SUTTON, Circuit Judge.
In this collective action, sales employees for Wyndham allege that the company violated the Fair Labor Standards Act by failing to compensate them for overtime. After a bench trial, the district court found that the employees were similarly situated and had presented sufficient representative evidence to show that Wyndham violated the Act. The court concluded that the 156 employees worked an average of 52 hours per week and awarded about $ 5 million in damages. The parties each appealed. We affirm in part, reverse in part, and remand for further proceedings.
I.
Wyndham owns and operates resorts around the world. At issue in today's case are its four destinations in Tennessee: two in the Smoky Mountains; one in Nashville; and one in Fairfield Glade, between Knoxville and Nashville. As part of its business model, it sells ownership interests (timeshares) and non-ownership trial packages to customers.
Wyndham employs three types of sales employees at the four locations. Front-line sales employees sell ownership interests to people who do not own Wyndham timeshares. In-house sales employees sell upgraded ownership interests to existing Wyndham owners. And discovery sales employees sell non-ownership trials to prospects.
Wyndham pays its salespeople based mainly on commission. While they receive a minimum-wage draw based on the hours they record each week, Wyndham deducts that amount from their commissions. In 2009, Wyndham began paying overtime to its sales force.
Jesse and Michael Pierce filed this collective action under the Fair Labor Standards Act in 2013, alleging that Wyndham required sales employees to underreport their hours or altered the employees' timesheets to avoid paying overtime. The district court granted Wyndham's motion for summary judgment on some of the employees'
*745claims. And it certified the rest of the group-156 employees from all three positions and all four locations-as a collective action and allowed them to proceed to trial based on representative evidence.
After a 14-day bench trial, the district court found that the employees' evidence was representative of the testifying and non-testifying sales employees. It found that the evidence showed that Wyndham violated the Act by "prohibiting Sales [employees] from recording or recovering overtime, despite working overtime, and by instructing sales managers to edit timecards to misrepresent the time that Sales [employees] worked to achieve that result." R. 427 at 127. It found that, on average, each employee had worked 52 hours per week during the recovery period-October 21, 2010, to October 31, 2013. And it awarded the sales employees $ 2,512,962.91 in overtime pay and an equal amount in liquidated damages. The parties each appealed.
II.
Wyndham bats first. It argues that the district court erred by (1) certifying the collective action, (2) finding that the representative evidence established that Wyndham violated the Act, and (3) finding that each employee worked 52 hours per week.
Collective-action certification. Wyndham argues that the district court erred in allowing the case to proceed as a collective action, a decision we review for an abuse of discretion. O'Brien v. Ed Donnelly Enters. , 575 F.3d 567, 584 (6th Cir. 2009), abrogated on other grounds by Campbell-Ewald Co. v. Gomez , --- U.S. ----, 136 S.Ct. 663, 193 L.Ed.2d 571 (2016). A mistake of law, it bears recalling, amounts to an abuse of discretion. Koon v. United States , 518 U.S. 81, 100, 116 S.Ct. 2035, 135 L.Ed.2d 392 (1996).
The Fair Labor Standards Act requires employers to pay overtime to most employees who work more than 40 hours a week. 29 U.S.C. § 207(a)(1). Employees may bring a collective action to enforce the Act on "behalf of ... themselves and other employees similarly situated." Id. § 216(b). Like-situated employees may join the lawsuit if they consent in writing. Id. To determine whether plaintiffs are similarly situated, we consider (1) "the factual and employment settings of the individual[ ] plaintiffs," (2) "the different defenses to which the plaintiffs may be subject," and (3) "the degree of fairness and procedural impact of certifying the action as a collective action." O'Brien , 575 F.3d at 584 (quotation omitted). The heart of the matter is whether the plaintiffs should be permitted to bring their claims of liability and damages as a group based on representative, rather than personal, evidence. See 7B Charles Alan Wright et al., Federal Practice and Procedure § 1807 (3d ed. 2005). Employees bear the burden of satisfying this requirement. O'Brien , 575 F.3d at 584.
The district court treated the 156 members of the collective action as similarly situated because Wyndham executed a common policy requiring off-the-clock work and altering timesheets to avoid paying overtime. The employees, it reasoned, "had the same job duties, were paid utilizing the same compensation method, and kept track of hours using the same time system, despite their different job titles." R. 427 at 154.
The district court did not abuse its discretion in certifying the collective action as to the in-house and front-line salespeople. Yes, the job titles differ. But the roles require identical tasks, just aimed at different customers. Both groups of employees arrive at work at approximately the same time each morning for a mandatory *746meeting. They participate in "tours" throughout the day-meetings with guests or other customers to pitch Wyndham timeshares-trying to sell the same product (deeded ownership interests). Both participate in "nightline" (pitching guests as they check into the hotel) and in "party weekends" (treating guests to a stay at the resort in hopes they will buy ownership interests). Wyndham compensates the employees under the same plan and tracks their hours under the same system. Through it all, Wyndham enforced a common policy of not paying them overtime, even when they worked over 40 hours per week.
In that sense, this case parallels Monroe v. FTS USA, LLC , in which cable technicians in various locations sued their employer for its failure to pay them overtime. 860 F.3d 389, 402 (6th Cir. 2017). Monroe treated the cable technicians as similarly situated because they did the same job and alleged a single time-shaving policy, even though they worked different amounts of overtime. Id. at 404. In that light, we held that the employer could assert any individualized defenses against the testifying employees, which would distribute those individualized defenses across the claims of the entire collective. Id. at 404-05. And we said that the Act was meant to allow employees to consolidate related claims alleging a single, illegal policy. Id. at 405. So also here. No abuse of discretion thus occurred with respect to the in-house and front-line salespeople.
Wyndham does not like that conclusion. It disclaims any company-wide time-shaving policy, noting that the employees pointed to a number of ways in which the company inaccurately recorded their hours: employees not clocking in at all, clocking out early, clocking out between tours, or working from home, and Wyndham manufacturing changes to their timecards. But Monroe forecloses Wyndham's attempt to slice the policy into so many parts. As in Monroe , the employees' essential claim is that Wyndham required them to work off the clock and altered their recorded hours in an effort to avoid paying overtime. And as in Monroe , that amounts to a single policy. Id. at 403.
Even so, the company insists, its employees had different jobs at different locations with different titles and worked a wide range of hours. But there are no meaningful differences between the in-house and front-line salespeople or for that matter between the jobs they performed at the four Tennessee locations. Wyndham asserted individual defenses regarding the number of hours that testifying employees worked. The district court distributed those defenses across the collective, resulting in the district court's decision that the salespeople did not work as many hours as they claimed. The same thing happened in Monroe , in which the jury awarded fewer damages than the technicians sought and found that the testifying technicians worked anywhere from 8 to 24 hours of overtime a week. Id. at 405 ; id. at 424 (Sutton, J., concurring in part and dissenting in part).
All in all, the district court did not abuse its discretion in treating the in-house and frontline sales employees as similarly situated.
The same cannot be said for the district court's decision to allow the discovery employees to proceed collectively with the in-house and front-line employees. The discovery salespeople not only had a different title, but they also sold a different product from the one sold by their in-house and front-line counterparts. That means the sales presentations and closings might generally have lasted for a different length of time. Only one of the discovery employee-plaintiffs testified at trial, and he *747did not indicate that discovery employees participated in party weekends, that Wyndham required the discovery team to work six days a week, or that the discovery salespeople at the other Tennessee locations performed the same functions he did, all in contrast to the consistent testimony from the in-house and front-line employees.
In addition, the discovery team started later than the in-house and front-line teams because the discovery team did not attend the mandatory morning meeting. The trial testimony suggested that the discovery employees generally arrived anywhere from a half hour to two hours later than the other sales employees. It makes little sense to treat the discovery salespeople, who sold a different product and regularly started an hour or two after the other sales team members, the same as the in-house and front-line employees. The end-all and be-all of a collective action is to determine the average number of hours each individual worked per week. At the least, the court should have created a separate subclass for the discovery employees.
The plaintiffs try to discount this point on the ground that the district court noted that the discovery employees "often stayed later." R. 427 at 136. But the district court made no factual findings about how often the discovery team members stayed late or how much later they stayed. The off-the-clock work study, which showed the dates and times of in-house and front-line employees' transactions, does not show similar data for the discovery employees. No evidence thus shows how often or at what time discovery salespeople initiated or closed contracts.
The record indicates that one or more discovery workers stayed until all of the other sales employees finished with customers-in case a deal fell through and a discovery salesperson needed to pitch the customer a trial package. But presumably the in-house and front-line employees who stayed late were those workers who completed a sale because the closing process often lasted several hours. In those cases, the discovery workers would not have talked to those customers and would have been able to leave just as early as the in-house or front-line employees who made the sales. That leaves us with consistent testimony that the discovery employees started later than the other salespeople and scattered testimony that some of the discovery workers stayed as late or later on occasion. But the evidence needed to show that the discovery team members generally stayed an hour or two later than the other salespeople each night-to make up for later start times-in order for all of the sales employees to be similarly situated.
Invoking Monroe , the employees persist that all three groups of salespeople are similarly situated because Wyndham had a uniform policy against paying them overtime. See 860 F.3d at 402. But, in Monroe , the record showed that all of the employees "work[ed] in the same position, ha[d] the same job description, and perform[ed] the same job duties." Id. Not so here. The discovery team sold a different product and started later in the day. A common policy cannot overcome the factual differences between the discovery employees and the other salespeople (what they sold and when they started work), which goes to determining the heart of the claim (the total hours worked each week).
Representative evidence of liability. Wyndham also argues that the employees did not provide sufficient representative evidence to prove that Wyndham violated the Act. To meet their burden under the Act, the employees had to show that they "performed work for which [they were] improperly compensated."
*748Anderson v. Mt. Clemens Pottery Co. , 328 U.S. 680, 687, 66 S.Ct. 1187, 90 L.Ed. 1515 (1946). Representative evidence may establish liability for testifying and non-testifying employees, as similarly situated employees may "testify as representatives of one another." Monroe , 860 F.3d at 408 (quotation omitted). At stake is whether the evidence-"representative, direct, circumstantial, in-person, by deposition, or otherwise"-supports the district court's finding that Wyndham violated the Act by failing to pay overtime. Id. at 407 (quotation omitted).
It does under Monroe . Out of the 156 opt-in employees, the plaintiffs chose 47 as their sample. They drew the sample from two groups-employees who had worked for Wyndham more than six months and those who had worked for Wyndham fewer than six months. At trial, the plaintiffs presented testimony from 30 members of the collective-24 from the sample and 6 others whom Wyndham had independently chosen to depose. Including deposition designations and counter-designations, the district court received testimony from 44 of the 156 employees. Discounting for the discovery plaintiffs, who were not similarly situated and should not have been part of the collective, the court heard testimony from 43 of the 145 similarly situated salespeople (29.66%). That is a far higher percentage than Monroe approved, in which only 17 of the 293 technicians testified (5.8%). Id. at 394, 410-11.
As in Monroe , the testifying employees, in addition to other evidence, showed that Wyndham systematically executed a policy to avoid paying overtime. All of the testifying plaintiffs consistently said that Wyndham required them to underreport their time or altered their recorded time. They all provided an average of the number of hours they worked each week, ranging from 50 to 80 hours per week, and their basis for that number: the mandatory morning meeting, tours throughout the day, frequent late-night work and special events, and six- or seven-day work weeks. But, through it all, they didn't worry about keeping an accurate account of their hours because the company told them it would recoup any overtime pay from their commissions.
The administrative manager at the Nashville location testified that upper management instructed that sales employees could not be paid overtime and that managers should alter employees' timecards to show no more than 40 hours per week. The vice president of sales and marketing at the two Smoky Mountain locations acknowledged that Wyndham performed an audit that showed that salespeople worked off the clock. Several emails from managers also mentioned Wyndham's no-overtime-pay policy. The evidence thus showed that Wyndham executed an across-the-board time-shaving policy that failed to compensate the employees for the hours they worked.
Wyndham tries to counter this conclusion on several grounds. None is persuasive. First, it argues that the employees' sample was not reliable because they "hand-picked" witnesses. Appellant's Br. 19. But Wyndham's expert agreed that the employees randomly selected the original sample of 47 employees. While it is true that the plaintiffs called six employees to testify at trial who were not part of the original sample, Wyndham deposed all six. Wyndham also deposed every member of the sample and could have called anyone it wanted to rebut the testifying employees. It failed to do so, leaving us with no warrant to doubt the reliability of the witnesses. Monroe , 860 F.3d at 410-11.
Relatedly, Wyndham argues that the plaintiffs did not show that the testifying employees were representative of the *749non-testifying employees. It points to the varying ways in which employees testified that Wyndham implemented its no-overtime-pay policy and to the differences in hours employees said they worked and wages they earned. Wyndham argues that, without expert testimony to show that "each class member could have relied on that sample to establish liability if he or she had brought an individual action," Tyson Foods, Inc. v. Bouaphakeo , --- U.S. ----, 136 S.Ct. 1036, 1046, 194 L.Ed.2d 124 (2016), the district court erred in relying on the representative sample in this case. But Monroe rejected the same argument, explaining that " Tyson did not discuss expert statistical studies because they are the only way a plaintiff may prove" his claim by representative evidence under the Act. Monroe , 860 F.3d at 401. Instead, Monroe said that "the collective-action framework presumes that similarly situated employees are representative of each other and have the ability to proceed to trial collectively." Id. at 409.
Wyndham tries to distinguish Monroe on the ground that the employer in Monroe consented to using a representative sample at trial. Not true. The parties in Monroe agreed to limit discovery to a representative sample and to propose a plan for using that sample at trial. Id. at 410. After discovery, however, the employer in Monroe "did object to the use of representative proof at trial." Id. Monroe held that the employer's objection did not change the result. Just so here.
Wyndham argues that its expert's testimony shows that the district court erred in relying on the representative evidence, noting that neither side offered an expert in Monroe . But the court did not err in discrediting Wyndham's expert. In reaching his conclusion about the evidence in this case, the expert relied heavily on the sales employees' recorded hours. As the district court found, the employees' recorded hours could not properly serve as the basis for any conclusion because the evidence at trial showed their rampant inaccuracy.
Last of all, Wyndham argues that the use of representative testimony violated its due process rights because it could not present individualized defenses. But the district court permitted Wyndham to cross-examine every witness at trial and gave Wyndham the freedom "to call anyone whom [it] wanted to call to the stand." R. 427 at 149.
Damages. Wyndham argues that, under Mt. Clemens , the district court could not have found that the employees averaged 52 hours of work per week "as a matter of just and reasonable inference." 328 U.S. at 687, 66 S.Ct. 1187. We need not resolve the point because the district court based its 52-hour average on all 156 sales employees. Because the court erred in finding that the discovery employees were similarly situated to the other salespeople and that error infected its hourly average determination, we vacate the damages award and remand for the court to reassess damages for the in-house and front-line employees. Cf. Hatahley v. United States , 351 U.S. 173, 182, 76 S.Ct. 745, 100 L.Ed. 1065 (1956).
III.
The employees appeal on the grounds that the district court erred in (1) determining the average number of hours they worked and (2) granting Wyndham summary judgment on Melissa Evans' claims. As to the first point, we will let the district court reassess the damages award in the first instance. As to the second point, we disagree.
The district court granted Wyndham summary judgment on former employee *750Melissa Evans' claims, finding her judicially estopped from asserting them because she failed to notify the bankruptcy court of the claims. Evans maintains that the court erred as to those claims that arose after she filed her bankruptcy petition on July 12, 2013. Because those causes of action were not part of the bankruptcy estate, she disclaims any obligation to notify the bankruptcy court about the claims and asserts a right to pursue them here. But Evans did not make this argument below and has forfeited the right to make it here. Armstrong v. City of Melvindale , 432 F.3d 695, 700 (6th Cir. 2006).
Evans resists on the ground that she argued below that no part of her claims should have been dismissed. True, but she did so on the basis that she unintentionally failed to notify the bankruptcy court of all the claims. That is not today's argument. Today's argument is that she had no obligation to tell the bankruptcy court about the subset of claims that allegedly did not constitute property of the bankruptcy estate. Evans offers no legitimate explanation for failing to raise her current argument in the district court. We therefore affirm on this score.
For these reasons, we affirm in part, reverse in part, vacate the damages award, and remand to the district court for further proceedings consistent with this opinion.
HELENE N. WHITE, Circuit Judge, concurring in part and dissenting in part.
I agree with the majority that the Front-Line and In-House Sales Representatives were similarly situated, that the district court properly relied on plaintiffs' representative evidence, and that the district court did not err in granting Wyndham summary judgment on plaintiff Melissa Evans's claim. However, because the record shows that the Discovery Sales Representatives were similarly situated to the other representatives and maintenance of the collective action was proper, I would affirm the district court in all respects.
The majority largely resolves whether the Discovery representatives are similarly situated to the Front-Line and In-House representatives on the first O'Brien factor-the "factual and employment settings of the individual[ ] plaintiffs." O'Brien v. Ed Donnelly Enterprises, Inc. , 575 F.3d 567, 584 (6th Cir. 2009), abrogated on other grounds by Campbell-Ewald Co. v. Gomez , --- U.S. ----, 136 S.Ct. 663, 193 L.Ed.2d 571 (2016) (internal quotation marks omitted). The heart of the majority's analysis is that factual differences between the Discovery representatives and the two other positions render the Discovery representatives not similarly situated. The problems with this analysis are two-fold: it fails to appreciate the import of Wyndham's FLSA-violating practices across all positions, and the record does not support the factual differences perceived by the majority.
"Key" to the similarly situated analysis is "evidence of a company-wide policy of requiring [employees] to underreport hours that originated with the [employer's] executives." Monroe v. FTS USA, LLC , 860 F.3d 389, 402 (6th Cir. 2017) ; see O'Brien , 575 F.3d at 585 (holding that "the plaintiffs were similarly situated, because their claims were unified by common theories of defendants' statutory violations, even if the proofs of these theories are inevitably individualized and distinct"). There is ample evidence of such a policy here, and the Discovery representatives faced the same policies as the In-House and Front-Line representatives. The Discovery-representative plaintiff testified that he, like other representatives, was told by his manager not to show more than 40 hours on his time card, and "[the Discovery *751representatives] were working off the clock a lot." (R. 389, PID 11235.) Also like the Front-Line and In-House Sales Representatives, the Discovery representative recalled closing "quite a lot" of contracts while he was off the clock. (Id. at PID 11245.) In addition, there was managerial testimony that Wyndham's policy applied across all positions. For example, the administrative manager at the Nashville location testified that all sales representatives, regardless of the line they worked in, were told to clock in only when they were with a customer.
In response, the majority asserts that the common theories of a violation or broad policy cannot overcome the factual differences between the Discovery representatives and the other representatives. But the factual differences the majority identifies either are not supported by the evidence or are not nearly as great as the majority suggests. Notably, the majority overrides or disregards reasonable factual findings made by the district court in contravention of our deferential standard of review.
Although the majority (Maj. Op. at 746-47) posits that Discovery representatives "sold a different product" from the one sold by the others, the district court properly found that each of the three positions' "job was to sell" and "the only difference" was to whom they were selling. (R. 427, PID 16324.) The district court's conclusion was supported by the evidence at trial. A representative testified that the selling process was "basically the same": "You're meeting the customer. You're closing. Then you're going through closing. And then you are following up." (R. 375, PID 9652.) Another testified that all three positions used the "same process, same product" even if what the Discovery representative sold was "not permanent." (R. 386, PID 11212.).
Proceeding from the faulty belief that Discovery representatives sold a different product, the majority supposes that sales presentations and closings "might generally" last for a different length of time for Discovery representatives. (Maj. Op. at 746-47.) However, that supposition is cast into doubt by the Discovery representative's testimony that tours lasted "about two to three hours" and closing lasted "[t]ypically, about 45 minutes," estimates that are in-line with-and indeed sometimes higher than-the estimates of employees in other positions. (Compare R. 389, PID 11229, 11232 with R. 378, PID 10470 ("a tour would last anywhere from 30 minutes to an hour, two hours sometimes"); R. 383, PID 10522 (a tour last "[a]round two hours usually, from the beginning to the end"); R. 384, PID 11041 (tour took typically two to three hours); R. 377, PID 10231-32 (average time to close was about "an hour").).1
The majority improperly discounts the Discovery-representative testimony, pointing out that the testifying Discovery-representative plaintiff did not state that he participated in a selling event called a party weekend. Although that plaintiff did not testify whether he participated in party weekends, there is also no evidence in the record that Discovery representatives generally did not participate in party weekends. Moreover, even if that Discovery representative did not participate in party weekends, some In-House and Front-Line representatives also did not testify to participating in party weekends.
*752Indeed, there was evidence that not every In-House or Front-Line representative was invited to party weekends (or nightlines or dinners) and that the invitation depended on a representative's sales. (See e.g. , R. 401, PID 12680-81; R. 383, PID 10528; R. 395, PID 11983-84.) Most importantly, even if all Discovery representatives did not participate in party weekends, not every plaintiff needs to testify to the same means and methods of violating the FLSA. See Monroe , 860 F.3d at 403-04.
The majority also claims that the testifying Discovery-representative plaintiff did not testify that he had to work "six-ones"-or six days of work each week. But, consistent with the district court's finding that "Plaintiffs often worked five to six days a week" (R. 427, PID 16316), the Discovery representative testified that he worked "at least five, sometimes more" days a week and that he would work on his days off. (R. 389, PID 11230.)
Finally, the majority relies on Discovery representatives' later start time to find them not similarly situated. However, the Discovery representatives' later start time does not mean they worked fewer hours and were not similarly situated to the other representatives. The district court found that "while [Discovery representatives] started later, they often stayed later." (R. 427, PID 16316.) This finding is supported by evidence in the record. The evidence shows that despite having a slightly different schedule, Discovery representatives did not work materially fewer hours. The Discovery representative testified that he arrived "around eight o'clock" every morning to attend a required sales meeting with the manager, the same time as other representatives. (R. 389, PID 11231.) He also estimated that he worked "at least 55 hours per week" and that he also worked from home about "six hours or so" per week. (Id. at PID 11236-37.) Those hours are similar to testifying plaintiffs in the other positions.
The Discovery representative's testimony on his work schedule is supported by other evidence in the record. For example, when asked whether he "often" worked late into the night, an In-House representative responded, "Yes .... So most of the time when I was there - - and, again, most of the reps, we all worked the same, frontline, in-house, discovery, working somewhere between 5:30 and 8:30 every night." (R. 375, PID 9669.) Another testified that Discovery representatives had to stay later because if Front-Line "didn't sell them and [ ] go to closing, then [the Discovery representatives] would be there after us." (R. 386, PID 11213.) And still another testified that Discovery representatives had to stay not only until all the tours were completed, but also had to stay if a Front-Line representative was closing in case the "deal blew out" and "[the customer] changed their mind." (R. 392, PID 11683.) This last representative further testified that Discovery representatives would "come in 30 minutes to an hour after we did" but "they were always the last ones to be able to leave the building." (R. 393, PID 11748-49, 11831-32.)
The majority disregards evidence that the Discovery representatives stayed through the closing process for the other positions, reasoning that if the sale was successfully completed, the Discovery representative would leave at the same time as the other representative. This analysis ignores testimony that the number of In-House and Front-Line representatives far exceeded the number of Discovery representatives. As a result, the Discovery representatives would be working later than the vast majority of the other representatives. For example, at the Nashville location, the Front-Line and In-House "average[d] 50 people, per *753line" while the Discovery "probably had five or ten people."2 (R. 378, PID 10316.) As a result, one would expect that at the end of the day those five or ten discovery representatives would stay as late as (if not later than) five or ten Front-Line or In-House representatives. That means that, either because their last pitch was unsuccessful or the sale was completed, the other Front-Line or In-House representatives would leave earlier than the Discovery representatives. Thus, even in the majority's scenario, the Discovery representatives were on a daily basis staying later than a large number-likely a vast majority-of the Front-Line and In-House representatives.
In light of this record, the district court did not abuse its discretion in including the Discovery Sales Representatives in the collective action. Contrary to the majority's conclusion, there was sufficient evidence in the record that Discovery Sales Representatives were similarly situated. The Discovery representatives had the same responsibilities, sold the same product, and worked a similar amount of hours as Front-Line and In-House representatives. The plaintiffs, including Discovery representatives, were subject to the same FLSA-violating policy, and their claims are unified by "common theories" of FLSA violations such as Wyndham's use of a company-wide time-shaving policy. Accordingly, I would hold that the Discovery representatives were properly included in the collective action, and affirm the district court in full.