discovery in class actions ... can lead to excessive billing"). Or maybe counsel devoted tremendous hours but achieved very little for the class. See id. at 667. And in any event, even attorneys’ fees based on a reasonable percentage of an unreasonable number of hours are still unreasonable.

We hold that district courts must apply the Bluetooth factors to scrutinize fee arrangements — even in post-class certification settlements — to determine if collusion may have led to class members being shortchanged. The class settlement here features all three red flags of potential collusion that we warned about in Bluetooth.

First, plaintiffs’ counsel "receive[d] a disproportionate distribution of the settlement." Bluetooth , 654 F.3d at 947. The lion's share of the money — almost $7 million — will end up in the pockets of attorneys, while the class receives relative scraps, less than a million dollars. So little goes to the class members in a claims-made settlement, such as this one, because the redemption rate is notoriously low, especially when it involves small-ticket items. The redemption rate shrinks even further if the settlement, as here, provides for no direct notice to class members. Despite ConAgra's claim that it could potentially provide over $95 million in payments and value to the class members, they ended up receiving only about 1% of that touted amount. And that was no surprise, given how the parties knowingly structured the settlement. This gross disparity in distribution of funds between class members and their class counsel raises an urgent red flag demanding more attention and scrutiny.

Second, the parties agreed to a "clear sailing arrangement" in which ConAgra agreed not to challenge the agreed-upon fees for class counsel. This flashes yet another red flag under Bluetooth . Id . A clear sailing provision signals the potential that a defendant agreed to pay class counsel excessive fees in exchange for counsel accepting a lower amount for the class members. Id. at 949. Indeed, the "very existence of a clear sailing provision increases the likelihood that class counsel will have bargained away something of value to the class." Id. at 948 (quoting Weinberger v. Great N. Nekoosa Corp. , 925 F.2d 518, 525 (1st Cir. 1991) ). When faced with a clear sailing provision, courts thus have a "heightened duty to peer into the provision and scrutinize closely the relationship between attorney's fees and benefit to the class, being careful to avoid awarding ‘unreasonably high’ fees simply because they are uncontested." Id.

Finally, the agreement contains a "kicker" or "reverter" clause in which ConAgra, not the class members, receives the remaining funds if the court reduces the agreed-upon attorneys’ fees. Id . We identified this, too, in Bluetooth , as a warning sign. The reason is obvious: If ConAgra is content to pay nearly $7 million to class counsel but the court finds the full amount unreasonable, there is no plausible reason why the class should not benefit from the spillover of excessive fees. As we warned in Bluetooth , "[u]nless the district court is able to conclude that in this particular case, a kicker provision is in the class’ best interest as part of the settlement package, the kicker makes it less likely that the settlement can be approved if the district court determines the clear sailing provision authorizes unreasonably high attorneys’ fees." Id. at 949.

Here, the parties managed to run afoul of all three Bluetooth factors. That also raises the question of whether the parties colluded to prevent any direct challenge to excessive fees. Typically, class members can challenge an excessive fee award under Rule 23(h). But when parties agree to a "kicker," a 23(h) challenge cannot increase class recovery because the excessive fees wind up back in the defendant's pockets. That means that a class member may not have standing to object to the excessive fees because any action taken by the court would not redress the class member's purported injury. Meanwhile, by agreeing to the "clear sailing" clause, the defendant has also waived its right to challenge the attorneys’ fees, foreclosing any action under Rule 23(h). See Pearson , 772 F.3d at 786 (describing this combination as "a gimmick for defeating objectors").

We stress that nothing in this opinion suggests that courts should unnecessarily meddle in class settlements negotiated by the parties or that courts have a duty to maximize the settlement fund for class members. Far from it. We instead follow the rules of our involvement in the class action process as set by Congress. Under those rules, the parties can agree on any "fair, reasonable, and adequate" settlement amount. Fed. R. Civ. P. 23(e). Nor do we seek to make any of the identified signs of collusion an independent basis for withholding settlement approval. Disproportionate fee awards, clear sailing agreements, and kicker clauses all may be elements of a good deal. But, as we explained in Bluetooth , they may also signal a collusive settlement, and district courts must scrutinize them where they appear. And here, the parties did more than just check every Bluetooth box; their settlement presented a Murderers’ Row of provisions out of left field that seemingly favor class counsel and the defendant at the expense of the class members. The district court thus should give a hard look at the settlement agreement to ensure that the parties have not colluded at class members’ expense.

II. The district court erred by failing to approximate the value of the injunction.

Next, we address Henderson's argument that the district court erred by failing to recognize the settlement's injunctive relief as worthless. Rather than attempt to quantify the value of the injunctive relief, it instead concluded that it had "some" value. We agree with Henderson that this constitutes reversible error. See Roes, 1–2 , 944 F.3d at 1055 (explaining that a district court must either quantify and explain the value of injunctive relief or exclude it from calculations).

We go further and also hold that the district court erred by placing even "some value" on the injunction because it was, and is, virtually worthless. A district court abuses its discretion when it approves a settlement despite "no evidence that the relief afforded by [a] settlement has any value to the class members, yet to obtain it they had to relinquish their right to seek damages in any other class action." Koby v. ARS Nat. Servs., Inc. , 846 F.3d 1071, 1079 (9th Cir. 2017).

Certainly, "the relief provided to the class cannot be assessed in a vacuum." Campbell , 951 F.3d at 1123. After all, a "class [does] not need to receive much for [a] settlement to be fair [when] the class [gives] up very little." Id. at 1124. And "[i]n evaluating what class members relinquished in [a] settlement, [courts] must also consider whether class members were required to release claims that were more meritorious than the theories Plaintiffs pursued in [the present] litigation." Id . at 1124.

But, critically, we also find illusory any injunction that "does not obligate [the bound party] to do anything it was not already doing ... voluntarily ... for its own business reasons ... not because of any court-or settlement-imposed obligation. . . [when a defendant] would therefore be unlikely to revert back to its old ways regardless of whether the settlement contained the stipulated injunction." Koby , 846 F.3d at 1080. ConAgra would like us to believe that, by agreeing to the settlement's injunctive relief, it bound its own hands and threw away the key. In reality, it did that years before reaching an agreement with the class.

Under the settlement, ConAgra agreed to refrain from marketing Wesson Oil as "100% Natural." That sounds great, except that ConAgra already abandoned that strategy in 2017 — two years before the parties hammered out their agreement — for reasons it claims were unrelated to this or any other litigation. Even worse, ConAgra's promise not to use the phrase "100% Natural" on Wesson Oil appears meaningless because ConAgra no longer owns Wesson Oil. In reality, this promise is about as meaningful and enduring as a proposal in the Final Rose ceremony on the Bachelor . Simply put, Richardson — the new owner of Wesson Oil — can resume using the "100% Natural" label at any time it wishes, thereby depriving the class of any value theoretically afforded by the injunction. ConAgra thus essentially agreed not to do something over which it lacks the power to do. That is like George Lucas promising no more mediocre and schlocky Star Wars sequels shortly after selling the franchise to Disney. Such a promise would be illusory. Granted, ConAgra also promised to abide by the injunction if it reacquires Wesson. Yet at the fairness hearing, ConAgra's counsel emphasized the company's lack of interest in buying back the brand. Indeed, ConAgra's attempts to sell Wesson predated the parties’ first mediation. Thus, for the injunction to have any value, ConAgra would not only need to abandon its long-term corporate strategy, purchase the brand, possibly at a premium, but then also redesign and execute a brand-new labeling and marketing campaign. We find this unlikely and speculative at best.

We do not question the district court's view of the plaintiffs’ likelihood of success in future litigation. But even if the class gave up very little, it has a right to receive something in exchange. Here, they did not because the injunctive relief is practically worthless.

Nothing in the testimony of the plaintiffs’ expert, Mr. Weir, convinces us otherwise. As Professor Fitzpatrick of Vanderbilt Law School explained, "[n]onmonetary relief is much more difficult to put a number on than a pot of cash is, and lawyers may give the court rosy numbers to justify fee awards." Fitzpatrick, The Conservative Case for Class Actions 71. Here, the lawyers relied on an expert to provide a rosy number untethered to reality. Courts must "stamp [ ] out" such attempts. Id. at 72.

Mr. Weir concluded that the annual value of the injunctive relief is a staggering $11.54 million . In other words, ConAgra had been charging a price premium — to the aggregate tune of $11.54 million a year — for the "100% Natural" label, and, because of the injunction, that full amount accrues to the class each year. So, according to Mr. Weir, this settlement is essentially minting money to the tune of eight figures each year for the class members. Despite ample opportunity, he never tested his theory. That failure is more telling because he had a real-life example he could have examined: ConAgra dropped the "100% Natural" label years ago, so he could have studied whether that led to the removal of the price premium. Mr. Weir claims other real-life market factors made performing a post-labelling price-check futile. But that admission appears to doom his own expert conclusion that the label led to a price premium — if he cannot separate other market factors to figure out a price premium in real-life, how can he do so in a hypothetical study that he did not conduct? In short, Mr. Weir effectively admits that his expert testimony turns on "unverifiable evidence." See Domingo ex rel. Domingo v. T.K. , 289 F.3d 600, 607 (9th Cir. 2002). Because Mr. Weir's testimony is unverifiable, it is ultimately worth as little as the settlement's injunctive relief.

III. Erie does not preclude the application of Rule 23(e)(2).

We next address — and reject — appellees’ argument that the Erie doctrine precludes the application of Rule 23(e) to a class settlement where state substantive law governs attorney's fees in fee-shifting cases.

Every year, tens of thousands of first year law students learn about Erie R. Co. v. Tompkins , and, soon after that, they become second-and-third-year law students with poor understanding of the doctrine. 304 U.S. 64, 58 S. Ct. 817, 82 L.Ed. 1188 (1938). Some of these students take the bar exam, fail, and try again or do other things. Others pass and become attorneys who still do not understand Erie . Granted, the difference between substantive and procedural law is sometimes fuzzy. While we seek to provide clarity when an Erie question presents itself to this court, we leave most of the work to professors like M. Todd Henderson and Samuel Issacharoff. That said, a "Federal Rule of Procedure is not valid in some jurisdictions and invalid in others — and valid in some cases and invalid in others — depending [solely] upon whether its effect is to frustrate [state law]." Shady Grove Orthopedic Assocs., P.A. v. Allstate Ins. Co. , 559 U.S. 393, 409, 130 S.Ct. 1431, 176 L.Ed.2d 311 (2010).

In any event, Henderson challenges settlement fairness under Rule 23(e), rather than an award of attorney's fees under Rule 23(h). Thus, Erie ’s effect on fee-shifting law, if it even has one, is simply not implicated in this appeal. Instead, this case concerns Congress's simple command that all federal district courts must withhold approval of any class settlement, absent a finding that it is "fair, reasonable, and adequate." Fed. R. Civ. P. 23(e)(2).

IV. The district court did not err by determining that Henderson failed to rebut its own conclusion that the settlement satisfied Rule 23(e)(2).

Finally, we address Henderson's argument that the district court improperly shifted to him the burden of rebutting the settlement's fairness, reasonableness, and adequacy at the fairness hearing. Contrary to his contentions, however, the district court never required Henderson to show that the settlement was "clearly inadequate."

Rule 23(e)(2) assumes that a class action settlement is invalid. See Roes, 1–2 , 944 F.3d at 1049 n.12 (explaining that "[a] presumption of fairness ... is very likely inappropriate under the standards now codified in Rule 23(e)(2)"). It also instructs that "[i]f the proposal would bind class members, the court may approve it only after a hearing and only on finding that it is fair, reasonable, and adequate." Fed. R. Civ. P. 23(e)(2). As class counsel points out, this is exactly what the district court tried to do. The record demonstrates that the district court conducted its own independent analysis, and then considered, and dismissed, Henderson's objections.

In Roes, 1–2 , we reversed a district court that applied a " ‘presumption that [a] settlement is fair and reasonable ’ " to a pre-certification class that "is the product of arms-length negotiations." 944 F.3d at 1049. Noting that this court has "never endorsed applying a broad presumption of fairness, but ha[s] actually required that courts do the opposite ... when it comes to settlements negotiated prior to class certification," we held that "the district court failed to apply the correct legal standard and to conduct the searching inquiry required, thereby abusing its discretion." Id . Critically, we explained that our decision turned on the amended text of Rule 23(e)(2). Id . at 1049 n.12 (" Rule 23(e)(2) now identifies ‘whether ... the proposal was negotiated at arm's length’ as one of four factors that courts must consider and does not suggest that an affirmative answer to that one question creates a favorable presumption on review of the other three") (citing Fed. R. Civ. P. 23(e)(2)(B) ). In contrast, though the district court did allude to a presumption of fairness, it confined it to its analysis of one of the Staton factors. It did not shift the burden to Henderson. Henderson's objection stems from the following textual juxtaposition:

In most situations, unless the settlement is clearly inadequate, its acceptance and approval are preferable to lengthy and expensive litigation with uncertain results.’ Nat'l Rural Telecommunications , 221 F.R.D. at 526 (quoting 4 A Conte & H. Newberg, Newberg on Class Actions , § 11:50 at 155 (4th ed. 2002)). The Objection fails to persuade the Court that this

Settlement Agreement is clearly inadequate.

Were that the crux of the district court's ruling, Henderson would have a point. But the district court already concluded, albeit improperly, that the settlement satisfied Rule 23(e)(2). Moreover, an assumption of invalidity does not demand disfavoring settlement. Quite the opposite, "we have repeatedly noted that ‘there is a strong judicial policy that favors settlements, particularly where complex class action litigation is concerned.’ " Allen , 787 F.3d at 1223 (quoting In re Syncor ERISA Litig ., 516 F.3d 1095, 1101 (9th Cir. 2008) (citing Class Plaintiffs v. City of Seattle , 955 F.2d 1268, 1276 (9th Cir. 1992) )). We do not intend to revisit that policy.

CONCLUSION

"Two Virginians and an immigrant walk into a room/ diametrically opposed/ foes/ They emerge with a compromise/ Having opened doors that were previously closed/ Bros/ ... No one else was in the room where it happened ... No one really knows how the game is played/ The art of the trade/ How the sausage gets made/ We just assume that it happens/ But no one else is in the room where it happens."

Hamilton: An America Musical (2016).

Though that process suffices for political compromise and even most settlements, it does not for class action settlements. Because they impose binding judgments on absent class members, federal courts must approve class action settlements and ensure that they are fair, reasonable, and adequate. And under the newly enacted Rule 23(e), federal courts must scrutinize attorneys’ fees for potential collusion that shortchanges the class, even in post-class certification settlements. We REVERSE the district court's approval of the settlement, and REMAND this case for further proceedings consistent with this opinion.