OPINION

KETHLEDGE, Circuit Judge.

A class action based, as this one is, on credible allegations that Michigan's largest health insurer engaged in price-fixing to the detriment of millions of Michigan citizens, is a case in which the public has a keen and legitimate interest. Yet the district court sealed most of the parties' substantive filings from public view, including nearly 200 exhibits and an expert report upon which the parties based a settlement agreement that would determine the rights of those millions of citizens. Class members who sought to object to the proposed settlement thus had no ability to examine the bases of what they were objecting to. To compound matters, the district court then approved the settlement without meaningful scrutiny of the settlement's fairness to unnamed members of the class. We vacate the district court's order approving the settlement, vacate the orders to seal, and remand for an open and vigorous examination of the settlement's fairness to the class.

I.

A.

Blue Cross Blue Shield of Michigan controls more than 60% of the commercial health-insurance market in Michigan. Meanwhile, privately insured patients are much more profitable for hospitals than are patients insured by Medicare or Medicaid. Few if any of the hospital systems in Michigan can afford to turn away an insurer that brings with it more than half of the privately insured patients in Michigan. And that means, for better or worse, that Blue Cross enjoys extraordinary market power in its negotiations with medical providers.

Beginning no later than 2007, according to a complaint filed in federal court by the United States Department of Justice, Blue Cross used that power for worse. Specifically, around that time, Blue Cross began insisting on so-called "MFN" and "MFN-plus" agreements with hospitals in Michigan. Per the MFN agreements, Blue Cross agreed to raise its own reimbursement rates for each hospital's services, so long as the hospital agreed to charge other commercial health insurers rates at least as high as the hospital charged Blue Cross. Blue Cross obtained MFN agreements with more than 40 hospitals, most of them small, community hospitals.

Blue Cross also entered into MFN-plus agreements with 22 of the largest hospital systems in Michigan, including three Beaumont hospitals in Southeast Michigan, Sparrow Hospital in Lansing, Metro Health Hospital in Grand Rapids, and Marquette General Hospital in the Upper Peninsula. Per the MFN-plus agreements, Blue Cross agreed to pay higher rates to each hospital so long as the hospital agreed to charge even higher rates — up to 40% higher, according to DOJ's complaint — to other commercial health-insurers. The record in this case also reflects that, the greater the spread between Blue Cross's rates and the minimum rates for other insurers, the higher the rates that Blue Cross was willing to pay. For example, Blue Cross would agree to pay a certain rate if a hospital agreed to charge other insurers 10% more than Blue Cross; but if the hospital agreed to charge other insurers even more — say, 20 or 30% higher rates than Blue Cross — then Blue Cross would agree to pay even higher rates. Thus, the effect of Blue Cross's market power was not to lower its customers' rates, as typically advertised. Instead the effect was to raise them, for Blue Cross's customers and everyone else — while preserving or expanding Blue Cross's market share.

B.

The Department of Justice filed its complaint in October 2010, seeking declarative and injunctive relief. Within two weeks, various individual and corporate plaintiffs filed putative class actions against Blue Cross based upon the same price-fixing scheme. Those complaints (including the operative complaint here) adopted almost verbatim many of the allegations in DOJ's complaint. Eventually the district court consolidated the putative class actions and appointed interim class counsel, who then filed an amended complaint on behalf of the consolidated plaintiffs. The amended complaint alleged damages of more than $13.7 billion, and sought an award of treble damages under the Sherman Act, see 15 U.S.C § 15, in addition to an award of costs and attorneys' fees.

In March 2013, the State of Michigan enacted legislation banning the use or enforcement of MFN clauses like those negotiated by Blue Cross. That legislation afforded DOJ the principal relief it sought in its complaint. A week later, DOJ voluntarily dismissed its suit.

Meanwhile, in July 2012, Blue Cross moved to dismiss the amended complaint for failure to state a claim. The district court denied the motion and the parties commenced discovery. To analyze the data obtained during discovery, interim class counsel hired an antitrust expert, Dr. Jeffrey Leitzinger. Thereafter interim class counsel filed a motion for class certification, attaching 90 exhibits. Blue Cross filed a brief in opposition, with 42 exhibits. All of these materials were filed under seal.

Leitzinger completed his report in October 2013. The report describes the MFN scheme in detail and tentatively concludes that damages of approximately $118 million "can be calculated in a class-wide, formulaic fashion." Report at 45. Whether Leitzinger thinks that additional damages can be calculated on an individual basis is unclear. In February 2014, Blue Cross filed a so-called Daubert motion to exclude Leitzinger's report and testimony, attaching his report and 34 other exhibits. All of these materials again were filed under seal, even though the report does not discuss any patient information and for the most part discusses only the antitrust implications of unsealed allegations in the amended complaint.

At that point the parties began settlement negotiations. They reached agreement in June 2014. Per the agreement, Blue Cross agreed to pay just under $30 million — about one-quarter the damages that Leitzinger thought calculable on a class-wide basis — into a settlement fund. Blue Cross further agreed to the following: (i) not to oppose class counsel's request for fees, so long as the amount of the request did not exceed approximately $10 million; (ii) not to oppose class counsel's motion for expenses, which purportedly totaled $3.5 million; and (iii) not to oppose so-called "incentive awards" to the named plaintiffs, in amounts up to $10,000 per individual and $50,000 per organization. All of these amounts — the fees, the expenses, and the incentive awards — would be paid out of the settlement fund. An additional $1 million from the fund would go to an administrator to facilitate disbursements to the class. After all these deductions, $14,661,560 — just over 12% of the damages calculated by Leitzinger, and less than 4% of the damages and fees the class would recover if successful at trial, see 15 U.S.C. § 15 — would remain for allocation among the three to seven million class members.

Three days after the parties reached agreement — still in June 2014 — the district court preliminarily approved the settlement's terms, granted the earlier motion to certify the class, and made the interim class counsel permanent. The court's order of preliminary approval required that notice of the proposed settlement be sent to members of the class within 40 days, and that any members who wished to opt out of the class or object to the settlement do so within 50 days after that. Class members who sought to examine the court record or the bases for the settlement agreement, however, found that most of the key documents were either heavily redacted or, as in the case of Leitzinger's report, completely sealed.

Various class members nonetheless objected to the settlement, arguing that its amount was too small and that too much of the settlement fund would go to attorneys' fees and expenses. The objectors also argued, among other things, that their lack of access to the court record impaired their ability to assess the settlement's fairness and that the claims process was unduly burdensome. A group of 26 self-insured businesses with health plans administered by Blue Cross (a group we call the "Varnum Group") also sought to unseal the court record by means of a motion to intervene for that limited purpose.

A month later, the district court held a fairness hearing. For the most part, the parties and each of the objectors presented their arguments without questions from the court. Thereafter the court issued an opinion approving the proposed settlement agreement as fair, reasonable, and adequate, and denying the objecting class members' motion to intervene.

This appeal followed.

II.

A.

We begin with the question whether the district court abused its discretion when, at the parties' behest, it sealed from public view most of the court filings and exhibits that underlay the proposed settlement in this case.

1.

By way of background, there is a stark difference between so-called "protective orders" entered pursuant to the discovery provisions of Federal Rule of Civil Procedure 26, on the one hand, and orders to seal court records, on the other. Discovery concerns the parties' exchange of information that might or might not be relevant to their case. "Secrecy is fine at the discovery stage, before the material enters the judicial record." Baxter Int'l, Inc. v. Abbott Labs., 297 F.3d 544, 545 (7th Cir. 2002). Thus, a district court may enter a protective order limiting the use or disclosure of discovery materials upon a mere showing of "good cause[.]" Fed. R. Civ. P. 26(c)(1). These orders are often blanket in nature, and allow the parties to determine in the first instance whether particular materials fall within the order's protection. The district court entered several such orders here.

"At the adjudication stage, however, very different considerations apply." Joy v. North, 692 F.2d 880, 893 (2d Cir. 1982). The line between these two stages, discovery and adjudicative, is crossed when the parties place material in the court record. Baxter, 297 F.3d at 545. Unlike information merely exchanged between the parties, "[t]he public has a strong interest in obtaining the information contained in the court record." Brown & Williamson Tobacco Corp. v. F.T.C., 710 F.2d 1165, 1180 (6th Cir. 1983). That interest rests on several grounds. Sometimes, the public's interest is focused primarily upon the litigation's result — whether a right does or does not exist, or a statute is or is not constitutional. In other cases — including "antitrust" cases, id. at 1179 — the public's interest is focused not only on the result, but also on the conduct giving rise to the case. In those cases, "secrecy insulates the participants, masking impropriety, obscuring incompetence, and concealing corruption." Id. And in any of these cases, the public is entitled to assess for itself the merits of judicial decisions. Thus, "[t]he public has an interest in ascertaining what evidence and records the District Court and this Court have relied upon in reaching our decisions." Id. at 1181; see also, e.g., Baxter, 297 F.3d at 546.

The courts have long recognized, therefore, a "strong presumption in favor of openness" as to court records. Brown & Williamson, 710 F.2d at 1179. The burden of overcoming that presumption is borne by the party that seeks to seal them. In re Cendant Corp., 260 F.3d 183, 194 (3d Cir. 2001). The burden is a heavy one: "Only the most compelling reasons can justify non-disclosure of judicial records." In re Knoxville News-Sentinel Co., 723 F.2d 470, 476 (6th Cir. 1983). Moreover, the greater the public interest in the litigation's subject matter, the greater the showing necessary to overcome the presumption of access. See Brown & Williamson, 710 F.2d at 1179. For example, in class actions — where by definition "some members of the public are also parties to the [case]" — the standards for denying public access to the record "should be applied... with particular strictness." Cendant, 260 F.3d at 194. And even where a party can show a compelling reason why certain documents or portions thereof should be sealed, the seal itself must be narrowly tailored to serve that reason. See, e.g., Press-Enter. Co. v. Superior Court of California, Riverside Cnty., 464 U.S. 501, 509-11, 104 S.Ct. 819, 78 L.Ed.2d 629 (1984). The proponent of sealing therefore must "analyze in detail, document by document, the propriety of secrecy, providing reasons and legal citations." Baxter, 297 F.3d at 548.

In like fashion, a district court that chooses to seal court records must set forth specific findings and conclusions "which justify nondisclosure to the public." Brown & Williamson, 710 F.2d at 1176. That is true even if neither party objects to the motion to seal, as apparently neither did in Brown & Williamson. (There, our court "reach[ed] the question" of the district court's seal "on our own motion." Id.) As our decision there illustrates, a court's obligation to explain the basis for sealing court records is independent of whether anyone objects to it. And a court's failure to set forth those reasons — as to why the interests in support of nondisclosure are compelling, why the interests supporting access are less so, and why the seal itself is no broader than necessary — is itself grounds to vacate an order to seal. Id.; see also United States v. Kravetz, 706 F.3d 47, 60 (1st Cir. 2013) ("Appellate courts have on several occasions emphasized that upon entering orders which inhibit the flow of information between courts and the public, district courts should articulate on the record their reasons for doing so"); SEC v. Van Waeyenberghe, 990 F.2d 845, 849 (5th Cir. 1993) (reversing because "[w]e find no evidence in the record that the district court balanced the competing interests prior to sealing the final order").

2.

These standards went unmet here. We review the court's orders to seal its records for an abuse of discretion — although, "[i]n light of the important rights involved, the district court's decision is not accorded" the deference that standard normally brings. Knoxville News-Sentinel, 723 F.2d at 476.

The documents placed under seal in this case include, among other things, the Plaintiffs' Amended Complaint, the Plaintiffs' Motion for Class Certification and Blue Cross's Response, and Blue Cross's motion to strike the report and testimony of the Plaintiffs' expert witness, Dr. Jeffrey Leitzinger (for whose services, under the settlement agreement, the class members would pay more than $2 million). And sealed along with the parties' filings were 194 exhibits to them — including Leitzinger's report, whose valuation of the class's claims by all accounts was the keystone of the settlement agreement. As a practical matter, therefore, both the general public and the class were able to access only fragmentary information about the conduct giving rise to this litigation, and next to nothing about the bases of the settlement itself.

The parties' asserted bases for sealing off all this information were brief, perfunctory, and patently inadequate. For example, when the Plaintiffs sought to seal their brief in support of their Motion for Class Certification — arguably the most important filing in any putative class action — and all 90 attachments thereto, the Plaintiffs' entire justification for filing these materials under seal was the following: "The Class Certification Brief includes quotations, information, and references to multiple depositions and documents designated as confidential by Blue Cross or the third party entity that produced the document or deposition." Those are protective-order justifications, not sealing-order ones. Indeed those grounds are precisely like the ones Judge Easterbrook rejected out of hand in Baxter; and here, as there, "[s]o perfunctory was this motion that it could have been summarily rejected." 297 F.3d at 546. Yet here the motion was granted, in an order reciting precisely the same justification — along with a reference to "good cause shown." And though that order by its terms addressed only the Plaintiffs' brief in support of their class-certification motion, Blue Cross, for its part, did not even bother to seek permission from the court before filing its response under seal. The parties' motions to seal the other filings and exhibits, and the court's orders granting them, follow the same perfunctory pattern, usually with the same one-sentence justification.

In sealing all these documents and exhibits, the parties and the district court plainly conflated the standards for entering a protective order under Rule 26 with the vastly more demanding standards for sealing off judicial records from public view. That Blue Cross presumed to decide for itself whether to file under seal its opposition to the class-certification motion is telling in that respect. So is the complete absence of any explanation, by the parties or the court, as to why the interests in favor of closure were compelling, or why those interests outweighed the public interest in access to court records — this, in a case of great importance to the public — or why the decision to seal off broad swaths of the court record was nonetheless narrowly tailored. One can only conclude that everyone in the district court was mistaken as to which standard to apply. But one point is unmistakable: on the showings set forth in this record, every document that was sealed in the district court was sealed improperly.

Blue Cross offers some arguments in response, the first being that the Varnum Group's attempt to unseal the court records was "untimely." That argument is meritless because Varnum should not have needed to ask permission to review the records in the first place. Nor — as our decision to address the same issue sua sponte in Brown & Williamson illustrates — is an argument against sealing the kind of argument that can be waived. A court's obligation to keep its records open for public inspection is not conditioned on an objection from anybody.

Blue Cross also argues that "this Court has repeatedly held that objectors' right to discovery is limited." Blue Cross Br. at 47. But there is an obvious difference (so obvious that one wonders how a party could overlook it) between limiting the right of class members to take new discovery after settlement, and denying them the right to view materials already in the court record. The latter is all that the Varnum Group sought here.

A third argument is that "the discovery conducted in this case and the [DOJ] action ... required more than 100 third-party hospitals and commercial insurers to submit competitively-sensitive financial and negotiating information, as well as confidential patient health information. The Protective Order prohibited the disclosure of such information to anyone other than outside counsel for the parties to the litigation." Blue Cross Br. at 43. Here again we see the standards for protective orders and sealing conflated: that a mere protective order restricts access to discovery materials is not reason enough, as shown above, to seal from public view materials that the parties have chosen to place in the court record. And as to materials actually in the record, our review of the sealed materials reveals scarcely any "confidential patient-health information," which is unsurprising given that such information would be irrelevant to the antitrust issues presented in this case.

That leaves the concern about "competitively-sensitive financial and negotiating information[.]" But that concern is inadequate for any number of reasons. One is that the proponents of closure bear the burden of showing that "disclosure will work a clearly defined and serious injury[.]" Cendant, 260 F.3d at 194 (internal quotation marks omitted). And "[i]n delineating the injury to be prevented, specificity is essential." Id. Blue Cross offers only platitudes here. Another is that, "[i]n civil litigation, only trade secrets, information covered by a recognized privilege (such as the attorney-client privilege), and information required by statute to be maintained in confidence (such as the name of a minor victim of a sexual assault)," is typically enough to overcome the presumption of access. Baxter, 297 F.3d at 546; see also Brown & Williamson, 710 F.2d at 1180 ("In such cases, a court should not seal records unless public access would reveal legitimate trade secrets, a recognized exception to the right of public access to judicial records"). Suffice it to say that "financial and negotiating information" about a practice since outlawed by the Michigan Legislature is not entitled to protection as a legitimate trade secret. Cf. Joy, 692 F.2d at 894 ("The potential harm asserted by the corporate defendants is in disclosure of poor management in the past. That is hardly a trade secret"). And otherwise nobody in the district court or on appeal has even tried to demonstrate that any of the third-party "financial and negotiation information" in the court record (as opposed to information merely exchanged in discovery) is comparable to a trade secret.

Finally, the point about third parties is often one to take seriously; "the privacy interests of innocent third parties should weigh heavily in a court's balancing equation." United States v. Amodeo, 71 F.3d 1044, 1050 (2d Cir. 1995). For example, in Knoxville News-Sentinel — a case involving litigation between a bank and the FDIC — we affirmed the district court's decision to seal bank records containing financial information about the bank's customers. Various "statutory provisions and regulatory rules" protected the customers' information from disclosure, 723 F.2d at 476; and the "order of the district court [was] narrowly tailored to serve that interest." Id. at 477. Moreover, the customers themselves had done nothing to give rise to the suit. Id.

Here, as to the third-party hospitals, there is no statutory or regulatory privilege that protects their information from disclosure, and the particulars of years-ago negotiations are unlikely to amount to a trade secret. Moreover, the conduct of the hospitals that agreed to MFN clauses with Blue Cross is intricately bound up with the subject of the suit; the public has a keen interest in that conduct; and the hospitals have a lesser privacy interest than did the bank customers. Finally, what little information there is in the record from third-party insurers is for the most part highly generalized (as in the expert report) or consists of the insurers' complaints about Blue Cross's MFN practices. In any event, the parties or the third parties themselves remain free on remand to demonstrate — on a document-by-document, line-by-line basis — that specific information in the court record meets the demanding requirements for a seal.

In sum, the district court abused its discretion when it sealed the various filings and 194 exhibits in this case. That is most acutely true with respect to Leitzinger's report. And one cannot say in any realistic sense that the error was harmless. It is true, as the district court said during the fairness hearing, that the self-insured employers in the Varnum Group know the amounts they paid for health care during the periods in which the MFN and MFN-plus agreements were in effect. What those employers and other class members cannot know, however, is whether it does or does not make sense to them to accept about 12 cents on the dollar of their damages as estimated in Leitzinger's report, or less than 4 cents on the dollar of the award they might obtain at trial. To make that judgment, the class members must be able actually to read the report — for which, under the settlement agreement, the class members would be required to pay more than $2 million — along with the scores of other documents they could not read in the court record below. Only then can they assess for themselves the likelihood of success on their claims.

Class members cannot participate meaningfully in the process contemplated by Federal Rule of Civil Procedure 23(e) unless they can review the bases of the proposed settlement and the other documents in the court record. All of the named parties to this case signed on to the proposed settlement only after reviewing the expert report and other sealed documents in the record. The unnamed class members are entitled to do the same, subject to the rights of the parties and third parties to make the showings necessary to seal. The Rule 23(e) objection process seriously malfunctioned in this case, and that is reason enough to vacate the district court's approval of the settlement.

B.

To guide the proceedings on remand, we discuss one very significant omission and several lesser ones in the proceedings leading up to this appeal.

1.

Again by way of background, "class-action settlements affect not only the interests of the parties and counsel who negotiate them, but also the interests of unnamed class members who by definition are not present during the negotiations. And thus there is always the danger that the parties and counsel will bargain away the interests of unnamed class members in order to maximize their own." In re Dry Max Pampers Litig., 724 F.3d 713, 715 (6th Cir. 2013). That is not an indictment of any parties or counsel in particular; it is merely a recognition of the adverse incentives at work in class-action settlements. And because those incentives are adverse, the district court "must carefully scrutinize" whether the named plaintiffs and counsel have met their fiduciary obligations to the class, id. at 718, and whether the settlement itself is "fair, reasonable, and adequate." Fed. R. Civ. P. 23(e)(2).

"The fairness of each settlement turns in large part on the bona fides of the parties' legal dispute." Int'l Union, UAW v. Gen. Motors Corp., 497 F.3d 615, 631 (6th Cir. 2007). Thus, a district court (like this court) "cannot judge the fairness of a proposed compromise without weighing the plaintiff's likelihood of success on the merits against the amount and form of the relief offered in the settlement." Id. (internal quotation marks omitted). That means the district court "cannot substitute the parties' assurances or conclusory statements for its independent analysis of the settlement terms." In re Pet Food Prods. Liab. Litig., 629 F.3d 333, 350-51 (3d Cir. 2010). Nor can the court merely recite the respective arguments of the objectors and parties, and then offer up platitudes about the risks of litigation generally. Instead, the district court must specifically examine what the unnamed class members would give up in the proposed settlement, and then explain why — given their likelihood of success on the merits — the tradeoff embodied in the settlement is fair to unnamed members of the class.

The district court did not make that examination here. The court's discussion of the plaintiffs' likelihood of success runs only a single paragraph, in which it stated that "extensive discovery has been held in this case" — which says nothing either way about the likelihood of success — and that "in light of the Plaintiff[s'] expert's analysis as to damages, the Court finds that the settlement amount reached by the parties is fair in light of any success the Plaintiffs may obtain on the merits of the case." Op. at 26-27. But the latter point is conclusory, and does not explain why it would be fair for the unnamed class members to receive only 12% of Leitzinger's estimate of their damages, and only 4% of the damages to which they would be entitled (based on that estimate) if they prevailed at trial. See 15 U.S.C. § 15. The district court also mentioned the "significant risk that the class members could receive nothing or some negligible amount in damages at trial or on appeal." Op. at 27. But again that statement is either conclusory, or just a statement about the risks of litigation generally. The district court's opinion thus leaves an "analytical gap" between the data on which it relies — namely the court record, much of which was sealed — and its conclusion that the settlement is fair to unnamed class members. Cf. Gen. Elec. v. Joiner, 522 U.S. 136, 146, 118 S.Ct. 512, 139 L.Ed.2d 508 (1997). On remand, the district court needs to fill that gap with a careful analysis that acknowledges what the unnamed class members would give up in this settlement and what the named class members and class counsel would receive, and includes a reasoned explanation as to whether — in light of the merits of this case specifically — the settlement is fair.

2.

Next, in approving class counsel's fee request of approximately $10 million, the district court erred in relying in part on the higher amount calculated by means of the "lodestar method." That method examines whether the rates and hours spent by class counsel are reasonable. The court's error was specific to the record here. One problem is that the rates claimed by class counsel are exceedingly high: some 20 lawyers billed the class more than $700 per hour, and some billed more than $900 per hour; and over 40 paralegals charged an average of $228 per hour, which is more than $10 per hour higher than the rates charged by the top 1% of paralegals nationwide. See Nat'l Ass'n of Legal Assistants & Paralegals, 2015 National Utilization and Compensation Survey Report, Section 3 Billing Rates, at 2-3 available at https://www.nala.org/sites/default/files/15SEC3.pdf. Class counsel also billed various administrative personnel at rates up to $175 per hour, including a receptionist billing $125 per hour. These are Bentley rates, not Cadillac rates, and if a district court thinks they are relevant to the fees that unnamed class members should actually pay, the court must explain why.

Moreover, class counsel provided no backup whatsoever — no time records, no descriptions of work done — in support of their hours spent working on the case. Instead, class counsel provided the district court with a single page of documentation for each firm, listing only the employee names, titles, rates, hours, and — by multiplying the rates and hours — the total lodestar for that firm. The documentation of fees in this case should have only heightened the district court's concerns about whether the settlement is fair to unnamed members of the class.

3.

The court's approval of so-called "incentive awards" for the named plaintiffs raises similar concerns. As an initial matter, contrary to the district court's assertion, our court has not "stressed that incentive awards are [] efficient ways of encouraging members of a class to become class representatives and rewarding individual efforts taken on behalf of the class." Op. at 32. Nothing in our opinion in Hadix v. Johnson, 322 F.3d 895 (6th Cir. 2003), adopts that proposition as our own. Instead, in Dry Max, we said that "[o]ur court has never approved the practice of incentive payments to class representatives, though in fairness we have not disapproved the practice either." 724 F.3d at 722. And in Hadix we recognized a "sensibl[e] fear that incentive awards may lead named plaintiffs to expect a bounty for bringing suit or to compromise the interest of the class for personal gain." 322 F.3d at 897.

We have concerns about a bounty here. The settlement agreement provides for incentive awards of up to $10,000 per individual named plaintiff and up to $50,000 per organization. Class counsel argues in conclusory terms that the awards compensate the named plaintiffs for their time spent on the case. To ensure that these amounts are not in fact a bounty, however, counsel must provide the district court with specific documentation — in the manner of attorney time sheets — of the time actually spent on the case by each recipient of an award. Otherwise the district court has no basis for knowing whether the awards are in fact "a disincentive for the [named] class members to care about the adequacy of relief afforded unnamed class members[.]" Dry Max, 724 F.3d at 722 (emphasis in original). That does not mean the court should necessarily approve the awards if counsel provides this documentation. But it does mean that, on the factual record at least, the "difficult issue" of the propriety of incentive awards would be properly presented. Hadix, 322 F.3d at 898.

4.

Finally, we note that the district court did not respond to at least one of the Varnum Group's objections, namely that the claims process is unduly burdensome. On remand, the court should respond to that objection along with any others that are presented to it. As to that objection and any other matter not specifically addressed herein, however, we express no opinion.

* * *

The district court must begin the Rule 23(e) process anew. We vacate the district court's approval of the settlement, vacate all of its orders sealing documents in the court record, and remand for further proceedings consistent with this opinion.