NBTY and its subsidiary Rexall Sundown manufacture vitamins and nutritional supplements, including glucosamine pills, which are dietary supplements designed to help people with joint disorders, such as osteoarthritis. See “Glucosamine,” Wikipedia, http://en.wikipedia.org/wiki/ Glucosamine (visited Nov. 18, 2014, as were the other websites cited in this opinion). Several class action suits have been filed in federal district courts across the country against NBTY, Rexall, and Target-(a retail distributor of the pills, which are sold under brand names like “Osteo Bi-Flex” as well as in generic versions sold by pharmacies, such as CVS and Walgreen). The suits charge the defendants with violating several states’ consumer protection laws by making false claims for glucosamine’s efficacy, such as that it will “help rebuild cartilage,” “support renewal of cartilage,” help “maintain the structural integrity of joints,” “lubricate joints,” and “support!] mobility and flexibility.” The district court has jurisdiction of the case under the Class Action Fairness Act, 28 U.S.C. § 1332(d)(2).
Eight months after the plaintiffs filed suit in a federal district court in Illinois, class counsel in all six cases negotiated a nationwide settlement, With NBTY and *780Rexall (for simplicity, we’ll pretend there is a single defendant and call it “Rex-all”) and submitted it to that court for approval. For it is typical in class action eases of this sort — cases in which class counsel want to maximize the settlement and the defendants don’t want to settle except for “global” peace — for the multiple class counsel to negotiate a single nationwide settlement and agree to submit it for approval to just one of the district courts in which the multiple actions had been filed.
The district judge approved the settlement, though with significant modifications. As approved, the settlement requires Rexall to cough up approximately $5.63 million — $1.93 million in fees to class counsel, plus an additional $179,676 in attorney expenses (attorneys’ fees cover billable time and overhead expenses such as office space and secretaries, but clients typically - are charged extra for such expenses as expert-consultant and expert-witness fees, PACER access, photocopies, and Westlaw-research), $1.5 million in notice and administration costs, $1.13 million to the Orthopedic Research and Education Foundation, $865,284 to the 30,245 class members who submitted claims, and $30,000 to the six named plaintiffs ($5,000 apiece) as compensation for their role as the class representatives. The version of the settlement that had received preliminary approval had provided for even higher attorneys’ fees — up to $4.5 million— with Rexall stipulating that it wouldn’t challenge any attorney-fee requests by class counsel up to that amount. Such a stipulation is called a “clear-sailing” agreement.
The parties further agreed that any part of the $4.5 million that the district judge thought excessive compensation for class counsel would revert to Rexall (such an agreement is called a “reversion” or more commonly a “kicker” clause), rather than becoming available for distribution to the class members or to the Orthopedic Research and Education Foundation, which was to receive the difference between $2 million and what the class members received if they received less than that amount, which they did. The Foundation is not a class member; anything it received would be a “cy pres ” award, explained later in this opinion. Finally, the approved settlement includes an injunction against Rexall’s making certain claims (alleged to mislead) in the advertising or marketing of its glucosamine products for 30 months.
There are six appeals (for remember that the settlement embraces six class actions), which we’ve consolidated for decision. Several of the appellants are class members, led by Theodore H. Frank of the Center for Class Action Fairness, who, as class members are authorized to do by Fed.R.Civ.P. 23(e)(5), objected to the approval of the settlement. Class counsel are several law firms, which have cross-appealed, arguing that the district court should not have modified the settlement that the parties had agreed to. But we’ll see that the problem with the district judge’s decision is not that it leans too far in favor of the objectors, as class counsel contend, but that it doesn’t lean far enough. Although appellate review of approval of class action settlements is limited, Williams v. Rohm & Haas Pension Plan, 658 F.3d 629, 634 (7th Cir.2011), it is far from pro forma, for we have described, the district judge as “a fiduciary of the class, who is subject therefore to the high duty of care that the law requires of fiduciaries.” Reynolds v. Beneficial National Bank, 288 F.3d 277, 280 (7th Cir.2002).
The district judge valued the settlement at the maximum potential payment that class members could receive, which came to $20.2 million. That valuation, which *781played a critical role in the judge’s decision as to how much to award class counsel in attorneys’ fees, comprises $14.2 million for class members (based on the contrary-to-fact assumption that every one of the 4.72 million class members who had received postcard rather than publication notice of the class action would file a $3 claim), $1.5 million for the cost of notice to the class, and $4.5 million for fees to class counsel (the judge cut this amount but allowed the amount cut to revert to Rex-all pursuant to the kicker clause and adhered to the $20.2 million estimate of the overall value of the settlement). The judge excluded, however, both the cy pres award of $1.13 million in calculating the benefit to the class, for the obvious reason that the recipient of that award was not a member of the class, and the injunction, which he valued at zero, which was proper too, as we’ll see.
The $20.2 million figure has barely any connection to the settlement’s value to the class. Notice and fees, which together account for $6 million of the $20.2 million, are costs, not benefits. The attorneys’ fees are of course not paid to the class members; and as we said in Redman v. RadioShack Corp., 768 F.3d 622, 630 (7th Cir.2014), “administrative costs should not have been included in calculating the division of the spoils between class counsel and class members. Those costs are part of the settlement but not part of the value received from the settlement by the members of the class. The costs therefore shed no light on the fairness of the division of the settlement pie between class counsel and class members.” The $14.2 million “benefit” to the class members was a fiction too. Only 30,245 claims were filed, yielding total compensation for the class members of less than $1 million.
Because the amount of the attorneys’ fees that the judge wanted to award class counsel — $1.93 million — was only 9.6 percent of $20.2 million, he thought the amount reasonable. But as we. said in the Redman case, the “ratio that is relevant ... is the ratio of (1) the fee to (2) the fee plus what the class members received.” Id. Basing the award of attorneys’ fees on this ratio, which shows how the aggregate value of the settlement is being split between class counsel and the class, gives class counsel an incentive to design the claims process in such a way as will maximize the settlement benefits actually received by the class, rather than to connive with the defendant in formulating elaimsfiling procedures that discourage filing and so reduce the benefit to the class. But $20.2 million is of course not the value of the settlement, defined as the sum of the awards to the class and to its lawyers. The class received a meager $865,284. This means the attorneys’ fees represented not 9.6 percent of the aggregate value but an outlandish 69 percent.
Had the judge approved class counsel’s request for $4.5 million in attorneys’ fees, those fees would have soared to 84 percent of the pot to be divided between class members and class counsel. Moreover, $4.5 million in attorneys’ fees would equal an average of $1,254 per hour of time put in by the lawyers and paralegals employed on the case by the law firms representing the class. Even $538 per hour, the average fee allowed by the district judge in cutting the total fees from $4.5 million to $1.93 million, would be excessive. It would imply that few if any associates or paralegals had actually been used on the case, even though most of the legal work was routine pretrial preparation. This is a further indication (if any were needed) that class counsel sought and were awarded excessive compensation.
It might make sense for the district judge in a large class action suit like this *782to appoint an independent auditor, on the authority of Fed.R.Evid. 706, to estimate the reasonableness of class counsel’s billing rates. This would of course be a discretionary decision by the judge, who would have to consider both possible delays and the cost of the auditor, a cost likely to be borne in part by the class members.
In regard to sizing the benefit of the settlement to the class, it is true that an option to file a claim creates a prospective value, even if the option is never exercised. Boeing Co. v. Van Gemert, 444 U.S. 472, 480, 100 S.Ct. 745, 62 L.Ed.2d 676 (1980), remarks that the right of class members “to share the harvest of the lawsuit upon proof of their identity, whether or not they exercise it, is a benefit in the fund created by the efforts of the class representatives and their counsel.” But in that case the “harvest” created by class counsel was an actual, existing judgment fund, and each member of the class had “an undisputed and mathematically ascertainable claim to part of a lump-sum judgment recovered on his behalf.” Id. at 479, 100 S.Ct. 745. “Nothing in the court’s order made Boeing’s liability for this amount contingent upon the presentation of individual claims.” Id. at 480 n. 5,100 S.Ct. 745. The class members were known, the benefits of the settlement had been “traced with some accuracy,” and costs could be “shifted with some exactitude to those benefiting.” Id. at 480-81, 100 S.Ct. 745. There is no fund in the present case and no litigated judgment, and there was no reasonable expectation in advance of the deadline for filing claims that more members of the class would submit claims than did.
We can imagine a case in which a lawyer sets to work diligently to make a powerful case for his client, agrees to a fee that compensates him for the work necessary to litigate the case to a successful conclusion, and has a reasonable expectation of obtaining a judgment that will exceed the agreed-upon attorney’s fee, but unforeseeable developments result in a judgment smaller than the agreed-upon fee, or even in a judgment for the defendant. In such a case the lawyer would have a right to his fee. But especially in consumer class actions, where the percentage of class members who file claims is often quite low (in this case it was 30,245 12 million = .0025, or one quarter of one percent), see, e.g., Daniel Fisher, “Odds of a Payoff in Consumer Class Action? Less Than a Straight Flush,” Forbes, May 8, 2014, www.forbes.com/sites/danielfisher/2014/05/ 08/odds-of-a-payoff-in-consumer-class-action-less-than-a-straight-flush/, the presumption should we suggest be that attorneys’ fees awarded to class counsel should not exceed a third or at most a half of the total amount of money going to class members and their counsel. In this case that range would be between $436,642 and $865,284 — a far cry from the $1.93 million that the judge awarded, and absurdly far from the $4.5 million that class counsel requested, with the connivance of Rexall, which doubtless looked forward to recapturing, as it did, a big chunk of that amount. The parties could not have thought that the attorneys’ fees that they agreed to pay class counsel, even if reduced by the district judge by almost one half, would be commensurate with the modest award to the clients, which is to say the class members. The parties may not have expected only 6.4 claims per each 1000 class members who received postcard notice to file claims. But they don’t claim to have been surprised by the low rate. They do say that the settlement they negotiated was indistinguishable from consumer class actions in which the claims rate has been much higher, but those may have been cases in which the parties did not structure the claims process with an eye *783toward discouraging the filing of claims, as appears to have happened in this case.
As experienced class action lawyers, class counsel in the present case must have known that the notice and claim forms, and the very modest monetary award that the average claimant would receive, were bound to discourage filings. The postcard sent to each of 4.72 million class members informs the recipient that to file a claim he must click on www.Glucosamine Settlement.com on his computer or cellphone or call a toll-free phone number. (The website has changed since the deadline for filing claims; we base our discussion on printed copies in the record of the relevant screens of the website as it appeared during the claims period.) The opening screen of the website contains links to six documents. One is entitled “Full Class Notice,” and if you clicked on it you would have seen a 10-page statement, largely repeating what was in the opening screen but adding more information about the case, and stating on the third page that “if you submit a claim postmarked or submitted online [by a specified date], you may be eligible to receive a check” of $3 for each bottle of the glucosamine pills you bought up to a total of 4 bottles, or $5 for up to 10 bottles if you provide proof of purchase.
Another of the links is captioned “Claim Form,” and if you clicked on that you’d see a “Glucosamine Settlement Claim Form.” The form required the claimant to list cash register receipts or other documentation indicating the date and place at which he or she had bought the product. The form advised the claimant that “The Claims Administrator and the Parties have the right to audit all claims for completeness, waste, fraud, and abuse. Filing a false claim may violate certain criminal or civil laws.” Further, the claimant was — in boldface— required to “certify under penalty of perjury that the foregoing is true and correct to the best of my knowledge.”
One would have thought, given the low ceiling on the amount of money that a member of the class could claim, that a sworn statement would-be sufficient documentation, without requiring receipts or other business records likely to have been discarded. The requirement of needlessly elaborate documentation, the threats of criminal prosecution, and the fact that a claimant might feel obliged to wade through the five other documents accessible from the opening screen of the website, help to explain why so few recipients of the postcard notice bothered to submit a claim. It’s hard to resist the inference that Rexall was trying to minimize the number of claims that class members would file, in order to minimize the cost of the settlement to it. Class counsel also benefited from minimization of the claims, because the fewer the claims, the more money Rexall would be willing to give class counsel to induce settlement. Rexall has no reason to care about the allocation of its cost of settlement between class counsel and class members; 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. When the parties to a class action expect that the reasonableness of the attorneys’ fees allowed to class counsel will be judged against the potential rather than actual or at least reasonably foreseeable benefits to the class, class counsel lack any incentive to push back against the defendant’s creating a burdensome claims process in order to minimize the number of claims. Class counsel could have done much better by the class had they been willing to accept lower fees in their negotiation with Rexall. But realism requires recognition that probably all that class counsel really care about is their fees — for $865,284 spread *784over 12 million class members is only 7 cents apiece.
The $1.13 million cy pres award to the orthopedic foundation did not benefit the class, except insofar as armed with this additional money the foundation may contribute to the discovery of new treatments for joint problems — a hopelessly speculative proposition. Cy pres (properly cy pres comme possible, an Anglo-French term meaning “as near as possible”) is the name of the doctrine that permits a benefit to be given other than to the intended beneficiary or for the intended purpose because changed circumstances make it impossible to carry out the benefactor’s intent. A familiar example is that when polio was cured, the March of Dimes, a foundation that had been established in the 1930s at the behest of President Roosevelt to fight polio, was permitted to redirect its resources to improving the health of mothers and babies.
Since the joint problems that glucosamine is supposed to alleviate are the domain of orthopedic medicine, the choice of an orthopedic institute as a recipient of money left over after, all approved class members’ claims are paid is consistent with cy pres. But there is no validity to the $1.13 million cy pres award in this case. A cy pres award is supposed to be limited to money that' can’t feasibly be awarded to the intended beneficiaries, here consisting of the class members. Notice costing $1.5 million reached 4.72 million class members. Granted, doubling the expenditure would not have doubled the number of class members notified. The 4.72 million who received postcards were all those whom Rexall knew (through pharmacy loyalty programs and the like) to have bought its glucosamine pills, while notice by publication or via the Internet tends to be ineffectual when the class consists of consumers. But the claims process could have been simplified. Or knowing that 4.72 million people had bought at least one bottle of its pills, Rexall could have mailed $3 checks to all 4.72 million postcard recipients. The Orthopedic Research and Education Foundation seems perfectly reputable, but it is entitled to receive money intended to compensate victims of consumer fraud only if it’s infeasible to provide that compensation to the victims — which has not been demonstrated.
The 30-month “injunction,” to which we now turn, is actually just a brief statement in the settlement agreement requiring Rexall to remove from its packaging claims listed in “Column 1” in an accompanying exhibit and replace them with claims listed in “Column 2.” As summarized in the parties’ memorandum urging the district judge to approve the settlement, the injunction provides that
defendants will not use any representations that the products will rebuild, b.uild, or renew cartilage (“rebuilding cartilage representations”) on the labeling and packaging for any Covered Products that they manufacture, sell, or supply to other retailers. For the same 30-month period, and as part of the negotiations related to the other types of claims made by Defendants (e.g., claims related to joint comfort or pain, flareups, soothing, cushioning, nourishing, mobility, lubrication, or support) (“palliation representations”), Defendants will be required to place on the packaging and labeling of the same products a statement that “individual results may vary.” Defendants will have a period of six months from the date of Final Approval [of the settlement] to begin shipping Covered Products with the label changes, (footnote omitted)
Notice that the practical effect of the last sentence is to shrink the 30-month term of the injunction to 24 months.
*785The passage we’ve quoted describes the final benefit that the settlement is alleged to confer on the class (as well as on consumers who first begin to shop for joint-related dietary supplements in the next 30 months and thus are not members of the class). The 30-month (actually 24-month) cutoff means that after 30 months Rexall can restore the product claims that form the foundation of this suit. It says it will be reluctant to do that because then fresh class actions will be brought against it. But if so, why would it prefer a 30-month injunction to a perpetual injunction? Were the injunction perpetual, Rexall could ask the district court to modify it should new research reveal that its allegedly false claims were true after all.
A larger objection to the injunction is that it’s superfluous — or even adverse to consumers. Given the emphasis that class counsel place on the fraudulent character of Rexall’s claims, Rexall might have an incentive even without an • injunction to change them. The injunction actually gives it protection by allowing it, with a judicial imprimatur (because it’s part of a settlement approved by the district court), to preserve the substance of the claims by making — as we’re about to see — purely cosmetic changes in wording, which Rexall in effect is seeking judicial approval of. For the injunction seems substantively empty. In place of “support[s] renewal of cartilage” Rexall is to substitute “contains a key building block of cartilage.” We see no substantive change. For “works by providing the nourishment your body needs to build cartilage, lubricate, and strengthen your joints,” is to be substituted “works by providing the nourishment your body needs to support cartilage, lubricate, and strengthen your joints.” Finally, in place of “rebuilds cartilage,” “repairs cartilage,” or “renews cartilage” Rexall may insteád advertise that its product “helps protect [or support] cartilage and helps with annoying flare-ups”; “helps to lubricate and cushion joints while supporting healthy connective tissue [or healthy cartilage]”; and that “just ‘one’ ... caplet is all you need to help protect cartilage and help ease occasional joint stress.” Again we can’t see any substantive change.
Equally dubious claims found on the original lgbel, moreover, are left unchanged, such as “maintain healthy connective tissue,” .“lubricate joints,” “maintain joint comfort,” and “improvements to joint comfort in seven days.” Rexall is required to add “individual results may vary,” but, to the extent not vague, that reminder is obvious. And no medical basis is suggested for any of the changes, or for not making changes that would bite.
Class counsel argue that “rebuilds” and “renews” are stronger claims than “strengthens” or “protects,” because most users of Rexall’s glucosamine pills are seeking to rebuild lost cartilage and that “to ‘maintain’ is a far cry from repairing or rebuilding what has been lost.” No survey or other evidence was presented that might support the proposition that purchasers are seeking to rebuild lost cartilage or would draw a sharp distinction between “rebuild” and “maintain.” (Although glucosamine is considered a viable treatment option for osteoarthritis, it may be no better, or only slightly better, than a placebo. “Osteoarthritis,” Wikipedia, http://en.wikipedia.org/wiki/Osteoarthritis.) One would think that a purchaser who was convinced by the label that taking Rexall’s glucosamine pills is all you heed to “experience true joint comfort for yourself’ would not be put off just because there is no claim that the pills achieve this miracle by rebuilding lost cartilage.
In light of the concerns we’ve expressed, we find it remarkable that at the oral argument of the appeal the lead class *786counsel told us, in justification of the injunction: “Well, from my perspective, the gain is that it’s [that is, the deleted label claims are] no longer on the labeling and they’re [that is, Rexall is] going to have to make a decision to put it back on. And so that in itself is an achievement because it has been on the labeling for over ten years.” When one of the judges replied, “That’s an achievement? I don’t understand,” the class lawyer replied: “I believe it’s actually probably an unprecedented achievement but obviously you have your doubts.” We’re not the only doubters; the district judge deemed the value of the injunctive relief unascertainable (and so treated- the value as zero). He did so, quite rightly, despite the report of an expert witness, an economist named Keith A. Reutter, which estimated that class members who continue to buy Rexall’s glucosamine pills will benefit to the tune of $8.7 million a year. He pegged the aggregate benefit of the label changes required by the injunction, to both class members and future purchasers of Rexall’s glucosamine pills, at $46.2 million during the 30-month term of the injunction. But future purchasers are not members of the class, defined as it is as consumers who have purchased Rexall’s glucosamine pills. Nor is the term really 30 months.
The basis of the expert’s estimate was a wild guess as to how much Rexall would have to cut the price of its pills in order to hold on to consumers no longer told that the product can renew or help rebuild lost cartilage. “Guess” is the word, as it depends on an arbitrary forecast — based as it was on a marketing study that had been conducted in 2002 and never updated and did not attempt to test the effect of labeling changes, similar to those required by the settlement, on consumers — as to how far Rexall will drop its price, and how many consumers it will retain or gain by doing so, as a result of the label changes. The changes seem, at least in the absence of evidence of any sort (the Reutter report is too shallow to be admissible as evidence), too trivial to have a significant effect on Rexall’s sales or price. The district judge was acting well within the scope of his discretion, therefore, in valuing the injunction at zero.
Our final concern is the reversion, or kicker, clause in the settlement agreement. This is the clause that provides that if the judge reduces the amount of fees that the proposed settlement awards to class counsel, the savings shall enure not to the class but to the defendant. This is a gimmick for defeating objectors. If the class cannot benefit from the reduction in the award of attorneys’ fees, then the objector, as a member of the class, would not have standing to object, for he would have no stake in the outcome of the dispute. The simple and obvious way for the judge to correct an excessive attorney’s fee for a class action lawyer is “to increase the share of the settlement received by the class, at the expense of class counsel.” Redman v. RadioShack Corp., supra, 768 F.3d at 632. This route is barred unless the judge invalidates the kicker clause.
Class counsel claim that often they negotiate for the benefits to the members of the class first, selflessly leaving for later any consideration of or negotiation for their award of attorneys’ fees. That claim is not realistic. For we know that an economically rational defendant will be indifferent to the allocation of dollars between class members and class counsel. Caring only about his total liability, the defendant will not agree to class benefits so generous that when added to a reasonable attorneys’ fee award for class counsel they will render the total cost of settlement unacceptable to the defendant. We invited class counsel to explain how, therefore, negotiating first for class benefits could actually benefit a class, and were left without an answer. Neither can we think *787of a justification for a kicker clause; at the very least there should be a strong presumption of its invalidity.
In closing we note with disapproval a quotation by class counsel from an opinion of ours that “because settlement of a class action, like settlement of any litigation, is basically a bargained exchange between the litigants, the judiciary’s role is properly limited to the minimum necessary to protect the interests of the class and the public. Judges should not substitute their own judgment as to optimal settlement terms for the judgment of the litigants and their counsel.” Armstrong v. Board of School Directors of City of Milwaukee, 616. F.2d 305, 315 (7th Cir.1980). That quotation is from 34 years ago and in the decades since judges have accrued much more experience with class actions and have learned that class action settlements are often quite different from settlements of other types of cases, which indeed are bargained exchanges between the opposing litigants. Class counsel rarely have clients to whom they are responsive. The named plaintiffs in a class action, though supposed to be the representatives of the class, are typically chosen by class counsel; the other class members are not parties and have no control over class counsel. The result is an acute conflict of interest between class counsel, whose pecuniary interest is in their fees, and class members, whose pecuniary interest is in the award to the class. Defendants are interested only in the total costs of the settlement to them, and not in the division of the costs between attorneys’ fees and payment to class members. We thus have “remarked the incentive of class counsel, in complicity with the defendant’s counsel, to sell out the class by agreeing with the defendant to recommend that the judges approve a settlement involving a meager recovery for the class but generous compensation for the lawyers— the deal that promotes the self-interest of both class counsel and the defendant and is therefore optimal from the standpoint of their private interests.” Eubank v. Pella Corp., 753 F.3d 718, 720 (7th Cir.2014), quoting Creative Montessori Learning Centers v. Ashford Gear LLC, 662 F.3d 913, 918 (7th Cir.2011), citing other opinions, in this and other circuits, recognizing the conflict of interest; see also Redman v. RadioShack Corp., supra, 768 F.3d at 629.
That is an accurate description of this case; and it is why objectors play an essential role in judicial review of proposed settlements of class actions and why judges must be both vigilant and realistic in that review. Theodore Frank and the other objectors flagged fatal weaknesses in the proposed settlement. The district judge made significant modifications in the settlement, but not enough. The settlement, a selfish deal between class counsel and the defendant, disserves the class. Class counsel shed crocodile tears over Rexall’s misrepresentations, describing them as “demonstrably false,” “consumer fraud,” “false representations,” and so on, and pointed out that most of the consumers of Rexall’s glucosamine pills are elderly, bought the product in containers the labels of which recite the misrepresentations — and number some 12 million. Yet only one-fourth of one percent of these fraud victims will receive even modest .compensation, and for a limited period the labels will be changed, in trivial respects unlikely to influence or inform consumers. And for conferring these meager benefits class counsel should receive almost $2 million?
The judgment is reversed and the case remanded for further proceedings consistent with this opinion.
Reversed and Remanded.