PATRICK E. HIGGINBOTHAM, Circuit Judge:

This is an interlocutory appeal from the district court's grant of class certification in a case involving allegations that the defendant title insurance company charged premiums for title policies that exceeded the refinance rates set by the Texas Department of Insurance in TEXAS INSURANCE CODE Rate Rule R-8. For the reasons that follow, we REVERSE the district court's grant of class certification and REMAND for further proceedings.

I.

Under TEXAS INSURANCE CODE Rate Rule R-8 ("R-8"), a title insurance company must provide a discount on the premium for a mortgagee title policy for a refinanced mortgage that was originally insured by a prior lender's policy if the new policy is issued within seven years of the date of the initial policy.1 R-8 provides:

R-8 provides this discount on a sliding scale, with the highest discount (40%) offered for refinances within two years of issuance of the original lender's policy and the lowest discount (15%) for refinances more than six years but less than seven years from the date of the policy insuring the prior mortgage.3

As this court has noted, "there is often no definitive way for a title insurer to determine, based on the documents available to it, whether or not a prior mortgage was covered by title insurance such that the new ... policy would qualify for the reissue discount."4 Title insurance companies thus have developed ad hoc policies which apply the discount when the borrower's file contains certain circumstantial evidence that the prior mortgage was insured.5 "[R]egardless of the circumstantial evidence in the borrower's file, the discount is mandatory for all borrowers who qualify."6

Gary and Mirvat Ahmad ("the Ahmads"), the named plaintiffs in this putative class action, filed a complaint alleging that Old Republic commonly fails to grant the discount to mortgagees who are entitled to it in violation of the federal Real Estate Settlement Procedures Act ("RESPA") and state common law. The defendant, Old Republic National Title Insurance Co. ("Old Republic"), moved for partial summary judgment, and the district court granted summary judgment for Old Republic on the Ahmads' RESPA claim.

Before the district court issued is order granting partial summary judgment, the Ahmads moved for Rule 23(b)(3) class certification, proposing the following class definition:

The Ahmads argued that class certification was appropriate because the transactions at issue "involve standard, form documents" and "[c]lass members and their damages can be identified and determined using objective criteria based on Defendant's business records." They maintained that such evidence could be summarized easily because it is an "established practice" in Texas "to assume that the refinanced mortgage was insured by title policy (and give the R-8 credit) if `(1) it has a GF number,7 (2) it is returned to a title company or (3) it is [a] first lien in favor of an institutional lender.'" According to the Ahmads, each of these three "proxy indicators" is sufficient to demonstrate that a policy was issued insuring the prior mortgage and to show that the borrower is entitled to the R-8 credit.

In response, Old Republic denied that any of the three proxy indicators is "legally sufficient" to establish a borrower's entitlement to the R-8 credit. It also disputed the Ahmads' contention that evidence necessary to show liability and damages could be easily summarized, noting that the guaranty files are not stored centrally by Old Republic but rather by numerous individual agents throughout the state; that much of the information is not available electronically; that where it is available electronically, it is not easily compiled because different agents use different programs for electronic filing; and that in most, if not all cases, the guaranty file for the refinance transaction will not contain a copy of the prior mortgage policy.

The district court granted the Ahmads' motion for class certification with regard to their remaining state law claims for unjust enrichment, money had and received, and breach of implied contract, which were brought pursuant to the court's diversity jurisdiction. The court narrowed the Ahmads' proposed class definition so that it included only individuals whose files contained one of the three proxy indicators of a prior title policy. The Ahmads had originally proposed eleven questions of fact or law common to the class:

From the Ahmads' eleven proposed common questions, the district court identified two questions of law (5 and 10) and two questions of fact (2 and 3) common to the class and not resolved by its ruling on partial summary judgment,8 and it concluded that common questions would predominate over individual issues.9

The same day the district court entered its order, this court issued Benavides v. Chicago Title Ins. Co.,10 affirming denial of class certification on similar facts. The district court in Benavides had concluded that the plaintiffs could not satisfy the predominance requirement of Rule 23(b)(3) because none of the central issues in the case constituted a "common question" of fact or law.11 This court affirmed, holding that the district court did not abuse its discretion when it found that there were no common questions capable of class-wide determination using class-wide proof.12 Arguing that the holding in Benavides controls the class certification analysis in this case, Old Republic filed a motion for reconsideration, which the district court denied. We granted permission for Old Republic to file an interlocutory appeal.13

II.

We review a district court's class certification decision "for abuse of discretion in `recognition of the essentially factual basis of the certification inquiry and of the district court's inherent power to manage and control pending litigation.'"14 However, whether the district court applied the correct legal standard in reaching its decision on class certification is a legal question that we review de novo.15 If the "district court premises its legal analysis on an erroneous understanding of governing law, it has abused its discretion."16

III.

For a class to be certified, plaintiffs must meet the four requirements of Rule 23(a) of the Federal Rules of Civil Procedure — numerosity, commonality of issues, typicality of the class representatives' claims in relation to the class, and adequacy of the class representatives and their counsel to represent the class.17 Plaintiffs also must satisfy at least one of the requirements of Rule 23(b).18 The plaintiffs in this case relied on Rule 23(b)(3), which requires that questions of law or fact common to the class predominate over questions affecting only individual members, and that a class action is superior to other available methods for the fair and efficient adjudication of the controversy.19

To satisfy Rule 23(a)(2)'s commonality requirement, plaintiffs must show that "[t]heir claims must depend upon a common contention," and the common contention "must be of such a nature that it is capable of classwide resolution — which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke."20 While the existence of even one common question of this nature satisfies Rule 23(a)(2),21 the Rule 23(b)(3) predominance inquiry "is more demanding,"22 testing "`whether proposed classes are sufficiently cohesive to warrant adjudication by representation.'"23 A court must consider how the case will be tried on the merits if the class is certified, "identifying the substantive issues that will control the outcome, assessing which issues will predominate, and then determining whether the issues are common to the class."24 Plaintiffs cannot satisfy Rule 23(b)(3)'s predominance requirement without satisfying Rule 23(a)'s commonality requirement.25

Old Republic challenges the district court's finding that the plaintiffs satisfied Rule 23(b)(3)'s predominance requirement. We agree with Old Republic that this finding that common questions would predominate misreads the demands of Rule 23.

A.

Our holding in Benavides controls the predominance analysis here. Benavides presented the same constellation of federal and state law claims as this case, with seven questions purportedly common to the putative class. Three of those questions pertained only to Benavides's RESPA claim, which the district court, as here, denied before deciding the class certification issue.26 The remaining four allegedly common questions in Benavides were:

The Benavides district court found that while the questions were "common in the colloquial sense,"28 none could be "definitively answered for all class members using a generalized set of facts and producing one unified conclusion."29 The resulting trial would require the factfinder to determine whether each individual qualified for the discount based on the evidence in his or her file — "[i]n some cases the answer [would] be `yes' and in others `no.'"30 On appeal, this court affirmed, approving the district court's denial of class certification because "[t]he only issues to be determined are ... individualized inquiries as to whether particular persons qualify for the discount and were denied it."31

B.

Because this court's affirmance in Benavides was based on the absence of any common question of fact or law,32 we turn to the question of whether the predominant issues here differ from those in Benavides and, if so, whether they are capable of class-wide determination. Old Republic argues that, as in Benavides, the question of whether a plaintiff actually qualified for the R-8 discount will require at trial a series of individualized inquiries rather than a single, class-wide determination.33 Indeed, two of the questions the district court identified as "common" — questions 2 and 10 — are virtually identical to questions that the Benavides district court found could not be determined on a class-wide basis using class-wide proof. We agree with Old Republic that there is no meaningful distinction between the class certification issues in this case and those in Benavides and that individual questions will predominate at trial.

The district court suggested that Benavides could be distinguished in two ways. First, the district court found that "whether extensive file-by-file review w[ould] be necessary" to answer question 2 — whether a plaintiff qualified for the discount — depended upon the resolution of question 3 — "what evidence is sufficient to demonstrate R-8 eligibility." The district court reasoned that question 3 is a common question of fact and that question 3 was not at issue in Benavides because there was no dispute in that case about what evidence was sufficient to establish entitlement to the discount. This reasoning is seductive but flawed.

In concluding that question 3 distinguishes this case from Benavides, the district court asked more work than it could perform of Chicago Title's concession in Benavides that the discount was mandatory for all who qualified. As Old Republic does here, Chicago Title conceded that when a borrower meets the requirements of R-8, the discount is mandatory.34 Chicago Title also acknowledged that it had an ad hoc practice of applying a presumption of prior title policy on the basis of any one of the three proxy indicators.35 But Chicago Title did not concede that any individual plaintiff or class of plaintiffs was "eligib[le]" for the R-8 discount. The Benavides district court rejected Benavides's proposal that the court "examine Chicago Title's internal policy that grants credits even in questionable cases, and rely on it (or something close) as the standard for imposing legal liability."36

The district court here also mistakenly characterized question 3 as a "common question" of fact. Question 3 is not a type of question that could satisfy Rule 23(a)'s commonality requirement, as it does not invite a "yes" or "no" answer,37 or present a contention whose "truth or falsity" can be established.38 During a trial on the merits, the fact-finder will not be tasked with answering the open-ended question, "What evidence is sufficient to qualify a borrower for the R-8 credit? Equally important, the answer to question 3 will not "resolve an issue that is central to the validity of each one of the claims in one stroke."39 Even if the proxy indicators constitute evidence from which a reasonable jury could conclude that a borrower's original mortgage was insured, the jury will have to engage in file-by-file review to determine whether individual plaintiffs had original mortgages covered by title insurance and met the other R-8 criteria.40

Second, the district court reasoned that it had "found four common questions where [the Benavides district] court found none." This justification is question-begging. The district court found that questions 2, 3, 5 and 10 were common to the class. For reasons just explained, question 3 is not a common question. And in Benavides, this court affirmed that issues virtually identical to questions 2 and 10 were not capable of class-wide determination by class-wide proof. As Old Republic notes, question 5 — "Whether Defendant or its agents could `earn' or lawfully keep the amount of the R-8 credit not given to eligible borrowers" — presupposes that "the R-8 credit not given" was required to be given. Question 5 thus is dependent on the resolution of question 2. In other words, question 5 presupposes the resolution of a question that cannot be answered on a class-wide basis with class-wide proof. Therefore, like questions 2, 3, and 10, question 5 is not a "common question."

Because none of the four questions identified by the district court is actually common to the class, common questions will not predominate at trial, and the Ahmads were not entitled to class certification under Rule 23(b)(3). Our holding in Mims v. Stewart Title Guaranty Co.41 is not to the contrary.42

IV.

The district court abused its discretion in finding that the requirements of Rule 23(b)(3) were satisfied. We REVERSE the district court's grant of class certification and REMAND the case for further proceedings consistent with this opinion.