*1071 Following a data breach at Home Depot, the information for tens of millions of credit cards was stolen, and a class of banks who issued the cards sued Home Depot to recover their resulting losses. Home Depot eventually settled with the class. As part of the settlement, Home Depot agreed to pay the reasonable attorney's fees of Class Counsel. The agreement specified that the attorney's fees would be paid separate from and in addition to the class fund, but the parties left the amount of those fees undetermined.

The District Court awarded Class Counsel $15.3 million in fees. It reached this award using the lodestar method, finding Class Counsel's hours to be reasonable and applying a multiplier of 1.3 to account for the risk the case presented. The Court also used the percentage method as a cross-check to ensure the amount of fees was reasonable.

On appeal, Home Depot argues that the District Court abused its discretion by applying a multiplier and by compensating Class Counsel for certain time spent on the case-namely, the substantial time spent litigating about a private dispute-resolution process separate from the litigation. Home Depot also says that the District Court's order is not capable of meaningful review. For its part, Class Counsel brings a conditional cross-appeal taking issue with how the District Court conducted the percentage cross-check.

The main issue underlying the appeal is whether the fee arrangement outlined in the settlement should be characterized as a constructive common fund or as a fee-shifting contract. We hold that this is a contractual fee-shifting case, and the constructive *1072 common-fund doctrine does not apply. Once we identify the proper legal framework, the parties' challenges are more easily resolved. We affirm the District Court's decision in all respects except one: it was an abuse of discretion to use a multiplier to account for risk in a fee-shifting case.

I.

Disputes over attorney's fees are fact-intensive inquiries. As such, a thorough review of the facts is necessary to decide this case.

A.

In 2014, Home Depot experienced a massive data breach. It started when hackers installed malware on Home Depot's self-checkout kiosks. The malware would siphon off the personal financial information of customers who paid at the kiosks using a credit or debit card. The hackers then made this information, including names, card numbers, expiration dates, and security codes, available for sale on a black-market website. Approximately fifty-six million cards were compromised. It did not take long for a large number of fraudulent transactions to occur using the stolen information.

A flood of lawsuits followed. Consumers whose personal information was stolen and banks that issued the compromised cards filed over 50 class actions. The United States Judicial Panel on Multidistrict Litigation consolidated these cases in the Northern District of Georgia, where the District Court split the litigation into two tracks: one for the consumers and one for the banks. This appeal arises from the bank track. 1 The District Court appointed Class Counsel to manage the sprawling litigation and ordered Class Counsel to submit quarterly reports to the Court in camera showing the hours billed and expenses incurred.

The banks filed a consolidated complaint accusing Home Depot of failing to secure its data. They brought claims for negligence and negligence per se on behalf of a national class, and for violations of state consumer-protection statutes on behalf of eight state-specific subclasses. They alleged that, as a result of the data breach, they were forced to cancel and reissue the compromised cards, investigate claims of fraudulent activity, and reimburse customers for fraudulent charges (among other things). The banks sought monetary damages for the cost of these responses, as well as declaratory and injunctive relief to force Home Depot to improve its security measures.

Home Depot moved to dismiss the complaint on numerous grounds. In the interim, and at the urging of the District Court, the parties proceeded with preliminary discovery. After the District Court ruled on the motion to dismiss, denying it in part, Home Depot answered the complaint. Shortly afterwards, the District Court stayed further action in the case pending settlement negotiations.

B.

While the litigation unfolded, another process played out that is central to this appeal: the card-brand recovery process. The card brand recovery process is essentially a private dispute-resolution arrangement between Home Depot, the banks, and the card brands (e.g., Visa and Mastercard). It's separate from the litigation, and instead is based on contracts with merchants (like Home Depot) that outline *1073 the terms for accepting credit and debit cards as payment. The process is relevant to this appeal because Class Counsel took issue with events that occurred in the card-brand recovery process that affected the class action, and Home Depot argues that Class Counsel should not be compensated for the substantial time spent pursuing the matter.

Here's how the card-brand recovery process works: The card brands, Visa and MasterCard, have contracts with the banks that issue their branded cards to customers. In turn, the banks have contracts with the merchants who accept the cards as payment. 2 These contracts include regulations for protecting payment-card data against the threat of a data breach. The contracts also establish procedures for merchants to reimburse the banks for losses in the event that card information is compromised in a data breach. As part of the procedure, the card brands investigate to determine if a breach occurred and the financial impact of the breach. The card brands then impose assessments on the merchant to reimburse the banks for their losses.

That process is what happened in this case. Following the breach, Visa and Mastercard together assessed $120 million against Home Depot to be paid to banks. 3 Home Depot had the option to challenge the assessment and possibly pay a reduced amount. In fact, Mastercard could not recall when a merchant had ever paid the full amount. But instead of fighting the assessments, Home Depot offered to pay the full amount plus a premium.

Home Depot did so in exchange for the banks releasing their claims against Home Depot. Normally, when a merchant pays these assessments, it does not include a release from liability. In Home Depot's view, if it was going to pay the assessments, it ought to be released from liability too. So Home Depot reached out to Visa, Mastercard, and some of the larger banks to negotiate a deal. These banks were putative class members, who represented up to 80% of the compromised payment cards.

These parties worked out a deal in which Home Depot would pay the full amount of the assessment, plus about a 10% premium payable to the banks that released their claims against Home Depot. Visa and Mastercard then sent the release offers to some of the larger banks, who had been part of the negotiations with Home Depot. Some of these larger banks then forwarded the release offers to smaller banks that were affiliated with the card brands only through their relationship with the larger banks and had not been involved in the release negotiations. The release offers specifically referenced *1074 the ongoing multidistrict litigation ("MDL"), making clear that any banks who accepted the terms would release their claims in that litigation.

Class Counsel objected. Home Depot had earlier moved the District Court for permission to reach out to putative class members to propose release offers. Because the District Court had not yet ruled on that motion, Class Counsel accused Home Depot of charging ahead without the Court's permission. 4 Class Counsel also complained that the release offers were misleading and coercive. Specifically, the offers did not say how much the banks would receive from the settlement or whether the banks would still receive their share of the assessments even if they did not agree to the settlement. Moreover, the offers were sent during the Thanksgiving holidays, and banks were given only a few days to respond. Class Counsel moved the District Court to vacate the releases, to send curative notices to class members, and to protect class members from misleading settlement attempts going forward.

The District Court agreed that the release offers were misleading and coercive. But the Court refrained from ruling on Class Counsel's request for relief-vacating the releases, etc.-until it had more information. To that end, the Court allowed Class Counsel to pursue discovery relating to the release offers. The parties clashed repeatedly over the scope of the discovery, leading to a flurry of motions to compel and other discovery disputes. The Court never resolved whether to vacate the release offers; it stayed discovery pending settlement negotiations before the issue came to a head.

Ultimately, but before Home Depot settled with the class, a significant number of banks accepted the releases, including most of the larger ones. These banks represented 70-80% of the compromised payment cards. In exchange for the releases, Home Depot paid these banks a total of $14.5 million (a premium on top of the $120 million in assessments).

C.

Returning to the litigation, after the District Court stayed discovery, the parties participated in three rounds of mediation, resulting in a preliminary settlement agreement that the parties presented to the District Court for approval.

The settlement agreement defined the class as follows:

All banks, credit unions, financial institutions, and other entities in the United States ... that issued Alerted-on Payment Cards. 5 Excluded from this class are entities that have released all of their claims against Home Depot, but not excluded from the class are independent sponsored entities whose claims were released in connection with [the release offers] made by Mastercard.

As one would expect, the class definition excludes those banks that released their claims against Home Depot by accepting the release offers. However, "independent sponsored entities"-smaller banks who did not contract directly with the card brands-are not excluded from the class. They are not excluded because Class *1075 Counsel contests the validity of their releases, maintaining that these smaller banks were misled and coerced by the offers.

In exchange for settling the case, Home Depot agreed to provide the following relief. First, Home Depot agreed to pay $25 million into a settlement fund. The fund would be used to pay any taxes due and to pay any service awards to class representatives that the District Court approved. 6 The remainder of the fund would be distributed to class members who had not released their claims. No money in the fund would revert to Home Depot. Second, Home Depot agreed to pay up to $2.25 million to some of the smaller banks (the "independent sponsored entities"). To be eligible, these banks must certify that they did not have sufficient time or information to appropriately consider the release offers-i.e., that they were misled and/or coerced. Home Depot did not create a fund for these payments; if less than $2.25 million was claimed, Home Depot would pay only the amount claimed.

Finally, Home Depot agreed to adopt security measures to protect its data. These measures include developing a "risk exception" process to identify risks in its data security; designing safeguards to manage any risks identified; monitoring its service providers and vendors to ensure compliance with those safeguards; and implementing an industry recognized security control framework.

On the matter of attorney's fees, the settlement agreement provided that Home Depot would pay the "reasonable attorneys' fees, costs and expenses" of Class Counsel. 7 But the agreement left the amount of fees undetermined. Pursuant to the agreement, Class Counsel would submit to the District Court a requested amount in fees and expenses, to which Home Depot was free to object. While each party reserved its right to appeal the District Court's decision on attorney's fees, the amount awarded-no matter how large or how small-would not affect the "finality or effectiveness" of the settlement. Notably, the agreement stated that Home Depot's payment of attorney's fees would be "separate from and in addition to" the settlement fund. In other words, payment would not come from the $25 million set aside for class members.

The District Court approved the settlement agreement, noting that the issue of attorney's fees would be decided separately.

D.

After the terms of the settlement were approved, the dispute over attorney's fees began.

*1076 Courts calculate attorney's fees using one of two methods: the percentage method or the lodestar method. Under the percentage method, courts award counsel a percentage of the class benefit. See Camden I Condo. Ass'n v. Dunkle , 946 F.2d 768 , 774 (11th Cir. 1991). The class benefit generally includes any benefits resulting from the litigation that go to the class. Id. In this Circuit, courts typically award between 20-30%, known as the benchmark range. Id.

Under the lodestar method, courts determine attorney's fees based on the product of the reasonable hours spent on the case and a reasonable hourly rate. Hensley v. Eckerhart , 461 U.S. 424 , 433, 103 S.Ct. 1933 , 76 L.Ed.2d 40 (1983). The product is known as the lodestar. Sometimes courts apply to the lodestar a multiplier, also known as an enhancement or an upward adjustment, to reward counsel on top of their hourly rates. See 5 William B. Rubenstein, Newberg on Class Actions § 15:91, p. 353 (5th ed. 2015).

Class Counsel advised the District Court that it had discretion to choose either the lodestar or the percentage method. Under either approach, Class Counsel requested $18 million in fees. In contrast, Home Depot argued that the District Court had to use the lodestar method, and based on its calculations, a reasonable fee would be about $5.6 million.

After entertaining a hearing on the motion for attorney's fees and reviewing the parties' briefings, the District Court issued a five-page decision. Following Home Depot's recommendation, the District Court adopted the lodestar approach. The District Court accepted the lodestar proposed by Class Counsel-about $11.7 million-as "an appropriate measure of the time expended by the plaintiffs in this case." Next, it applied the same multiplier used in the consumer-track settlement, 1.3, to arrive at a reasonable fee of $15.3 million.

Home Depot argued that Class Counsel was not entitled to a multiplier. Home Depot did not suggest that a multiplier was prohibited, only that it was not warranted. In Home Depot's view, Class Counsel did not achieve a great result, the case was not more complex than the consumer track, and Class Counsel did not face greater risks than counsel for the consumer class faced. But the District Court disagreed, finding that a multiplier of 1.3 was "appropriate and justified in light of the exceptional litigation risk that class counsel took in litigating this case."

While the District Court agreed with Home Depot on using the lodestar method, it declined to adopt the lodestar proposed by Home Depot: about $5.6 million. It rejected the argument that Class Counsel's lodestar should be the same as the one used for counsel in the consumer track:

The Court accepts that the lawyers for the [banks] have expended more effort than the lawyers who represented consumers, that they had to expend more effort than did the consumer lawyers in arriving at a settlement, and that dealing with [banks] rather than consumers added difficulty to the process of litigating this case, such as finding adequate class representatives, and thus required more time and effort.

The District Court also rejected the argument that Class Counsel should not be compensated for the time spent litigating about the card-brand recovery process, finding that the issues relating to the release offers "were appropriate for plaintiffs to address in this case."

Finally, the District Court employed the percentage method as a cross-check on the lodestar. The parties agreed that the class benefit should include the $25 million settlement *1077 fund, the $2.25 million Home Depot agreed to pay to some smaller banks (the sponsored entities), and $710,000 in expenses. The parties did not agree on two other potential inputs into the class benefit.

Class Counsel thought the class benefit should include the $14.5 million premiums that Home Depot paid to banks in exchange for releases as part of the card-brand recovery process. Home Depot urged the District Court not to include the $14.5 million for three reasons. First, the premiums did not go to class members; they went to former putative class members who were no longer part of the class (because they accepted the premiums). Second, the premiums were not prompted by the litigation. And third, Class Counsel tried to stop the premiums, so they should not now receive compensation for them. The District Court sided with Class Counsel and included the premiums, finding that they were "substantially motivated by the pendency of this litigation."

Class Counsel also asked the District Court to include the $18 million in requested fees in the class benefit. Home Depot objected to including a self-selected fee in the class benefit, pointing out that allowing Class Counsel to determine the size of the benefit by selecting the size of the fee is circular. Instead, Home Depot effectively made the same circular request, proposing that the District Court use the lodestar amount as the fees to include in the benefit. The District Court declined to follow either recommendation, and did not include any attorney's fees in the class benefit, because this was not a "true common fund analysis."

Thus, adding the $25 million settlement fund, the $2.25 million that Home Depot agreed to pay to the sponsored entities, the $710,000 in expenses, and the $14.5 million premiums, the class benefit equaled about $42.5 million. 8 As an attorney's fee of $15.3 million is slightly more than a third of the class benefit, the District Court concluded that the percentage cross-check supported the reasonableness of the fee award.

In sum, the District Court ordered Home Depot to pay Class Counsel $15.3 million in fees. It reached this award using the lodestar method, under which it accepted the lodestar proposed by Class Counsel and applied a multiplier of 1.3 to account for risk. The Court also used a percentage cross-check, which, after including the $14.5 million premiums in the class benefit and excluding any attorney's fees, showed that the fee award was slightly more than a third of the class benefit, which the Court found to be reasonable.

E.

Home Depot appeals the award of attorney's fees, raising four issues for our consideration. First, whether it was an abuse of discretion for the District Court to apply a multiplier. Second, whether it was an abuse of discretion to compensate Class Counsel for time spent litigating about the card-brand recovery process. Third, whether it was an abuse of discretion to compensate Class Counsel for time spent soliciting class representatives. And fourth, *1078 whether the District Court's order fails to provide sufficient detail for meaningful appellate review.

Class Counsel also brings a cross-appeal-conditioned on the outcome of Home Depot's appeal. Class Counsel asks us to reach their cross-appeal only if, in response to Home Depot's appeal, we reverse or modify the attorney's fee award and remand to the District Court for reconsideration. In that event, Class Counsel challenges the District Court's decision not to include attorney's fees in the class benefit when it conducted the percentage cross-check. Thus, if we remand the case, Class Counsel asks us to instruct the District Court to include attorney's fees in the class benefit when it performs the percentage method-either as a cross-check or in the first instance. 9 Necessarily, then, we address the issues raised in Home Depot's appeal first.

II.

We review a district court's award of attorney's fees for abuse of discretion. Muransky v. Godiva Chocolatier, Inc. , 922 F.3d 1175 , 1194 (11th Cir. 2019). "An abuse of discretion occurs if the judge fails to apply the proper legal standard or to follow proper procedures in making the determination, or bases an award upon findings of fact that are clearly erroneous." ACLU of Ga. v. Barnes , 168 F.3d 423 , 427 (11th Cir. 1999) (quotation omitted). "Under this standard, district courts have great latitude in setting fee awards in class action cases." Muransky , 922 F.3d at 1194 .

A.

Before tackling the specific issues raised in Home Depot's appeal, we address a preliminary question on which much of the subsequent analysis turns: whether this is a common-fund or fee-shifting case. Different rules and principles govern common-fund cases and fee-shifting cases. Because this fee arrangement defies easy categorization, we start with some background on these concepts.

1.

In the American legal system, each party is traditionally responsible for its own attorney's fees. Hardt v. Reliance Standard Life Ins. Co. , 560 U.S. 242 , 253, 130 S.Ct. 2149 , 176 L.Ed.2d 998 (2010) ("Each litigant pays his own attorney's fees, win or lose, unless a statute or contract provides otherwise."); see also Alyeska Pipeline Serv. Co. v. Wilderness Soc'y , 421 U.S. 240 , 247, 95 S.Ct. 1612 , 44 L.Ed.2d 141 (1975) ("In the United States, the prevailing litigant is ordinarily not entitled to collect a reasonable attorneys' fee from the loser."). This principle is known as the American Rule. Hardt , 560 U.S. at 253 , 130 S.Ct. 2149 .

There are three exceptions to the American Rule: (1) when a statute grants courts the authority to direct the losing party to pay attorney's fees; (2) when the parties agree in a contract that one party will pay attorney's fees; and (3) when a court orders one party to pay attorney's fees for acting in bad faith. 10 See *1079 Rubenstein, supra § 15:25, p. 59-60 ; see also Alyeska Pipeline , 421 U.S. at 257-59 , 95 S.Ct. 1612 . These exceptions-when one party pays for the other's attorney's fees-describe fee-shifting cases.

Some courts, including this one, have described common-fund cases as an exception to the American Rule. See Camden I , 946 F.2d at 771 ("One of the recognized exceptions to the American Rule is the 'common fund' case."). That is incorrect. 11 And it is important to understand why.

A common-fund case is when "a lawyer who recovers a common fund for the benefit of persons other than himself or his client is entitled to a reasonable attorney's fee from the fund as a whole." Boeing , 444 U.S. at 478, 100 S.Ct. 745 . This is typical in class actions, where the class might receive a large payout, from which the attorney derives his fees. Common-fund cases are consistent with the American Rule, because the attorney's fees come from the fund, which belongs to the class. In this way, the client, not the losing party, pays the attorney's fees. See Rubenstein, supra , § 15:25, p. 60 n.3 ("Occasionally courts state that common fund fee awards are, too, an exception to the American Rule. But that is not quite right: when fees are extracted from the common fund to pay class counsel, the class members' recoveries are reduced accordingly and hence those class members themselves are paying their own fees." (citation omitted)).

Thus, the key distinction between common-fund and fee-shifting cases is whether the attorney's fees are paid by the client (as in common-fund cases) or by the other party (as in fee-shifting cases). 12

2.

Applying this understanding of attorney's fees, we are convinced that this is a fee-shifting case.

On its face, the settlement agreement provides that Home Depot will pay the attorney's fees. The agreement states that "Home Depot agrees to pay the reasonable attorneys' fees, costs and expenses of counsel for the Financial Institution Plaintiffs." Even more explicit, the agreement goes on to state that "[a]ny award of attorneys' fees, costs, and expenses shall be paid separate from and in addition to the Settlement Fund." That sounds like fee *1080 shifting. Indeed, it is hard to imagine how the settlement agreement could be any clearer that Home Depot will pay the attorney's fees, and that payment will not come out of the class fund. A settlement agreement is a contract, which we construe "to effectuate the intent of the parties," Pottinger v. City of Miami , 805 F.3d 1293 , 1298 (11th Cir. 2015), and the parties' intent seemed to be for the fees to be paid separately by Home Depot-i.e., a fee-shifting arrangement.

Still, Class Counsel insists that we should treat this arrangement as a constructive common fund. Where class action settlements are concerned, courts will often classify the fee arrangement as a "constructive common fund" that is governed by common-fund principles even when the agreement states that fees will be paid separately. See, e.g. , In re Heartland Payment Sys., Inc. , 851 F. Supp. 2d 1040 , 1072 (S.D. Tex. 2012) (providing an overview of the constructive common-fund doctrine). Based on a proper understanding of the doctrine of constructive common funds, we find that it does not apply to this case.

The rationale for the constructive common fund is that the defendant negotiated the payment to the class and the payment to counsel as a "package deal." Id. (quoting Johnston v. Comerica Mortg. Corp. , 83 F.3d 241 , 246 (8th Cir. 1996) ). The defendant is concerned, first and foremost, with its total liability. See In re Gen. Motors Corp. , 55 F.3d 768 , 819-20 (3d Cir. 1995). Thus, courts have recognized that, as a practical matter, defendants undoubtedly take into account the amount of attorney's fees when they agree on an amount to pay the class. See id. ; see also Brytus v. Spang & Co. , 203 F.3d 238 , 246 (3d Cir. 2000) ("[C]onsideration of the attorney's fees was likely factored into the amount of settlement."). By taking the amount of attorney's fees into account, the defendant effectively reduces the class' recovery accordingly. Commentators have endorsed this reasoning:

It is fair to assume that the class members' recoveries have been indirectly reduced already in that the settling defendant, in agreeing to pay the class, say, $8 million and class counsel an additional $2 million, is effectively agreeing to pay the class $10 million and to not contest class counsel's pursuit of a 20% fee from the $10 million recovery.

Rubenstein, supra , § 15:76, p. 267 n.7; see also Manual for Complex Litigation (Fourth) § 21.7 (2004) ("If an agreement is reached on the amount of a settlement fund and a separate amount for attorney fees and expenses, ... the sum of the two amounts ordinarily should be treated as a settlement fund for the benefit of the class, with the agreed-on fee amount constituting the upper limit on the fees that can be awarded to counsel.") 13

But this package-deal reasoning does not apply here. Put simply, there was no package: Home Depot did not negotiate the attorney's fees simultaneously with the settlement fund. The fees were left entirely *1081 to the District Court's discretion. The parties did not even agree to a cap, often referred to as a "clear-sailing agreement." 14 So it cannot be said that Home Depot took into account the amount of attorney's fees when it negotiated the size of the class award, because the amount of attorney's fees was completely undetermined.

Usually, when courts have applied the constructive common-fund doctrine, the parties at least agreed to a cap on the attorney's fees. See, e.g. , Johnston , 83 F.3d at 243-44 ; Dennis v. Kellogg Co. , 697 F.3d 858 , 863 (9th Cir. 2012). But see In re Gen. Motors Corp. , 55 F.3d at 781, 821 (characterizing the fee arrangement as a common fund even though the fees were paid separately and left completely undetermined). In that scenario, the constructive common fund makes more sense because the defendant used the cap to determine its total exposure and (theoretically) limited the class' recovery accordingly. Class Counsel argues that whether the amount is agreed-to or capped is immaterial: the amount of attorney's fees is always unsettled because it is subject to the court's approval. True enough. In reality, though, a defendant can make a much more reliable estimate of its liability when the parties make a joint recommendation than when the parties present widely divergent proposals.

Admittedly, a defendant could (and probably does) make an educated guess concerning the amount of attorney's fees, even when the amount is left undetermined. 15 But if this were enough to create a constructive common fund, it would be virtually impossible to contract for fee-shifting. The purported rule would be that any class settlement-no matter whether the fees are paid by the defendant or out of the class award, or whether the fees are negotiated separately or as part of the settlement-should be treated as a common fund. As a result, construing the agreement here as a constructive common fund would effectively eliminate the ability to contract for fee-shifting absent perhaps some magic-word requirement.

In sum, we hold that the constructive common fund does not apply when the agreement provides that attorney's fees will be paid by the defendant separately from the settlement fund, and the amount of those fees is left completely undetermined. We construe the settlement agreement here as a fee-shifting arrangement.

3.

Ordinarily, after classifying the fee arrangement, the next question would be which method the court should use to calculate the attorney's fees. In common-fund cases, we have directed courts to use the percentage method. Camden I , 946 F.2d at 774 ("Henceforth in this circuit, attorneys' fees awarded from a common fund shall be based upon a reasonable percentage of the fund established for the benefit of the class."). In statutory fee-shifting cases, the Supreme Court has said that courts should use the lodestar method. See City of Burlington v. Dague , 505 U.S. 557 , 562, 112 S.Ct. 2638 , 120 L.Ed.2d 449 (1992) ("The 'lodestar' figure has, as its name suggests, become the guiding light of our fee-shifting jurisprudence.").

*1082 This case, however, is a contractual fee-shifting case, and the appropriate method for such a case is not clearly governed by any binding precedent.

But the parties do not challenge the District Court's selection of the lodestar method. Even though Class Counsel believes this is a common-fund case, they say the District Court had discretion to choose either the percentage or the lodestar method because several of the claims raised in the complaint were under state statutes with fee-shifting provisions. Class Counsel may be right that the District Court had discretion to choose, but the proper method here has nothing to do with the state statutes. The District Court awarded the attorney's fees pursuant to a contract-the settlement agreement-not pursuant to a statute. See Brytus , 203 F.3d at 246 ("Where there has been a settlement, the basis for the statutory fee has been discharged, and it is only the fund that remains."); see also Florin v. Nationsbank of Ga., N.A. , 34 F.3d 560 , 563 (7th Cir. 1994) (finding that "the terms of [the statute's] fee-shifting provision do not purport to control fee awards in cases settled with the creation of a common fund"). Nevertheless, because the parties do not challenge the District Court's use of the lodestar method, we do not question it. 16

With these preliminary matters decided, we resume with the issues raised by Home Depot on appeal.

B.

First, Home Depot argues that the District Court abused its discretion by applying a multiplier to Class Counsel's lodestar. Home Depot bases its argument on Supreme Court precedent outlining the use of multipliers in statutory fee-shifting cases. Although we are not bound in a contractual fee-shifting case by statutory fee-shifting cases, we agree that it was error for the District Court to enhance Class Counsel's lodestar based on risk.

1.

We begin by summarizing the Supreme Court's precedent on statutory fee-shifting cases. Fee-shifting statutes allow counsel for the prevailing party to recover a reasonable fee. Perdue v. Kenny A. ex rel. Winn , 559 U.S. 542 , 550, 130 S.Ct. 1662 , 176 L.Ed.2d 494 (2010). A reasonable fee is one sufficient to attract competent counsel to represent the case, but not one that provides a windfall for attorneys. Id. at 552 , 130 S.Ct. 1662 . There is a strong presumption that the lodestar yields a reasonable fee for this purpose. Id. Because the lodestar is presumed to be sufficient, a multiplier will be appropriate only in "rare and exceptional" cases. Id. (quotations omitted). To warrant a multiplier, the fee applicant must produce "specific evidence" that an enhancement is necessary to provide a reasonable fee. Id. at 553 , 130 S.Ct. 1662 (quotation omitted). An enhancement may be necessary if the lodestar does not reflect the true value of counsel's work. Id. at 554 , 130 S.Ct. 1662 .

The question becomes, what specific evidence would satisfy this standard.

*1083 The Supreme Court has made it plain that "most, if not all," of the factors used to determine a reasonable fee are already subsumed in the lodestar, and it is not permissible to enhance a fee based on a factor that is subsumed. Id. at 553 , 130 S.Ct. 1662 . That would be "double counting"-i.e., a windfall. Dague , 505 U.S. at 563 , 112 S.Ct. 2638 . In a series of cases, the Court has expanded on which factors are subsumed and why. 17

For example, the novelty and complexity of the issues are reflected in the number of hours spent on the case, as complicated litigation will demand more time. Blum v. Stenson , 465 U.S. 886 , 898, 104 S.Ct. 1541 , 79 L.Ed.2d 891 (1984). Similarly, the skill and experience of the attorneys will be reflected in the hourly rates. Id. Thus, courts should not use these factors to justify a multiplier. Id. at 898-99 , 104 S.Ct. 1541 .

As for the results obtained, this factor should be folded into the quality-of-representation factor. 18 Perdue , 559 U.S. at 554 , 130 S.Ct. 1662 . This is so because the results obtained are relevant to attorney's fees only if those results are attributable to counsel's performance, rather than, say, the other side dropping the ball. Id. And the quality of representation should be used to enhance the fee only in the rare cases where the fee applicant demonstrates that the "superior attorney performance is not adequately taken into account in the lodestar calculation." Id.

The Court offered three examples of when the lodestar may not adequately capture counsel's superior performance. First, "where the method used in determining the hourly rate ... does not adequately measure the attorney's true market value." Id. at 554-55 , 130 S.Ct. 1662 . "This may occur if the hourly rate is determined by a formula that takes into account only a single factor (such as years since admission to the bar) or perhaps only a few similar factors." Id. at 555 , 130 S.Ct. 1662 (footnote omitted). Second, if counsel incurs an "extraordinary outlay of expenses" in the case. Id. Third, if there is an "exceptional delay in the payment of fees." Id. at 556 , 130 S.Ct. 1662 .

Finally, the Court determined that risk is not an appropriate basis for a multiplier in statutory fee-shifting cases. The Court explained that the "risk of loss ... is the product of two [inputs]: (1) the legal and factual merits of the claim, and (2) the difficulty of establishing those merits." Dague , 505 U.S. at 562 , 112 S.Ct. 2638 . The second input is subsumed in the lodestar-"either in the higher number of hours expended to overcome the difficulty, or in the higher hourly rate of the attorney skilled *1084 and experienced enough to do so." Id. While the first input is not reflected in the lodestar, "there are good reasons" not to enhance fees for the risk presented by meritless claims. Id. at 563 , 112 S.Ct. 2638 .

Namely, if fees are enhanced for contingency fee cases as a class-rather than based on a risk assessment of each case-it would inevitably overcompensate some cases and undercompensate others. Id. at 564-65 , 112 S.Ct. 2638 . Conversely, if fees are enhanced based on the riskiness of each particular case, it would reward lawyers for taking cases with relatively little merit and incentivize bad claims. 19 Id. at 563 , 112 S.Ct. 2638 . For this reason, not adjusting fees for risk is consistent with fee-shifting statutes. Id. at 565 , 112 S.Ct. 2638 . These statutes limit fees to prevailing parties, and adjusting fees for risk effectively subsidizes the attorney's losing cases-a result at odds with the prevailing party requirement. Id. Plus, enhancing for risk "would make the setting of fees more complex and arbitrary, hence more unpredictable, and hence more litigable." Id. at 566 , 112 S.Ct. 2638 . For all of these reasons, the Supreme Court decreed that courts could not use a multiplier in statutory fee-shifting cases to account for risk. Id. at 567 , 112 S.Ct. 2638 .

If these precedents apply, it was an abuse of discretion for the District Court to apply a multiplier. The District Court's only stated reason for using a multiplier was the exceptional risk taken by counsel in litigating the case. 20 And risk, according to the Supreme Court, is not an appropriate basis for enhancing an attorney's fee in statutory fee-shifting cases. But this is a contractual fee-shifting arrangement. As such, we must consider whether and to what extent these precedents apply.

2.

There is no question that the Supreme Court precedents stretching from Hensley to Perdue are specific to fee-shifting statutes. See Perdue , 559 U.S. at 552 , 130 S.Ct. 1662 ("Our prior decisions concerning the federal fee-shifting statutes have established six important rules that lead to our decision in this case."). These cases were about interpreting statutory language. See Blum , 465 U.S. at 893 , 104 S.Ct. 1541 ("Resolution of these two arguments [about the proper way to calculate attorney's fees] begins and ends with an interpretation of the attorney's fee statute."); see also Dague , 505 U.S. at 562 , 112 S.Ct. 2638 ("This language is similar to that of many other federal fee-shifting statutes *1085 ... [and] our case law construing what is a 'reasonable' fee applies uniformly to all of them." (citation omitted)). Thus, these precedents are not binding outside of the statutory context.

For this reason, we have held that the Supreme Court precedent requiring the use of the lodestar method in statutory fee-shifting cases does not apply to common-fund cases. Muransky , 922 F.3d at 1194-95 (" Perdue addresses fee-shifting statutes and says nothing about the award of attorney's fees from a common fund."). And other circuits have held that the Supreme Court's jurisprudence restricting the use of multipliers in statutory fee-shifting cases does not apply to common-fund cases. See Florin , 34 F.3d at 564-65 ("[W]e conclude that the holding in Dague ... forbidding risk multiples in statutory fee-shifting cases ... has no application to common fund cases."); see also In re Wash. Pub. Power Supply Sys. Sec. Litig. , 19 F.3d 1291 , 1299 (9th Cir. 1994) (" Dague 's rationale for barring risk multipliers in statutory fee cases does not operate to bar risk multipliers in common fund cases."). But see In re Gen. Motors Corp. , 55 F.3d at 822 (suggesting that even in a common-fund case, the "pertinence" of Supreme Court precedent on the use of multipliers in lodestar analysis is "patent").

But this is a contractual fee-shifting case, not a common-fund case. As such, it is more closely related to the Supreme Court precedent governing fee-shifting statutes. And just because precedent is not technically binding does not mean we should blithely disregard it. To promote consistency in the law, we should adhere to precedent where its reasoning applies.

For example, it makes sense to draw a clear line between fee-shifting cases and common-fund cases. The common-fund doctrine serves a different purpose-preventing unjust enrichment-and the fees are paid by the client, rather than by the opposing party. See Rubenstein, supra , § 15:53, p. 181-83 (explaining the purpose of the common fund doctrine). In contrast, the Supreme Court cases reviewed above and the present case are all fee-shifting arrangements, where the fees are paid by the other party and the purpose is to fairly compensate counsel for the value of their work. The only difference is that the Supreme Court cases are fee-shifting by statute, and this one is fee-shifting by contract. Thus, while the statutory cases are not binding on contractual arrangements, we will not lightly cast aside the statutory fee-shifting precedent if its reasoning applies with full force. Obviously, the reasoning does not apply if it is specific to statutory interpretation.

With that in mind, we consider whether the Supreme Court's reasons for limiting the use of multipliers in statutory fee-shifting cases apply to contractual fee-shifting cases. The Court's reasons largely turn on the point that most of the factors used to justify an enhancement are already subsumed in the lodestar, so it would result in a windfall to count them again with a multiplier. See, e.g. , Perdue , 559 U.S. at 553 , 130 S.Ct. 1662 . This reasoning makes it just as unreasonable to double-count in a contractual fee-shifting case as it is in a statutory fee-shifting case.

Here, the District Court used a multiplier to account for risk. The Supreme Court forbade adjusting for risk for a number of reasons. Dague , 505 U.S. at 562-67 , 112 S.Ct. 2638 . To start, the Court said that risk was partly reflected in the lodestar. Id. at 562 , 112 S.Ct. 2638 . For the part of risk that is not reflected in the lodestar, the Court gave one reason for not using it to justify an enhancement that is specific to statutes-subsidizing losing claims contrary to the prevailing-party requirement found in most fee-shifting statutes.

*1086 Id. at 565 , 112 S.Ct. 2638 ; see also infra part II.C. But the Court gave other reasons that apply equally in contractual fee-shifting settings, such as incentivizing meritless claims and making fees less predictable. Thus, on the whole, we find that the Court's prohibition on enhancements for risk applies to contractual fee-shifting cases when courts use the lodestar method.

Because it is inappropriate to enhance a lodestar in a fee-shifting case to account for risk, the District Court abused its discretion in applying a multiplier on the basis of the "exceptional litigation risk that class counsel took in litigating this case."

3.

Class Counsel insists, however, that Home Depot waived the multiplier issue. It's a close call, but we do not think Home Depot waived the issue.

We generally will not review issues raised for the first time on appeal. Blue Martini Kendall, LLC v. Miami Dade County , 816 F.3d 1343 , 1349 (11th Cir. 2016). But there is a difference between raising new issues and making new arguments on appeal. If an issue is "properly presented, a party can make any argument in support of that [issue]; parties are not limited to the precise arguments they made below." Yee v. City of Escondido , 503 U.S. 519 , 534, 112 S.Ct. 1522 , 118 L.Ed.2d 153 (1992) ; see also Sec'y, U.S. Dep't of Labor v. Preston , 873 F.3d 877 , 883 n.5 (11th Cir. 2017) ("Parties can most assuredly waive positions and issues on appeal, but not individual arguments .... Offering a new argument or case citation in support of a position advanced in the district court is permissible-and often advisable." (citation omitted)). This principle begs the question: does Home Depot raise a new argument or a new issue?

Home Depot argued below that the District Court should not apply a multiplier to Class Counsel's lodestar. In support of this position, Home Depot argued that Class Counsel did not achieve a great result, the case was not more complex than the consumer case, and Class Counsel did not face greater risk than counsel for the consumers. Now, on appeal, Home Depot makes a different pitch: it's not that the level of risk did not justify a multiplier; it's that the District Court cannot use a multiplier to account for risk, period. The new argument is based on a different line of precedents, see supra part II.B.1., and is inconsistent with the old argument, which seemed to accept that multipliers for risk could be appropriate in the right circumstances.

Nevertheless, in the final analysis, we think this is a new argument, not a new issue. Home Depot asked the District Court not to apply a multiplier. On appeal, Home Depot makes the same request, albeit for different (and contradictory) reasons. The issue was not waived.

C.

The second issue Home Depot raises in its appeal is whether the District Court abused its discretion by compensating Class Counsel for the time spent litigating about the card-brand recovery process.

Home Depot says that it was an abuse of discretion because our precedent requires courts to "deduct time spent on discrete and unsuccessful claims." Norman v. Hous. Auth. of Montgomery , 836 F.2d 1292 , 1302 (11th Cir. 1988). And the time spent litigating about the card-brand recovery process-whether the release offers were misleading and coercive, whether Home Depot improperly directed the releases, *1087 and whether the releases should be vacated-was, in Home Depot's view, discrete and unsuccessful. However, the rule that Home Depot relies on does not apply here.

The rule against compensating counsel for time spent on discrete and unsuccessful claims comes from fee-shifting statutes. Specifically, this rule derives from language commonly found in such statutes that limits recovery to a "prevailing party." See Hensley , 461 U.S. at 433, 435 , 103 S.Ct. 1933 ("A plaintiff must be a 'prevailing party' to recover an attorney's fee under § 1988.... The congressional intent to limit awards to prevailing parties requires that these unrelated claims be treated as if they had been raised in separate lawsuits, and therefore no fee may be awarded for services on the unsuccessful claim."). Notably, some fee-shifting statutes do not contain the prevailing-party language, in which case, as you would expect, the prevailing-party limitation does not apply. See Hardt , 560 U.S. at 252 , 130 S.Ct. 2149 ("The words 'prevailing party' do not appear in this provision. ... We therefore hold that a fee claimant need not be a 'prevailing party' to be eligible for an attorney's fees award.").

Here, of course, the fees are awarded pursuant to a contract, not a statute, and there is no prevailing-party limitation in the settlement agreement. 21 Accordingly, the prevailing-party limitation does not apply, and the District Court did not need to deduct time spent on discrete and unsuccessful claims. Instead, the question is simply whether the time spent was reasonable, which is the standard set in the agreement.

Time spent is reasonable, and thus compensable, if it would be proper to charge the time to a client. See Norman , 836 F.2d at 1301 . As with a client, counsel should not include in the lodestar hours that are "excessive, redundant or otherwise unnecessary." Id. (quoting Hensley , 461 U.S. at 434 , 103 S.Ct. 1933 ). In other words, counsel must exercise "billing judgment." Id. If counsel does not exercise billing judgment, "courts are obligated to do it for them." Barnes , 168 F.3d at 428 . Thus, "[i]n the final analysis, exclusions for excessive or unnecessary work on given tasks must be left to the discretion of the district court." Norman , 836 F.2d at 1301 .

In this case, it was firmly within the District Court's discretion to compensate Class Counsel for time spent challenging the release offers. To be clear, the release offers were effectively settlement offers: they promised additional payment in exchange for releasing the class claims. Indeed, institutions representing around 70-80% of the compromised payment cards settled their claims through the release offers. Class Counsel thought these were lousy settlement offers and that the class could recover more from the litigation. Moreover, Class Counsel was concerned that the offers were misleading and coercive. It was perfectly reasonable for Class Counsel to take action to ensure that class members' releases were voluntary and informed, especially since Class Counsel thought the terms were unfavorable. To hold otherwise would be to say, as a matter of law, that it is unreasonable for Class Counsel to ever oppose a settlement.

We also note that the District Court specifically authorized Class Counsel to conduct discovery into the card-brand recovery *1088 process. The District Court agreed that the release offers were misleading and coercive, and it approved the discovery in order to determine whether to vacate the releases. We are not willing to say that time spent on court-sanctioned discovery was unreasonable.

It was not an abuse of discretion to compensate Class Counsel for time spent on the card-brand recovery process.

D.

The third issue Home Depot raises in its appeal is whether the District Court abused its discretion by compensating Class Counsel for time spent soliciting class representatives.

A significant chunk of Class Counsel's lodestar included time spent selecting and vetting class representatives. Class Counsel wanted to ensure that if the proposed national class was not certified, there would be state-specific classes as an alternative. To that end, Class Counsel needed to find and select a class representative from each state. Ultimately, Class Counsel secured representatives from 44 states. This is a sound (and not uncommon) strategy. It is also a time-consuming process.

In Barnes , we said that "hours spent looking for and soliciting potential plaintiffs should not have been included in the time billed." 168 F.3d at 435 . We explained that, based on fee-shifting statutes , counsel is entitled to compensation for time reasonably spent "on the litigation." Id. (emphasis and quotation omitted). Thus, "time spent procuring potential plaintiffs" is not compensable, "because until the attorney has a client, there is no case to litigate." Id. This reasoning made sense in Barnes ; it does not apply to this case.

As an initial matter, it is questionable whether the formal limit on compensation to time spent "on the litigation" even applies, since this case is not governed by a fee-shifting statute and the settlement agreement says only that Class Counsel should be compensated with a reasonable attorney's fee. That aside, in a class action (which Barnes was not), it is not true that a case does not exist until the class names a representative. Here, for example, numerous cases were filed across the country before being consolidated as an MDL. As a fact, then, the litigation existed before naming class representatives.

Furthermore, it would be seriously misguided to say that Class Counsel cannot be paid for time spent vetting class representatives. Selecting proper class representatives is an important part of what class counsel does. And counsellors should be paid for work reasonably done on behalf of their clients. See Norman , 836 F.2d at 1305 ("The law seeks to compensate attorneys for work reasonably done actually to secure for clients the benefits to which they are entitled."). There is no question that Class Counsel's efforts in this instance meet that standard.

For these reasons, we hold that it was not an abuse of discretion to pay Class Counsel for their time spent finding and vetting class representatives.

E.

Finally, Home Depot argues that the District Court's order does not allow for meaningful review.

As noted earlier, a district court has ample discretion in awarding fees-and with good reason. See Hensley , 461 U.S. at 437 , 103 S.Ct. 1933 ("We reemphasize that the district court has discretion in determining the amount of a fee award. This is appropriate in view of the district court's superior understanding of the litigation and the desirability of avoiding frequent appellate review of what essentially *1089 are factual matters."). But "that discretion is not without limits." Norman , 836 F.2d at 1304 . A district court's "order on attorney's fees must allow meaningful review-the district court must articulate the decisions it made, give principled reasons for those decisions, and show its calculation." Id. In other words, the court must "provide a concise but clear explanation of its reasons for the fee award." Hensley , 461 U.S. at 437 , 103 S.Ct. 1933 . Home Depot contends that the District Court failed to satisfy this requirement. We disagree.

The District Court explained that it would use the lodestar method to calculate fees, agreeing with Home Depot's argument that the percentage method was not appropriate in this case because it was not a common fund. The District Court then accepted the lodestar proposed by Class Counsel and explained why it rejected the lodestar proposed by Home Depot. It disagreed with Home Depot's argument that the lodestar should be the same as the one used for counsel in the consumer track, finding that the financial track "required more time and effort." It also disagreed with Home Depot's argument that Class Counsel should not be compensated for time spent litigating the card-brand recovery process, finding that such issues "were appropriate for plaintiffs to address in this case." Finally, the District Court decided to enhance the lodestar with a multiplier of 1.3, which it said was "appropriate and justified in light of the exceptional litigation risk that class counsel took in litigating this case." In short, we are not left in doubt about what the District Court decided and why.

Home Depot raises two alleged deficiencies in the order. First, Home Depot complains that the District Court did not deduct a single hour from Class Counsel's lodestar. Home Depot suggests that it was unreasonable for the District Court to accept more than 21,000 hours without showing any analysis of those hours specifically. While it is the obligation of district courts to ensure that the hours claimed are reasonable, see Barnes , 168 F.3d at 428 , we have said that courts "need not engage in an hour-by-hour analysis" when "the fee motion and supporting documents are so voluminous" that "an hour-by-hour review is simply impractical and a waste of judicial resources," Loranger v. Stierheim , 10 F.3d 776 , 783 (11th Cir. 1994).

The level of specificity required by district courts is proportional to the specificity of the fee opponent's objections. See Barnes , 168 F.3d at 428-29 ("[W]here specific objections are made a court's order should consist of more than conclusory statements. ... The more specific the objections to a fee application are, the more specific the findings and reasons for rejecting those objections can be."). Put differently, if a party objects to a subset of hours as unreasonable, the court should respond to that objection.

The problem for Home Depot is that it did not make specific objections to Class Counsel's lodestar. 22 Instead, Home Depot *1090 offered two general reasons why Class Counsel's hours were excessive: first, because Class Counsel spent more than twice as many hours as counsel in the consumer track; second, because it was unreasonable to spend time litigating the card-brand recovery process. The District Court responded to both arguments. It found that it was reasonable to spend more time in the financial track:

The Court accepts that the lawyers for the financial institutions have expended more effort than the lawyers who represented consumers, that they had to expend more effort than did the consumer lawyers in arriving at a settlement, and that dealing with financial institutions rather than consumers added difficulty to the process of litigating this case, such as finding adequate class representatives, and thus required more time and effort.

And it determined that the "issues relating to the card brand recovery processes were appropriate for plaintiffs to address in this case." The latter response is a little conclusory, but Home Depot's argument below for why this time was not reasonably spent was equally conclusory: "The Court should not compensate Plaintiffs for this time," accompanied by a citation stating that only hours reasonably expended are included in the lodestar. Given the lack of specificity of Home Depot's objections, the District Court's response was adequate.

The second deficiency raised by Home Depot is that the District Court did not address the Johnson factors in its order. 23 Essentially, Home Depot asks us to declare that, in order to allow for meaningful review, an order awarding attorney's fees must explicitly consider the Johnson factors. We have never announced such a rule, and we decline to do so now. Such a rule would be especially misguided in cases using the lodestar method.

With the percentage method, courts use the Johnson factors to help determine what percentage of the fund to award to counsel. See Camden I , 946 F.2d at 775 . But the Johnson factors have a much more limited role in determining fees under the lodestar method.

In Hensley , the Supreme Court opted to use the lodestar method instead of the Johnson factors to calculate a reasonable attorney's fee under fee-shifting statutes. 24 See Hensley , 461 U.S. at 429-34 , 103 S.Ct. 1933 . Nonetheless, the Court said that the Johnson factors could be used to determine whether an upward (or downward) adjustment, i.e., a multiplier, was warranted. Id. at 434 & n.9, 103 S.Ct. 1933 . In other words, the Court consigned the Johnson factors to the adjustment stage. Since then, the Court has explained that "most, if not all," of the Johnson factors are subsumed in the lodestar. Pennsylvania v. Delaware Valley Citizens' Council for Clean Air , 478 U.S. 546 , 566, 106 S.Ct. 3088 , 92 L.Ed.2d 439 (1986). As a result, the Johnson factors are largely redundant, and an enhancement will be warranted only in the rare and exceptional case where the fee applicant provides specific evidence showing that the lodestar does not adequately reflect the true market value of the attorney's performance. See Perdue , 559 U.S. at 553-54 , 130 S.Ct. 1662 .

Our precedent puts this a little differently. We have said that courts may use the *1091 Johnson factors to determine "what is a 'reasonable' hourly rate and what number of compensable hours is 'reasonable.' " Bivins v. Wrap It Up, Inc. , 548 F.3d 1348 , 1350 (11th Cir. 2008) (per curiam). But that does not mean that courts should march through the Johnson factors-considering the time and labor required, the novelty and difficulty of the issues, the results obtained, etc.-to arrive at an hourly rate. Instead, after counsel proposes an hourly rate based on the prevailing market rate in the community, courts may consider the Johnson factors to determine if the proposed rate accurately reflects the true worth of counsel. See Norman , 836 F.2d at 1299-1300 ("We still believe that at least some of the Johnson factors have utility in establishing the hourly rate. In evaluating the comparability of the market rates being attested to, the district court may wish to consider any of the Johnson factors to the extent that they suggest that comparables offered may not be relevant to the issues before the court or as they may affect the weight to be given to the comparables being offered the court.").

While the Supreme Court reserves this analysis-whether the market rate is an accurate reflection of counsel's true worth-for the adjustment stage, the result is the same. We use the Johnson factors to adjust the hourly rate, the Supreme Court uses the Johnson factors to adjust the overall lodestar. Either way, the Johnson factors are relevant only in the rare cases where they are not fully captured in the lodestar.

The crucial point, under both line of precedents, is that the Johnson factors are largely redundant to the lodestar analysis because they are almost always subsumed in the lodestar. Consequently, it would be inefficient, to say the least, to require district courts to slog through the Johnson factors when those factors have little independent bearing on the analysis.

For all of these reasons, there is no merit to Home Depot's contention that the District Court's order does not allow for meaningful review.

III.

Because we held that the District Court abused its discretion by applying a multiplier to account for risk, we reach Class Counsel's conditional cross-appeal.

Fittingly, the cross-appeal challenges the way the District Court performed the cross-check. Courts often use a cross-check to ensure that the fee produced by the chosen method is in the ballpark of an appropriate fee. See In re Gen. Motors Corp. , 55 F.3d at 820 ("[I]t is sensible for a court to use a second method of fee approval to cross check its conclusion under the first method."). 25 Accordingly, after calculating a fee using the lodestar method, the District Court cross-checked the fee with the percentage method. As the percentage method awards class counsel a percentage of the class benefit, the first step is to determine what constitutes the class benefit.

It is with this step that each party finds error. Class Counsel maintains that the District Court should have included the attorney's fees in the class benefit. For its part, Home Depot argues that the District Court should not have included in the class benefit the $14.5 million premiums that Home Depot paid to banks in exchange for *1092 releases as part of the card-brand recovery process. We take up each claim in turn.

A.

While Class Counsel is correct that attorney's fees are generally included in the class benefit in common-fund cases, it does not make sense to do so in fee-shifting cases. In typical common-fund cases, attorney's fees are necessarily included in the class benefit, see, e.g. , Gascho v. Global Fitness Holdings, LLC , 822 F.3d 269 , 282 (6th Cir. 2016) (stating that the class benefit includes attorney's fees), because the defendant pays a lump sum of cash, a percentage of which is awarded to class counsel. The analysis is straightforward because there is no need to determine the amount of attorney's fees to include in the class benefit-it all comes from the same lump sum, so the class benefit is obvious.

In constructive common-fund cases, the parties may designate the attorney's fees to be paid separately, but at the same time they agree on the amount of attorney's fees or at least set a cap on the amount. Courts, of course, are not bound by the parties' agreement on fees. Waters , 190 F.3d at 1296 n.9. So the agreed-upon fees, or the agreed-upon cap, are better thought of as the expected attorney's fees. Courts have included the expected attorney's fees in the class benefit, reasoning that the payment to the class and the payment to counsel were negotiated as a package deal, so that the defendant reduced the payment to the class to account for the expected payment to counsel. See In re Sw. Airlines Voucher Litig. , 898 F.3d 740 , 745 (7th Cir. 2018) ("Fee awards for class counsel are part of a constructive common fund because they are a benefit to the class."). In mathematical terms, the equation for the percentage method in constructive common-fund cases effectively works like this: the actual payment to counsel is the product of (1) the percentage the court decides to award, and (2) the payment to the class plus the expected payment to counsel (together, the class benefit). 26

As we explained in part II.A.2, there is no constructive common fund in this case because the parties left the amount of attorney's fees completely undetermined. Instead, they negotiated a pure fee-shifting arrangement. For this reason, Class Counsel's argument to include attorney's fees in the class benefit fails. Conceptually, if the fees are paid separately, they never belonged to the class, so they should not be included in the class benefit. Class Counsel maintains that the class benefit is reduced indirectly by the amount of attorney's fees-essentially the same argument we rejected for classifying this arrangement as a constructive common fund. We acknowledge, again, that there is some truth to this argument, but it simply does not work as a practical matter.

It would be impossible for us to determine the expected payment to counsel to plug into the math formula. Undoubtedly, Home Depot made an educated guess about the amount it might have to pay. As should be clear by now, though, calculating an attorney's fee is not an exact science-there are many variables subject to discretion-and an educated guess would at best produce a range. What are we supposed to do with that? We couldn't even replicate the range with any confidence. How could we possibly account for all the unknowns? Unlike a constructive common fund, there is no agreed-upon amount or cap to identify *1093 the amount the class benefit was reduced by.

What happened below illustrates the problem. Class Counsel told the District Court that the amount of attorney's fees to be included in the class benefit should be $18 million-the amount that Class Counsel was requesting in fees. Home Depot responded that this was unfair: by requesting an inflated figure for fees, Class Counsel inflated the size of the class benefit, thus increasing the final payout. Of course, Home Depot's proposal below did the same thing in reverse. It told the District Court that the amount of fees to be included in the class benefit should be $5 million. By proposing a deflated figure for fees, Home Depot deflated the size of the class benefit, thus reducing the final payout. Either way, the reasoning was circular. And it always will be when the attorney's fees are left completely undetermined.

Thus, the District Court properly excluded the attorney's fees from the class benefit.

Class Counsel complains that this ruling unfairly reduces their compensation. But Class Counsel fails to account for a trade-off. In common-fund cases, attorney's fees are included in the class benefit, but class counsel is not entitled to fees incurred for time spent litigating about the amount of fees (known as fee-on-fees). See Rubenstein, supra , § 15:93, p. 367 ("[T]o permit counsel to collect for hours spent seeking that fee would effectively reward them for reducing the size of the common fund, or diminishing their client's return."). In contrast, in fee-shifting cases, though the attorney's fees are not included in the class benefit, counsel can recover for the time reasonably spent pursuing fees in the case. See id. at § 15:93, p. 366. So we think the law fairly balances things out.

B.

Home Depot also takes issue with the calculation of the class benefit. It says the District Court should not have included the $14.5 million premiums it paid to banks in exchange for releases.

Counsel is entitled to compensation for its efforts that create, enhance, preserve, or protect a common fund. 27 Rubenstein, supra , § 15:59, p. 195-96 ("In assessing the question of whether the attorney's efforts generated the common fund, courts look to whether counsel's work creates, enhances, preserves, or protects the fund."); see also Mashburn v. Nat'l Healthcare, Inc. , 684 F. Supp. 679 , 686 (M.D. Ala. 1988). Thus, the question is whether Class Counsel deserves credit for the $14.5 million premiums that Home Depot paid to putative class members to settle their class claims.

Home Depot argues that the releases were unrelated to the class litigation. We're not buying. Following the data breach, there was an established card-brand recovery process that would have taken place regardless of whether a class action or any other litigation was filed. Typically, the process results in assessments that the merchant would pay (at least partially). But the assessments usually do not include a release from liability. In this case, after numerous lawsuits were filed and consolidated in an MDL, not only did Home Depot pay the assessments in full, which Mastercard could not recall ever happening before, but Home Depot offered to pay certain banks a premium on *1094 top of those assessments in exchange for releases from the exact liability faced in the class action. Home Depot would have us believe that payments that effectively settled the class claims had nothing to do with the class. We are not ostriches with our heads in the sand. The District Court was well within its discretion to conclude that the release payments were "substantially motivated by the pendency of this litigation."

Home Depot further argues that Class Counsel is not entitled to compensation for benefits that did not go to class members. And the banks who received the release payments are excluded from the class. This argument is shaky from the start. Cf. Boeing , 444 U.S. at 478, 100 S.Ct. 745 ("[T]his Court has recognized consistently that a litigant or a lawyer who recovers a common fund for the benefit of persons other than himself or his client is entitled to a reasonable attorney's fee from the fund as a whole." (emphasis added)). Of course, these banks were putative class members at the time, and they are excluded from the class now because they accepted the release payments and thus already settled their claims. A rule establishing that class counsel can get no credit for settlements with putative class members done before the class as a whole settles would entrench the very unjust enrichment and collective-action problem that class actions are designed to solve. Plus, "[t]here is no question ... that federal courts may award counsel fees based on benefits resulting from litigation efforts even where adjudication on the merits is never reached, e.g., after a settlement." Kopet v. Esquire Realty Co. , 523 F.2d 1005 , 1008 (2d Cir. 1975) ; see also Rubenstein, supra , § 15:57, p. 190 ("[A] fee may be sought regardless of whether a formal judgment, a settlement, or some other disposition-such as the mooting of a suit-created the common fund." (footnotes omitted)).

In sum, the District Court did not abuse its discretion by including the $14.5 million premiums in the class benefit. 28

IV.

For the foregoing reasons, we affirm the judgment of the District Court in part, vacate in part, and remand for further proceedings consistent with this opinion.

AFFIRMED in part, VACATED in part, and REMANDED.