EQT Production Company v. Robert Adair

Vacated and remanded by published opinion. Judge DIAZ wrote the opinion, in which Judge WILKINSON and Judge KEENAN joined.

This appeal arises from the district court’s decision to certify five related class action suits. The plaintiffs in each of the five classes generally allege that EQT Production Co. and CNX Gas Co. have unlawfully deprived the class members of royalty payments from the production of coalbed methane gas (“CBM”) in Virginia. Four of the five classes claim that EQT and CNX have improperly remitted royalty payments to escrow or suspense accounts instead of to the royalty owners. All five classes allege that EQT and CNX have been underpaying royalties.

The defendants petitioned for permission to appeal the five orders granting class certification pursuant to Federal Rule of Civil Procedure 23(f). We deferred ruling on the petitions, consolidated the cases, and ordered formal briefing.

We now grant the appeal and conclude that the district court abused its discretion when it certified the five classes. As we explain below, Rule 23 requires a more rigorous analysis as to whether the requirements for class certification have been satisfied. We therefore vacate and remand for reconsideration of the plaintiffs’ motions for class certification.

I.

A brief explanation of the historical and statutory background is necessary to assess the implications of class certification in this case.

A.

CBM is a form of natural gas that resides in the pores of coal. When the pressure on coal is reduced — for example, from natural geologic shifts or mining — CBM is released from the surface of coal.

*353 Like any form of methane, CBM is highly explosive. Historically, miners viewed CBM as a dangerous waste product and ventilated it into the atmosphere as a safety measure. By the 1970s, however, it became apparent that CBM could be used as an energy resource, and producers began to capture it for commercial use. CBM has since been recognized in Virginia as a “distinct mineral estate,” Harrison-Wyatt, LLC v. Ratliff, 267 Va. 549, 593 S.E.2d 234, 238 (2004), which means that the rights to CBM can be severed from the land.

Questions regarding ownership of the CBM estate have long plagued its commercial development in Virginia. CBM drilling often occurs on tracts of land where different persons own the subsurface gas rights (the “gas estate”) and coal mining rights (the “coal estate”). Until recently, severance deeds generally did not mention CBM, much less assign ownership rights. At times, both gas estate owners and coal estate owners have claimed title to CBM. Further complicating matters, a CBM drilling unit — the area of land underlying and surrounding a CBM well — typically encompasses 60 to 80 acres. Multiple, separately owned tracts of land often underlie a single unit, and each tract has the potential for an ownership conflict if the coal estate has been severed from the gas estate.

B.

In 1990, the Virginia legislature enacted the Virginia Gas and Oil Act, Va.Code Ann. § 45.1-361.1 et seq., to enable producers to capture CBM “[w]hen there are conflicting claims to the ownership of coalbed methane gas.” Id. § 45.1-361.22. Upon application from a CBM producer, the Act authorizes the Virginia Gas and Oil

Board to enter orders “pooling all interests or estates in the [CBM] drilling unit for the development and operation thereof.” Id. Once issued, a pooling order consolidates all adjoining tracts of land with subsurface CBM into a single pool or unit of interests, enabling the CBM producer to extract the gas from a common reservoir. Under the Act, a pooling order deprives potential CBM owners of the right to prevent CBM extraction but does entitle CBM owners to a royalty payment.

To apply for a pooling order, producers must send notice to every “potential owner of an interest” in the CBM underlying a planned drilling unit. Id. § 45.1-361.22.1. The notices typically give each interest holder the option of reaching a voluntary lease agreement with the CBM producer prior to the entry of the final pooling order. A person who does not reach such an agreement is typically “deemed ... to have leased his gas or oil interest to the [CBM] well operator.” Id. § 45.1-361.22.6. Under the provisions of the Board’s pooling orders, deemed lessors are entitled to a royalty of one-eighth of the net proceeds received by the CBM producer for their share of the CBM.

To identify the persons to whom they must send notice, CBM producers have historically prepared ownership schedules listing all of the potential interest holders — and conflicting claimants — to the CBM involved in each drilling unit. Preparing these schedules is often an arduous process, requiring extensive research and the preparation of numerous lease reports and title opinions. The Board’s pooling orders adopt the ownership schedules submitted by the CBM producers and memorialize the ownership conflicts identified therein. 1

*354 Whenever a CBM ownership conflict is identified, the Board must establish an escrow account to receive the royalties attributable to the disputed interest. See id. § 45.1-361.22.2. The CBM producer must “deposit into the escrow account one-eighth of all proceeds attributable to the conflicting interests plus all proceeds in excess of ongoing operational expenses.” Id. § 45.1-361.22.4. As of January 2010, the Board’s escrow account contained over $25 million.

The Act provides three ways for persons with a disputed ownership claim to CBM to gain release of the escrowed funds. A claimant can obtain “(i) a final decision of a court of competent jurisdiction adjudicating the ownership of [CBM] as between [conflicting claimants]; (ii) a determination reached by an arbitrator ...; or (iii) an agreement among all claimants owning conflicting estates in the tract in question or any undivided interest therein.” Id. § 45.1-361.22.5.

II.

In this consolidated appeal, we consider the claims of five separate plaintiff classes, comprising actual or potential CBM interest holders, against two CBM producers, EQT and CNX. The Adair, Adkins, and Kiser cases involve claims against EQT, while Hale and Addison involve claims against CNX.

A.

Defendants EQT and CNX operate numerous CBM wells in Virginia, many of which are subject to Board pooling orders. 2 To apply for Board pooling orders, both EQT and CNX prepared schedules attempting to identify every potential CBM interest holder and any ownership conflict involved in each drilling unit.

In their submissions to the Board, EQT and CNX have consistently taken the position that a CBM interest is conflicted if, for a given tract of land that is part of a drilling unit, different persons own the gas estate and the coal estate. Because Board pooling orders incorporate the defendants’ schedules, those orders memorialize the ownership conflicts identified by EQT and CNX.

Buckhorn Coal Co. LLP, Commonwealth Coal Corp., and Harrison-Wyatt, LLC (collectively, the “BCH-W defendants”) intervened as defendants in the two cases against CNX — Hale and Addison. All of the BCH-W defendants have lease arrangements with CNX granting it the right to drill wells into coal seams owned by the BCH-W defendants. Based on these agreements, the BCH-W defendants claim an interest in the CBM at issue in this case. 3

*355 B.

The plaintiff classes can be categorized by their shared circumstances and requested relief.

1.

Four of the five classes — Adair, Addison, Hale, and Kiser — consist of persons who have never received CBM royalties for a CBM interest they claim to own. 4 As defined by the district court, the classes include (1) all persons or their successors, (2) whom EQT or CNX have identified as being the owners of the gas estate in a tract underlying a CBM drilling unit, (8) whose interest in the CBM is “in conflict” because a different person owns the coal estate in the same tract.

The ownership classes can be further broken down. In two cases (the “force pooled” classes) — Adair and Hale — the plaintiffs’ purported CBM interests have been force pooled by a Board order.

In the other two ownership cases (the “voluntary lease” classes) — Kiser and Addison — the defendants entered voluntary lease arrangements with the putative class members. Nonetheless, the class members’ CBM interests have been subject to pooling, and their royalties have either been paid into Board escrow accounts or internally withheld by EQT and CNX. 5

The primary object of the ownership classes is to obtain the release of escrowed or suspended royalties. To that end, they seek a declaratory judgment that: (1) the ownership conflict EQT and CNX identified between gas estate owners and coal estate owners is “illusory”; (2) as gas estate owners, the class members are entitled to the CBM royalties withheld; and (3) any royalties held in escrow or internally suspended by EQT and CNX as a result of the “illusory” ownership conflict must be paid to the class members.

2.

The fifth class — Adkins—is unique, as it consists of persons whose CBM ownership interest is not disputed. Instead, the putative class includes persons who have received a royalty from EQT at some point since January 1, 1995. The Adkins plaintiffs allege that EQT has systematically underpaid CBM royalties. The four other classes make similar claims against the defendants. Each of the classes seek a complete accounting of the royalties EQT and CNX have remitted to class members, paid into escrow, or internally suspended.

In addition to the declaratory judgment relief sought by the ownership classes, each class alleges a variety of other theories of recovery, including tort, property, and contract, and they all seek punitive damages.

C.

The lead plaintiffs filed the various complaints between June 2010 and April 2011. The district court coordinated discovery and pretrial proceedings in the five cases, referring many of the preliminary motions to a magistrate judge.

After discovery and numerous hearings, the magistrate judge issued a report and recommendation (“R & R”) supporting class certification of the proposed classes and claims with two exceptions. First, the *356 magistrate judge found the claims of the class representative in Kiser — then Eva Mae Adkins 6 — atypical of the other class members, and thus recommended against certifying that class until a suitable representative could be substituted. See Adair v. EQT Prod. Co., Nos. 1:10-cv-00037, 1:10-cv-00041, 1:11-cv-00031, 1:10-cv-00059, 1:10-cv-00065, 2013 WL 5429882, at *42, *44-*45 (W.D.Va. Sept. 5, 2013). Second, the magistrate judge recommended against certifying the breach of contract claims related to the underpayment of royalties in Kiser and Adkins because the class members had different lease agreements with EQT. See id. at *42. Such variation, the magistrate judge concluded, defeated Rule 23’s requirement that class claims be typical of one another.

The district court adopted the magistrate judge’s R & R but certified additional classes and claims. See Adair v. EQT Prod. Co., No. 1:10-CV-00037, 2013 WL 5442369 (W.D.Va. Sept. 30, 2013); Addison v. CNX Gas Co., No. L10-CV-00065, 2013 WL 5442373 (W.D.Va. Sept. 30, 2013); Adkins v. EQT Prod. Co., No. 1:11-CV-00031, 2013 WL 5442378 (W.D.Va. Sept. 30, 2013); Hale v. CNX Gas Co., No. 1:10-CV-00059, 2013 WL 5429901 (W.D.Va. Sept. 30, 2013); Legard v. EQT Prod. Co., No. L10-CV-00041, 2013 WL 5429885 (W.D.Va. Sept. 30, 2013). Specifically, the district court substituted Julie A. Kiser as the class representative in Kiser and certified the class. See Adkins, 2013 WL 5442378, at *2. Additionally, and without explanation, the court certified the breach of contract claims in Kiser and Adkins. See id. at *1; Legard, 2013 WL 5429885, at *1.

Finally, the court revised the class definitions for each of the classes. Relevant to this appeal, it added language in Adkins — the pure royalty underpayment case — to limit the class to include only those royalty owners whose leases are “silent as to the deduction of costs, according to the business records maintained by EQT.” See Legard, 2013 WL 5429885, at *1. In Kiser, one of the voluntary lease cases, the court certified a class of all lease holders, but also certified a subclass of persons “whose lease is silent as to the deduction of costs.” Adkins, 2013 WL 5442378, at *1. The district court did not clarify what it meant by “silent as to the deduction of costs” in either of the certification orders.

The defendants timely filed petitions pursuant to Rule 23(f) for permission to appeal the five orders granting the plaintiffs’ motions for class certification. We deferred ruling on the petitions, consolidated the actions, and ordered briefing on the merits.

III.

As a threshold matter, we first consider the defendants’ petitions for permission to appeal under Federal Rule of Civil Procedure 23(f). That rule authorizes courts of appeals to review decisions granting or denying class certification on an interlocutory basis. See Fed.R.Civ.P. 23(f).

We apply a five-factor test to assess the appropriateness of granting a Rule 23(f) petition. See Lienhart v. Dryvit Sys., Inc., 255 F.3d 138, 145 (4th Cir.2001). The relevant factors are:

*357 (1) whether the certification ruling is likely dispositive of the litigation; (2) whether the district court’s certification decision contains a substantial weakness; (3) whether the appeal will permit the resolution of an unsettled legal question of general importance; (4) the nature and status of the litigation before the district court (such as the presence of outstanding dispositive motions and the status of discovery); and (5) the likelihood that future events will make appellate review more or less appropriate.

Id. at 144. We consider these factors on a holistic basis, but the court should grant the petition, notwithstanding the other factors, “[w]here a district court’s certification decision is manifestly erroneous and virtually certain to be reversed on appeal.” Id. at 145.

As discussed in greater detail below, class certification in this ease was manifestly improper. We therefore grant the petitions for review and assess the merits of the district court’s certification orders.

IV.

We review a district court’s decision to certify a class for abuse of discretion. Br own, v. Nucor Corp., 576 F.3d 149, 152 (4th Cir.2009). A district court abuses its discretion when it materially misapplies the requirements of Rule 23. See Gunnells v. Healthplan Servs., Inc., 348 F.3d 417, 424 (4th Cir.2003).

Rule 23(a) requires that the prospective class comply with four prerequisites: (1) numerosity; (2) commonality; (3) typicality; and (4) adequacy of representation. See Fed.R.Civ.P. 23(a). In addition, “the class action must fall within one of the three categories enumerated in Rule 23(b).” Gunnells, 348 F.3d at 423.

Here, the plaintiffs seek certification under Rules 23(b)(2) and 23(b)(3). Rule 23(b)(2) authorizes class treatment when “the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” Fed.R.Civ.P. 23(b)(2). As the Supreme Court has instructed, “[t]he key to the (b)(2) class is the indivisible nature of the ... remedy warranted.” Wal-Mart Stores, Inc. v. Dukes, — U.S. -, 131 S.Ct. 2541, 2557, 180 L.Ed.2d 374 (2011) (internal quotation marks omitted). Certification under this provision is appropriate “only when a single injunction or declaratory judgment would provide relief to each member of the class.” Id.

By contrast, certification under Rule 23(b)(3) is appropriate when all of the prerequisites of Rule 23(a) are satisfied and two other requirements are met. See id. at 2558. Specifically, (1) common questions of law or fact must predominate over any questions affecting only individual class members; and (2) proceeding as a class must be superior to other available methods of litigation. See Fed.R.Civ.P. 23(b)(3).

A party seeking class certification must do more than plead compliance with the aforementioned Rule 23 requirements. See Walr-Mart, 131 S.Ct. at 2551 (“Rule 23 does not set forth a mere pleading standard.”). Rather, the party must present evidence that the putative class complies with Rule 23. See Comcast Corp. v. Behrend, — U.S. -, 133 S.Ct. 1426, 1432, 185 L.Ed.2d 515 (2013).

To determine whether the party seeking certification has carried its burden, a district court may need to “probe behind the pleadings before coming to rest on the certification question.” Id. (internal quotation marks omitted). Although *358 Rule 23 does not give district courts a “license to engage in free-ranging merits inquiries at the certification stage,” a court should consider merits questions to the extent “that they are relevant to determining whether the Rule 23 prerequisites for class certification are satisfied.” Amgen Inc. v. Conn. Ret. Plans & Trust Funds, - U.S. -, 133 S.Ct. 1184, 1194-95, 185 L.Ed.2d 308 (2013).

It is the plaintiffs’ burden to demonstrate compliance with Rule 23, but the district court has an independent obligation to perform a “rigorous analysis” to ensure that all of the prerequisites have been satisfied. See Wal-Mart, 131 S.Ct. at 2551.

V.

In light of the foregoing principles, we first consider the district court’s decision to certify the four classes asserting CBM ownership claims. At bottom, the ownership classes seek a declaration that the class members are the true owners of CBM, as well as payment of the royalties they believe EQT and CNX have improperly escrowed or withheld.

After reviewing the magistrate judge’s R & R and the district court’s certification orders, we conclude that the district court abused its discretion in at least two ways. First, it failed to rigorously analyze whether the administrative burden of identifying class members in the ownership cases would render class proceedings too onerous. Second, the court improperly lowered the burden of proof the plaintiffs must satisfy to demonstrate the prospective classes’ compliance with Rule 23(a)’s commonality requirement. We address each issue in turn.

A.

We have repeatedly recognized that Rule 23 contains an implicit threshold requirement that the members of a proposed class be “readily identifiable.” Hammond v. Powell, 462 F.2d 1053, 1055 (4th Cir.1972); see also In re AH. Robins Co., 880 F.2d 709, 728 (4th Cir.1989) (“Though not specified in [Rule 23], establishment of a class action implicitly requires ... that there be an identifiable class.... ”), abrogated on other grounds, Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997). Our sister circuits have described this rule as an “ascertainability” requirement. See, e.g., Marcus v. BMW of N. Am., LLC, 687 F.3d 583, 592-94 (3d Cir. 2012); John v. Nat’l Sec. Fire & Cas. Co., 501 F.3d 443, 445 (5th Cir.2007); In re Initial Pub. Offerings Sec. Litig., 471 F.3d 24, 44-45 (2d Cir.2006).

However phrased, the requirement is the same. A class cannot be certified unless a court can readily identify the class members in reference to objective criteria. See Marcus, 687 F.3d at 593; see also Crosby v. Soc. Sec. Admin., 796 F.2d 576, 579-80 (1st Cir.1986) (finding that a class failed to satisfy Rule 23 requirements because it would be impossible to identify class members without “individualized fact-finding and litigation”).

The plaintiffs need not be able to identify every class member at the time of certification. But “[i]f class members are impossible to identify without extensive and individualized fact-finding or ‘mini-trials,’ then a class action is inappropriate.” Marcus, 687 F.3d at 593; see also 7A Charles Alan Wright et al., Federal Practice & Procedure § 1760 (3d ed. 2005) (“[T]he requirement that there be a class will not be deemed satisfied unless ... it is administratively feasible for the court to determine whether a particular individual is a member.”).

*359 Here, the proposed classes raise serious ascertainability issues because they are defined to include both former and current gas estate owners.

The district court defined the classes to include all persons, and their successors-in-interest, who EQT or CNX identified in their filings with the Board as being the owners of a gas estate, whose interest in CBM is conflicted because a different person owns the coal estate in the same tract. 7 The court correctly concluded that some class members will be easy to identify because the classes are all defined in reference to the ownership schedules that EQT and CNX submitted to the Board. When ownership has not changed hands, identifying class membership may be as simple as cross-referencing the ownership schedules the defendants themselves prepared. See Adair, 2013 WL 5429882, at *33.

Complications arise, however, because ownership of the gas estate has not been static since EQT and CNX first prepared the ownership schedules. Some of the schedules were prepared some twenty years ago, and they have not been updated to account for changes in ownership. The schedules therefore cannot aid a court in ascertaining those class members who obtained their interest in the gas estate after the schedules were first prepared. 8

The district court largely glosséd over this problem, merely noting that any ownership changes could be determined by reference to local land records. See id. But resolving ownership based on land records can be a complicated and individualized process. Cf. Johnson v. Kan. City S., 224 F.R.D. 382, 389 (S.D.Miss.2004) (denying certification on ascertainability grounds when determining class membership “would require individualized review of thousands of title documents containing differing and diverse conveyance language that would have to be analyzed according to the specific language used and applicable case law to ascertain the intention of the parties to the conveyances and the legal effect of the instruments”), aff'd sub nom. Johnson v. Kan. City S. Ry. Co., 208 Fed.Appx. 292, 297 (5th Cir.2006). As the record in this case highlights, numerous heirship, intestacy, and title-defect issues plague many of the potential class members’ claims to the gas estate. In our view, these complications pose a significant administrative barrier to ascertaining the ownership classes.

On appeal, the plaintiffs minimize these challenges, arguing that a court can identify current gas estate owners at the back-end. According to them, ownership issues only affect the plaintiffs’ entitlement to royalties, not the ascertainability of class membership. See Appellees’ Br. at 58-60.

We disagree. The fact that verifying ownership will be necessary for the class members to receive royalties does not mean it is not also a prerequisite to identifying the class. Without even a rough estimate of the number of potential succes *360 sors-in-interest, we have little conception of the nature of the proposed classes or who may be bound by a potential merits ruling. Lacking even a rough outline of the classes’ size and composition, we cannot conclude that they are sufficiently ascertainable.

On remand, the district court should reconsider the ascertainability issues posed by the ownership classes. At a minimum, the district court should endeav- or to determine the number of potential class members who have obtained their interest in the gas estate after the defendants first prepared the ownership schedules. The court should also give greater consideration to the administrative challenges it will face when using land records to determine current ownership, and assess whether any trial management tools are available to ease this process. The district court should also determine whether it is possible to adjust the class definitions to avoid or mitigate the administrative challenges we have identified. 9

B.

In addition to questioning the ascertain-ability of the ownership classes, the defendants challenge the district court’s conclusion that the ownership classes comply with Rule 23(a)’s commonality requirement. As discussed previously, Rule 23(a)(2) requires a plaintiff to show that “there are questions of law or fact common to the class.” Fed.R.Civ.P. 23(a)(2).

Although the rule speaks in terms of common questions, “what matters to class certification ... [is] the capacity of a classwide proceeding to generate common answers apt to drive the resolution of the litigation.” Wal-Mart, 131 S.Ct. at 2551 (internal quotation marks omitted). A single common question will suffice, id. at 2556, but it must be of such a nature that its determination “will resolve an issue that is central to the validity of each one of the claims in one stroke,” id. at 2551.

As we explain below, the plaintiffs in the ownership classes have yet to identify such a question.

1.

To a great extent, commonality for the ownership classes turns on the proper meaning of the Supreme Court of Virginia’s decision in Harrison-Wyatt. In that case, the court considered a 19th century severance deed conveying “all the coal in, upon, and underlying” certain tracts of land. Harrison-Wyatt, 593 S.E.2d at 235 (internal quotation marks omitted). The court held that the conveyance of coal did not transfer title to the CBM estate, and that the grantor — the surface owner — retained ownership of the CBM. See id. at 238. The Virginia legislature subsequently codified that holding as part of the Virginia Oil and Gas Act, providing that “[a] conveyance, reservation, or exception of coal shall not be deemed to include coalbed methane gas.” Va.Code Ann. § 45.1-361.21:1.

The plaintiffs interpret Harrison-Wyatt and the Act to mean that a severance deed conveying coal never transfers title to CBM, and that the owner of the gas estate in a tract of land owns the underlying CBM as a matter of law. Since the plaintiffs have all been identified as gas estate *361 owners by EQT and CNX, they believe the question of CBM ownership can be resolved on a classwide basis — and in their favor.

The defendants say that the relevant authorities only establish that deed language conveying coal — and only coal — does not transfer title to CBM. But, they contend, deed language varies significantly, and broader conveyances may transfer CBM. They maintain that CBM ownership can only be determined on a deed-by-deed basis by examining the intent of the parties. According to the defendants, the need for such individualized review defeats commonality.

Although the district court did not rule on the meaning of Harrison-Wyatt, it agreed with the plaintiffs that the case gave rise to at least one common question capable of classwide resolution. See Adair, 2013 WL 5429882, at *36. Specifically, the court agreed that whether Harrison-Wyatt entitles the plaintiffs to CBM royalties is a question “subject to a common resolution.” Id. 10

We conclude that certification based on this question was premature. Prior to certifying a class, a district court must definitively determine that the requirements of Rule 23 have been satisfied, even if that determination requires the court to resolve an important merits issue. See Gariety v. Grant Thornton, LLP, 368 F.3d 356, 365-66 (4th Cir.2004). The district court failed to do so here by refusing to resolve — one way or the other — the implications of Har-risorWWyatt for commonality purposes.

Here, the meaning of Harrison-Wyatt is inescapably part of the Rule 23(a) analysis. To even demonstrate commonality, the plaintiffs must prevail on their reading of the case. That is, they must establish that the common question — who owns the CBM — will be answered in their favor. If Harrison-Wyatt does not support such a conclusion, the plaintiffs have no other argument as to how CBM ownership can be resolved on a classwide basis, and they will have failed to carry their burden of establishing even a single common question. Cf Phillips v. Asset Acceptance, LLC, 736 F.3d 1076, 1081 (7th Cir.2013) (concluding that the district court erred when it declined to decide a merits issue before certifying the class when resolving the question would “determine whether the suit could be maintained as a class action at all”).

The district court abused its discretion by failing to resolve the meaning of Harrison-Wyatt prior to certification. Although the court noted its probable agreement with the plaintiffs’ reading of the case, it declined to decide the matter one way or the other. By leaving the issue unresolved, the court improperly left open, at the time of certification, whether CBM ownership is an individual or common question. Certifying a class in the face of such uncertainty runs afoul of the rule that *362 “actual, not presumed, conformance with Rule 23(a) [is] ... indispensable.” Gen. Tel. Co. of Sw. v. Falcon, 457 U.S. 147, 160, 102 S.Ct. 2864, 72 L.Ed.2d 740 (1982).

2.

We also do not believe Harrisonr-Wyatt and the Virginia Oil and Gas Act can provide classwide answers to the question of CBM ownership, at least as the classes are currently defined. Although we do not hold that the plaintiffs can never satisfy Rule 23’s commonality requirement, we believe the district court misread the implications of those authorities when it certified the ownership classes.

We read Harrison-Wyatt and the Act to establish only that a surface owner’s conveyance of coal — and only coal — does not automatically transfer title to CBM. But many of the severance deeds at issue in this case explicitly convey much more than coal. For example, one deed in Hale confers “[a]ll the coal, minerals, petroleum, metallic substances, fluids and gas of every description, in, upon, or underlying that certain tract of land.” J.A. 1780. A different Hale deed grants “all the coal and mineral of every description, in, on and underlying that certain tract.” J.A. 1784. Yet another deed from the same class transfers “all the coal and other substances, properties, rights and interests in and upon that certain tract of land.... ” J.A. 1793. Neither Harrisonr-Wyatt nor the Act fully resolves who owns the CBM under these broader deeds.

We also note that lower Virginia courts have not adopted the plaintiffs’ reading of Harrison-Wyatt. Instead, they continue to resolve CBM ownership conflicts on a deed-by-deed basis, looking at the language of the deeds in each ease. 11 See, e.g., Wade v. Hugh MacRae Land Trust, CL09000476-00, at 3 (Va.Cir.Ct. Aug. 31, 2010) (suggesting that Harrison-Wyatt gives rise to a presumption that a severance deed conveying only coal does not transfer title to the CBM estate, but noting that such a presumption is rebutta-ble). 12

The plaintiffs’ reading is also at odds with longstanding principles of Virginia contract law, which require courts to review deed language to ascertain the parties’ intent. See, e.g., Vicars v. First Va. Bank-Mountain Empire, 250 Va. 103, 458 S.E.2d 293, 294-95 (1995) (stating that ownership rights are determined by the construction of deeds, which requires a court to determine the grantor’s intent); Virginian Ry. Co. v. Avis, 124 Va. 711, 98 S.E. 638, 639 (1919) (“The purpose of all written ... conveyances is to say what the parties mean, and the only legitimate or permissible object of interpreting them is to determine the meaning of what the parties have said therein.”). 13

*363 If ownership cannot be established on the basis of Harrisorir-Wyatt and the Act alone, we see no way for the district court to answer the ownership question on a common basis. Rather, the court will need to resolve each ownership conflict with reference to specific deed language. Such individualized review precludes a finding of commonality. See, e.g., Isaacs v. Sprint Corp., 261 F.3d 679, 682 (7th Cir.2001) (finding class certification “decidedly inappropriate” when the case involved “different conveyances by and to different parties made at different times over a period of more than a century”); Johnson, 208 Fed.Appx. at 297 (concluding that a class failed to satisfy Rule 23(a) when the case involved “a multitude of property owners, each with individual conveyances stating different things”).

This is not to say that certification could never be proper for any of the ownership classes or some subdivision thereof. Harrison-Wyatt may provide a common answer to the ownership question for a class of gas estate owners whose severance deeds convey coal and only coal. Likewise, the plaintiffs may be able to identify a finite number of variations in deed language, such that the ownership question is answerable on a subclass basis. Cf Fisher v. Va. Elec. & Power Co., 217 F.R.D. 201, 216-17 (E.D.Va.2003) (granting certification when the easements at issue were “the product of a limited set of substantially similar conveyances,” so that “determining the relevant property interest [would] require analysis of only a limited array of easement language and the vast majority of conveyances at issue contain[ed] substantially similar language”). That the deeds may be classifiable will not, by itself, mean that there is an adequate common question. But it may aid the district court’s analysis of Rule 23(a)’s requirements. 14

As it stands, however, neither the plaintiffs nor the district court have conducted the necessary substantive analysis of the severance deeds at issue in this case. Neither we nor the district court knows the number of deed variations or the materiality of the discrepant language. Without such evidence, the plaintiffs have failed to carry their burden of demonstrating commonality. By certifying the classes notwithstanding this failure, the district court abused its discretion by relaxing the plaintiffs’ burden of proof with respect to Rule 23’s commonality requirement. 15

VI.

The district court also certified the class claims relating to EQT’s and CNX’s al *364 leged underpayment of royalties. We again conclude that the district court abused its discretion when it certified these classes.

A.

Before turning to the merits of the district court’s certification decision, we first clarify the scope of our review. The defendants have asked us to exercise pendent appellate jurisdiction over an earlier ruling of the district court. Specifically, they ask that we consider the district court’s determination that Virginia courts would apply a doctrine called the “first marketable product” rule to determine whether the defendants have underpaid royalties.

Broadly speaking, the first marketable product rule holds that all oil and gas lessees have an implied duty of marketability. That is, lessees have an implied duty to bear the cost of putting the oil and gas in a marketable condition after it is removed from the well, including common postproduction expenses for gathering, compressing, and dehydrating oil and gas. See generally Byron C. Keeling & Karolyn King Gillespie, The First Marketable Product Doctrine: Just What is the Product, 37 St. Mary’s L.J. 1, 5 (2005) (summarizing the doctrine). A number of state courts have adopted variations of the doctrine to guide their interpretation of oil and gas leases. See, e.g., Rogers v. Wes-terman Farm Co., 29 P.3d 887, 902-03 (Colo.2001) (en banc); Gilmore v. Superior Oil Co., 192 Kan. 388, 388 P.2d 602, 606-07 (1964); Mittelstaedt v. Santa Fe Minerals, Inc., 954 P.2d 1203, 1205 (Okla.1998).

Many of the plaintiffs’ theories of royalty underpayment in this case depend, either explicitly or implicitly, on the existence of an implied duty of marketability. For example, according to some of the classes, the first marketable product rule renders illegitimate many of the deductions the defendants have taken from the plaintiffs’ royalty payments.

In an earlier ruling denying the defendants’ motion to dismiss, the district court held that Virginia courts would apply the first marketable product rule, and that the doctrine would guide its analysis of the royalty underpayment claims in this case. 16 On appeal, the defendants ask us to review that non-final judgment.

Under the doctrine of pendent appellate jurisdiction, “we retain the discretion to review issues that are not otherwise subject to immediate appeal when such issues are so interconnected with immediately appealable issues that they warrant concurrent review.” Rux v. Republic of Sudan, 461 F.3d 461, 475 (4th Cir.2006). We exercise jurisdiction under this exception sparingly, and only when: (1) “an issue is inextricably intertwined with a question that is the proper subject of an immediate appeal” or (2) “review of a ju-risdictionally insufficient issue is necessary to ensure meaningful review of an immediately appealable issue.” Id. (internal quotation marks omitted).

We decline to exercise such discretion here. The district court did not mention the implied duty of marketability in its certification decision, which suggests that the issue was not inextricably intertwined with its determination that the plaintiffs satisfied Rule 23’s requirements.

Additionally, we need not revisit the district court’s marketability ruling to decide the central issue on appeal: whether the *365 district court abused its discretion when it certified the claims of the five classes alleging underpayment of royalties. As we discuss in greater detail below, the classes do not satisfy Rule 23’s requirements even if we assume the first marketable product rule applies to their claims.

B.

We next turn to the substance of the district court’s decision to certify the classes asserting claims of royalty underpayment. The classes’ theories of underpayment vary, but there are some common threads. For example, all five classes allege that the defendants sold the CBM at too low a price, in part, by selling the gas to affiliates in non-arms-length transactions. Most of the classes also contend that EQT and CNX have taken improper or excessive deductions, for example, for common postproduction expenses. Based on these and other diverse theories, 17 the plaintiffs assert a host of property, tort, and breach of contract/unjust enrichment claims arising from the defendants’ purported underpayments.

The district court certified these classes as Rule 23(b)(3) class actions. See Adair, 2013 WL 5429882, at *38. 18 As noted above, a class certified under that provision must satisfy all of Rule 23(a)’s prerequisites and two additional requirements: predominance and superiority. See Fed. R.Civ.P. 23(b)(3).

As with the ownership classes, the primary issue on appeal for the underpayment claims is whether the plaintiffs have demonstrated common questions of law or fact. Because the district court certified these classes under Rule 23(b)(3), however, we consider that issue in conjunction with the court’s further conclusion that common questions also predominate. See Lienhart, 255 F.3d at 146 n. 4 (“In a class action brought under Rule 23(b)(3), the ‘commonality’ requirement of Rule 23(a)(2) is ‘subsumed under, or superseded by, the more stringent Rule 23(b)(3) requirement that questions common to the class predominate over’ other questions.” (quoting Am-chem Prods., 521 U.S. at 609, 117 S.Ct. 2231)); see also Comcast, 133 S.Ct. at 1432 (noting that “[t]he same analytic principles” governing the Rule 23(a) commonality analysis apply to Rule 23(b)(3), but the latter’s predominance requirement is “more demanding”).

For a variety of reasons, we conclude that the district court abused its discretion *366 when it certified the five classes under Rule 23(b)(3).

1.

We first review the aspects of the district court’s analysis that apply to all five royalty underpayment classes.

At bottom, the district court believed that both the commonality and predominance requirements of Rule 23 were satisfied by the same basic fact: the defendants employed numerous uniform practices related to the calculation and payment of CBM royalties. These common practices are not irrelevant to Rule 23(b)’s predominance requirement. But we hold that the district court abused its discretion by failing to consider the significance of this common conduct to the broader litigation.

The district court identified numerous common royalty payment practices. For example, it noted that EQT sells all of the CBM it produces in Virginia to an affiliate, EQT Energy, and that “all royalty owners within the same field have been paid royalties based on the same sales price for the CBM.” Adair, 2013 WL 5429882, at *38. With respect to CNX, it noted that CNX “has uniform policies and procedures which governed its calculation of CBM revenues,” and that “it has deducted severance and license taxes when calculating royalties since January 1, 2004.” Id. at *39.

That the defendants engaged in numerous common practices may be sufficient for commonality purposes. As noted above, the plaintiffs need only demonstrate one common question of sufficient importance to satisfy Rule 23(a)(2).

But the mere fact that the defendants engaged in uniform conduct is not, by itself, sufficient to satisfy Rule 23(b)(3)’s more demanding predominance requirement. The predominance inquiry focuses not only on the existence of common questions, but also on how those questions relate to the controversy at the heart of the litigation. See Amchem Prods., 521 U.S. at 623, 117 S.Ct. 2231 (noting that the predominance inquiry “trains on the legal or factual questions that qualify each class member’s case as a genuine controversy”). Even a plethora of identical practices will not satisfy the predominance requirement if the defendants’ common conduct has little bearing on the central issue in the litigation^-in this case, whether the defendants underpaid royalties. Absent such a relationship, there is no basis for concluding that individual issues will not predominate.

We believe the district court placed an inordinate emphasis on the sheer number of uniform practices without considering whether those practices are relevant to assessing the defendants’ ultimate liability. Some of the common practices that the district court identified — e.g., the fact that EQT sold all of its CBM into one of two interstate pipelines — have little relevance to the validity of the defendants’ royalty payment practices.

The district court did identify common practices that may be pertinent to the predominance inquiry — e.g., the fact that “EQT calculated all royalties based on the same methodology.” Adair, 2013 WL 5429882, at *38. But the district court’s analysis fell short because it never analyzed why those common practices were sufficient to ensure that the class members’ common issues would predominate over individual ones.

The defendants have highlighted a number of uncommon practices that might cause individual, issues to predominate. For example, EQT notes that it calculates royalties in different ways for different class members, depending on where the CBM is produced. Its method of calculat *367 ing royalties — and the deductions it applies — have also changed over time. CNX submitted evidence that it takes different deductions depending on where it sells the CBM, and that its deduction calculations sometimes vary between and even within wells during different time periods.

We do not decide today whether the disparate practices identified by the defendants are sufficient to defeat the predominance requirement. On remand, the district court may well conclude that the defendants’ common conduct is sufficient to ensure the predominance of common issues over individual ones. But it was an abuse of discretion for the district court to focus only on the number of common practices without considering the significance of the defendants’ disparate conduct in the broader litigation. 19

2.

We also remand for the district court to give greater consideration to Rule 23 factors that affect only certain classes. In particular, the district court should consider how variations in the defendants’ royalty obligations to the class members implicate the commonality and predominance inquiries in Kiser, Adkins, and Addison.

The defendants have relatively uniform royalty obligations with respect to the class members in the two force pooled cases — Adair and Hale. All plaintiffs in those classes are deemed lessors, which means that Board pooling orders dictate the terms of the defendants’ royalty obligations. Those terms are largely uniform among the class members. 20

The issue is more complicated in Kiser, Adkins, and Addison, because those class members all have voluntary lease arrangements with the defendants. As the district court recognized, “these leases vary as to the language as to the payment of royalties and post-production deductions.” Adair, 2013 WL 5429882, at *42. For example, while some leases require the defendants to calculate royalties based on the proceeds they receive from the sale of CBM, others require the defendants to use the market value of CBM. Some leases specify that the price for CBM must be determined at the well, while others permit calculation at the point of sale.

Although the district court recognized the problem of lease language variation, it did not see it as a barrier to class certification in any of these cases. In our view, *368 however, these variable terms will make it difficult, if not impossible, for a court to assess the validity of the defendants’ royalty payment practices on a classwide basis.

For example, the question of whether a gathering charge 21 is legitimate will produce different answers for class members whose leases specifically authorize that charge versus those whose leases specifically forbid it. Such dissimilarity will preclude the generation of a common answer to the plaintiffs’ common question. See, e.g., Wallace B. Roderick Revocable Living Trust v. XTO Energy, Inc., 725 F.3d 1213, 1218-1219 (10th Cir.2013) (concluding that the plaintiffs failed to demonstrate commonality when there was significant evidence of lease language variation); Chieftain Royalty Co. v. XTO Energy, Inc., 528 Fed.Appx. 938, 942-44 (10th Cir.2013) (remanding to allow the district court to examine whether lease language variations in a similar royalty underpayment case defeat commonality).

The plaintiffs argue that the first marketable product rule renders lease variation a moot point because that rule prohibits producers from deducting any postproduction costs. But even the plaintiffs concede that an express lease term — e.g., authorizing a particular post-production charge — supersedes any implied duty under the rule. Based on the sampling of deeds in the record, we know at least some of the class members’ leases expressly negate part or all of the implied duty. See, e.g., J.A. 2556-57 (requiring the lessor to pay a proportionate share of common postproduction charges, including the cost of gathering and dehydrating gas).

It was the plaintiffs’ burden to demonstrate commonality on the implied duty of marketability. See Thom v. Jeffersonr-Pi-lot Life Ins. Co., 445 F.3d 311, 321 (4th Cir.2006). Yet they have made no attempt to do so. Neither they nor the district court engaged in any substantive analysis of the lease terms to determine whether language variations destroy the possibility of resolving the common question(s) on a classwide basis. Assuming the first marketable product rule does apply, the plaintiffs have yet to demonstrate even the lesser requirement of commonality on the implied duty of marketability.

Contrary to the plaintiffs’ assertions, the district court’s class definitions do not solve this problem. In Kiser and Adkins, the court defined the classes to include only those gas owners whose leases are “silent” with respect to the deduction of costs. 22 According to the plaintiffs, this limitation obviates the need for them to review the leases individually because the class members’ “leases are the same with respect to the one issue that is material to their claims: they do not contain language allocating to the lessor the costs of making gas ... marketable.” Appellees’ Br. at 37.

But the “silence” requirement raises as many problems as it solves. The court never explained what it meant by “silent as to the deduction of costs” in either Kiser or Adkins. See Adkins, 2013 WL 5442378, at *1; Legard, 2013 WL 5429885. Would a lease requiring the lessor to pay “all excise, depletion, privilege and production taxes” 23 but not postproduction *369 charges qualify? See J.A. 1069. What about a lease that permits a lessee to use any gas produced from the premises “for fuel in its operations ... free of charge”? J.A. 1073. We agree with the defendants that disputes will inevitably arise regarding the meaning of “silence,” and the court will have to sort out these differences based on the particular lease language.

The issues are slightly different in the other voluntary lease case, Addison, because the class definition does not contain a “silence” requirement. The district court nonetheless concluded that Rule 23 was satisfied because it found that CNX— the defendant in that case — employs a standard gas lease. Thus, it assumed there would be no lease language variation that could affect the uniformity of CNX’s royalty obligations. See Adair, 2013 WL 5429882, at *39.

But the fact that CNX now uses a form lease for CBM royalties does not establish that all of the Addison class members’ leases are uniform. CNX has inherited a large number of leases from predecessor companies, many of which contain different royalty provisions. Compare J.A. 2556-57 (providing a gas royalty of “12.5% of the value of gas produced from the leased premises and sold on or off the leased premises ... less a proportionate part of the costs incurred by Lessee in heating, sweetening, gathering, transporting, dehydrating, compressing, exacting, processing, manufacturing, or any other post-production costs incurred by Lessee in making such gas or other substance merchantable”), with J.A. 4914-15 (providing a royalty of “the value of l/8th of the gas so sold or used,” where “value” means “the selling price stipulated in a bona fide contract entered into by Lessee as a result of an arms-length negotiation with a third party not a subsidiary, parent or affiliate of Lessee,” or, if the transaction is with an affiliate without the lessor’s permission, “on the basis of the current market value of the production so disposed of’). Perhaps the legality of CNX’s deduction practices can be assessed as to only those class members who signed its standard lease. But the class definition is not limited to those persons, and the plaintiffs have made no effort to explain how commonality might be established for the other Addison class members.

In short, the plaintiffs have failed to demonstrate that variations in lease language in Kiser, Adkins, and Addison do not defeat even the lesser requirements of Rule 23(a). On remand, after reviewing the leases in this case, the plaintiffs may be able to show that there are a limited number of lease forms, such that the validity of the defendants’ conduct can be assessed on a subclass basis. See, e.g., Foster v. Merit Energy Co., 282 F.R.D. 541, 556 & n. 12 (W.D.Okla.2012). The district court may also be able to craft more definite class definitions, thus eliminating or mitigating some of the problems described above. At this point, however, the plaintiffs have not yet carried their burden of demonstrating the classes’ compliance with all of Rule 23’s requirements.

3.

The plaintiffs in Adkins face additional complications, which arise from the defining characteristic of that class: all of the class members have received a royalty payment from EQT at some point in the past twenty years. This fact raises at least two issues that are likely to implicate the district court’s Rule 23 analysis.

First, at least with respect to the breach of contract claims, the court will likely need to consider course of performance evidence. See Video Zone, Inc. v. KF & F Props., L.C., 267 Va. 621, 594 S.E.2d 921, 924 (2004) (“Generally, the parties’ *370 interpretation and dealings with regard to contract terms are entitled to great weight and will be followed unless doing so would violate other legal principles”)- The record highlights the individualized nature of such evidence. See, e.g., J.A. 3855-98 (documenting one Adkins plaintiff’s individual communications with EQT regarding its royalty obligations under her lease). At a minimum, the need for individualized proof strongly affects the predominance analysis of Rule 23(b). Yet, as the defendants note, the district court failed to discuss course of performance evidence entirely. See Appellants’ Br. at 53-54.

Second, the district court should reevaluate the implications of the defendants’ statute of limitations defense for Rule 23’s predominance requirement. 24

Below, EQT moved to dismiss several of the plaintiffs’ claims on the grounds that they were time-barred by applicable statutes of limitations. In response, the plaintiffs argued that the limitations period should have been tolled because EQT issued misleading reports about the kinds of deductions it was taking from its royalty payments.

The district court “refused to grant EQT’s motion to dismiss ... based on its finding that the plaintiffs had alleged sufficient facts to plead fraudulent concealment by which EQT may be estopped from asserting th[e statute of limitations] defense.” Adair, 2013 WL 5429882, at *39. The court elaborated that “the doctrine of fraudulent concealment does not focus on the actions or knowledge of the plaintiffs, but on the actions of the defendant.” Id. Because the defendants’ representations to the plaintiffs regarding their royalty deductions were relatively uniform, the court concluded that the defendants’ common conduct was again sufficient to satisfy the commonality and predominance requirements. See id.

The district court misapplied the doctrine of fraudulent concealment. Although a defendant’s conduct is not irrelevant, attention must also be paid to the plaintiffs knowledge and actions. “A party seeking to invoke the doctrine of fraudulent concealment must demonstrate that ‘(1) the party pleading the statute of limitations fraudulently concealed facts that are the basis of plaintiffs claim, and (2) the plaintiff failed to discover those facts within the statutory period, despite (3) the exercise of due diligence.’ ” Detrick v. Pan-alpina, Inc., 108 F.3d 529, 541 (4th Cir. 1997) (quoting Supermarket of Marlinton, Inc. v. Meadow Gold Dairies, Inc., 71 F.3d 119, 122 (4th Cir.1995)). In this context, a plaintiffs knowledge typically requires individual evidence, Thom, 445 F.3d at 321, which will frequently defeat Rule 23’s requirements.

Here, the district court abused its discretion by failing to give any consideration to what proof the plaintiff-focused elements of the doctrine of fraudulent concealment might require, even if the court is ultimately correct that the statute of limitations is no bar to class certification. 25

4.

We conclude by briefly discussing Rule 23(b)(3)’s superiority requirement. Be *371 cause all of the royalty underpayment classes and claims- were certified under Rule 23(b)(3), the plaintiffs must be able to demonstrate that proceeding as a class “is superior to other available methods for fairly and efficiently adjudicating the controversy.” Fed.R.Civ.P. 23(b)(3).

The district court concluded that the royalty underpayment classes satisfied this requirement, focusing on the barriers to individual litigation that many CBM royalty claimants face. See Adair, 2013 WL 5429882, at *40. As the court noted, “many CBM royalty claimants own only a fractional interest in a 12.5 percent royalty,” a fact that, “no doubt, has resulted in the sparse number of individual cases filed to date over ... the calculation of royalties.” Id. Additionally, the court found that concerns of judicial economy supported a finding of superiority because a collective action would allow a court to resolve all of the royalty owners’ claims in a single forum and lessen the risk of inconsistent judgments against the defendants. See id. We agree with the district court that the factors it identified are relevant to the superiority analysis. Indeed, for many of these claimants, collective action may offer the only realistic opportunity to recover.

Nevertheless, the district court should give further thought to other factors that may bear on the superiority analysis. Without intending to limit the scope of the relevant inquiry, the court should consider how the dominance of state law issues may affect the suitability of this litigation in a federal forum, and what state-law mechanisms may be available to resolve the underpayment claims as an alternative to a class action.

We also think it proper for the district court to assess the extent of the defendants’ efforts to resolve and pay undisputed claims. A finding that the defendants have not acted in good faith toward that end may weigh strongly in favor of a finding of superiority of a class action.

Where the proper balance lies in the superiority analysis we leave to the district court on remand as part of its broader consideration of the other Rule 23(b)(3) factors.

VIL

We ultimately hold that the district court’s analysis lacked the requisite rigor to ensure the requirements of Rule 23 were satisfied by any of the certified classes. On remand, the district court may conclude that one or more subclasses should be certified. It may also find that class certification should be denied entirely. At this point, we only conclude that certification was premature.

We recognize that there are numerous CBM owners in Virginia who haven’t received a penny of CBM royalties and others who may have gotten less than their due. We are not unsympathetic to their plight.

But sympathy alone cannot justify certification under Rule 23. We therefore vacate the district court’s grant of the plaintiffs’ motions for class certification, and remand the case for further proceedings consistent with this opinion.

VACATED AND REMANDED