Before DENNIS, HIGGINSON, and WILLETT, Circuit Judges. DON R. WILLETT, Circuit Judge: Seventy-five years ago in Anderson v. Mt. Clemens Pottery Company, the Supreme Court fashioned a burden-shifting framework for federal wage claims where an employer fails to maintain proper records.[1] Under Mt. Clemens, if “the employer’s records are inaccurate or inadequate,” a plaintiff need only show by “just and reasonable inference” that she was an employee, worked the hours, and wasn’t paid.[2] It’s a lenient standard rooted in the view that an employer shouldn’t benefit from its failure to keep required payroll records, thereby making the best evidence of damages unavailable. In this unpaid-overtime case, the district court applied Mt. Clemens because Five Star’s bare-bones timesheets left numerous evidentiary gaps. The Department of Labor filled those gaps with consistent testimony that Five Star urged employees not to record their pre-and post-shift work hours. DOL used this testimony to estimate unpaid hours and calculate back wages. Five Star’s only rebuttal evidence was a summary chart based on the company president’s memory. As this chart failed to negate any raised inferences of unpaid work, we affirm the district court’s judgment. I Five Star Automatic Fire Protection, LLC is a fire-sprinkler installation and service company based in El Paso. Luis Palacios and his wife, Veronica, run the company as President and Vice President, respectively. Five Star has five separate departments—this lawsuit implicates only the construction department. During the relevant timeframe, Five Star had 53 construction employees. Construction employees typically work in two-man crews with one foreman (sprinkler fitter) and one helper (laborer). Most of the time, the crews work at client jobsites, not at Five Star’s facility where pipe is cut and welded (the “shop”). But occasionally, the construction employees work in the shop or at Palacio’s personal ranch. Most of the jobsites are close to Five Star’s shop, but others are up to an hour away. Several jobsites are out of state and require crews to stay out of town during the workweek. During typical day shifts at jobsites, construction employees work from 7 am to 3:30 pm.[3] The crews must first report to the shop and load the materials needed for the workday. The crews then drive a company truck to the jobsite. When the day’s work is completed, the crew drives back to the shop to drop off the company vehicle. The foreman usually drives the truck to and from the jobsite. Five Star pays its construction employees by the hour. Employees must record their own time, by handwriting on the company timesheets how many hours they worked each day. Employees only include the total number of hours worked at a jobsite, the shop, or the ranch. So when an employee has worked at two or more locations in one day, he does not record his start and stop time for each location nor does he indicate the order in which he worked at those places. In September 2015, DOL’s Wage-and-Hour Investigator Sandra Alba initiated an inquiry into Five Star’s compensation practices. Alba interviewed nine employees as well as Mr. and Mrs. Palacios. And she analyzed all timesheets spanning the two-year investigative period, except for two weeks for which time records were missing. Alba presented her findings to Mr. and Mrs. Palacios. She told them that construction employees were working, without compensation, before and after their recorded shifts. Alba told Mr. and Mrs. Palacios that they owed back wages for this uncompensated time. Mr. Palacios disagreed, stating that employees needed to record their hours, and if they were working before and after the regular shift hours, they should have recorded that time. He declined to pay the back wages or consider Alba’s calculations. DOL then filed a complaint against Five Star in federal court, alleging overtime and recordkeeping violations of the FLSA and seeking back wages and liquidated damages for the affected employees. The case was tried by consent before a magistrate judge.[4] DOL called six former employees to testify. The district court first made preliminary factual findings about Five Star’s liability, without calculating damages. After recounting the evidence presented at trial, the court found that Five Star failed to keep accurate records of off-the-clock time for the investigative period. The court then found that while the typical construction shift was 7 am to 3:30 pm, Five Star required employees to arrive at the shop no later than 6:45 am and didn’t compensate its employees for the 15-minute gap. The court further found that, while the typical workday ended at 3:30 pm, that was the time employees left the jobsite. And Five Star didn’t compensate employees for the required travel time back to the shop. Finally, the court found that Five Star had some face-of-the-record violations concerning errors on the payroll records; the parties do not dispute this finding. Following these preliminary conclusions on liability, the court granted the parties’ request to submit additional briefing on damages. In its final order, the court adopted the preliminary findings concerning liability and proceeded to evaluate damages. The court agreed with DOL’s calculations and held that Five Star was liable to 53 construction employees for $121,687.37 in back wages, $121,687.37 in liquidated damages, and $2,604.35 for face-of-the-record violations. Five Star appeals the court’s findings as to liability for the 47 non-testifying employees and the back-wages calculation for all 53 employees. II After a bench trial, we review findings of fact for clear error and legal conclusions de novo.[5] The calculation of unpaid overtime is a mixed question of law and fact—the number of overtime hours is a finding of fact, but the methodology used to calculate back wages based on that number is a question of law.[6] “When reviewing mixed questions of law and fact, this court reverses only if the findings are based on a clearly erroneous view of the facts or a misunderstanding of the law.”[7] III Five Star argues that the district court erred in relying on the testimony of six former employees to (1) find Five Star liable to 53 employees and (2) calculate the damages resulting from that liability. The court permitted this representative evidence under the Mt. Clemens burden-shifting framework. In Mt. Clemens, the Supreme Court noted that, typically, a plaintiff who brings an unpaid-wages claim under the FLSA “has the burden of proving that he performed work for which he was not properly compensated.”[8] But “where the employer’s records are inaccurate or inadequate and the employee cannot offer convincing substitutes,” an employee can attempt to fill the evidentiary gap.[9] “[A]n employee has carried out his burden if he proves that he has in fact performed work for which he was improperly compensated and if he produces sufficient evidence to show the amount and extent of that work as a matter of just and reasonable inference.”[10] The burden then “shifts to the employer to come forward with evidence of the precise amount of work performed or with evidence to negative the reasonableness of the inference.”[11] When an action involves a group of employees, a “representative sample,” if reliable, can shift the burden to the employer.[12] The representative proof is reliable “if the sample could have sustained a reasonable jury finding . . . in each employee’s individual action.”[13] If the employer fails to negate the inferences raised by the representative evidence, “the court may then award damages to the employee[s], even though the result be only approximate.”[14] As a preliminary matter, Five Star argues that its records were adequate because nobody, including DOL, has explained what adequate records should look like. Five Star misses the point. The adequacy of the records has to do with the evidence available to establish liability and damages, not the employer’s failure to conform to a certain recordkeeping standard. As the Court noted in Mt. Clemens, “[w]hen the employer has kept proper and accurate records,” then “the employee may easily” satisfy his burden to show he worked overtime without overtime compensation.[15] But where, as here, the records do not allow employees to show the uncompensated overtime work they completed, the burden-shifting framework applies. Five Star next argues that, even if Mt. Clemens applies, the district court erred because the representative evidence offered at trial failed to raise “just and reasonable” inferences of liability. Alternatively, Five Star argues that if DOL did raise such inferences, Five Star negated them. Finally, Five Star contends that the damages calculation failed to account for the variances among employees’ schedules and work assignments. We first address Five Star’s liability then turn to the damages calculations. A To raise just and reasonable inferences as to Five Star’s liability, DOL called six former employees (representing both foremen and helpers) at trial. Those employees consistently testified that: Jorge Cobian, Five Star’s lead supervisor, required them (at the risk of discipline) to report to Five Star as early as 6:30 am and no later than 6:45 am, even though the official shift (and compensation clock) began at 7 am. Before 7 am, the employees engaged in compensable activ ities, such as loading material onto company trucks. Employees didn’t leave the jobsites until 3:30 pm, and the amount of time to drive the company truck back to Five Star’s headquarters (to return the truck) varied depending on the location of the jobsite. The average drive time was 30 minutes. Cobian either instructed employees not to record time before 7 am and after 3:30 pm or told them that, if they did record the time, Five Star wouldn’t compensate them for it. DOL acknowledges that employees performed work at different jobsites but argues that all employees “typically started and ended their workday at Five Star’s premises and witnessed one another performing uncompensated work.” Five Star offers three main arguments to undermine or negate these inferences. None is persuasive. First, Five Star argues that the former employees’ testimony was unreliable. Five Star points to inconsistent statements regarding whether Cobian (or anyone at the company) actually told employees they couldn’t record, or wouldn’t receive compensation for, time before 7 am and after 3:30 pm. For example, one former employee testified that no one instructed him to write down his time before 7 am, although he never asked about it. Another stated that he just thought he would only be paid from 7 am to 3:30 pm. Others claimed that Cobian specifically told them that they would only be paid for eight hours per day. Despite these slight variations, all of this testimony supports the inference that the employees believed they could not, or should not, record their pre-and post-shift time, and that the company failed to compensate for this time. Relatedly, Five Star argues that the employee testimony varied when it came to what loading work employees did before 7 am. Five Star notes that it has two types of crews—underground and overhead. For the underground crews, a third party delivers most materials directly to the jobsite. On the other hand, the overhead crews have to load their own materials for each workday before heading to the jobsite. But as DOL points out, most employees work on overhead crews, and those who worked on underground crews still had to load some materials for most of their jobsites. Second, Five Star contends that the testifying employees lacked “personal knowledge” of the work performed by those who didn’t testify.[16] Five Star claims that because some crew members worked out of town or performed different activities during the day, the testifying employees couldn’t know what the non-testifying employees were doing. But the employees who testified stated that they personally saw other employees completing similar pre-and post-shift work. Finally, Five Star offers a string of arguments concerning its general efforts to correct timesheet errors and its openness to addressing employee concerns.[17] But these general efforts do not undermine the specific testimony that employees worked, per company instruction, before and after their recorded hours. Our decision in Brennan v. General Motors Acceptance Corporation confirms the district court’s liability determination.[18] In Brennan, employees had three different job titles, all of which involved collecting on overdue accounts and repossessing vehicles.[19] The employees had long, irregular hours so the employer depended on the employees to report their own time on company timesheets.[20] Even though upper management encouraged employees to record their overtime accurately, the employees’ immediate supervisor pressured them not to report overtime hours.[21] Fifteen of the company’s twenty-six employees testified, and the district court found that the company violated the FLSA as to all twenty-six employees.[22] The company argued on appeal that it was unaware that employees were not recording overtime hours.[23] We rejected that argument, holding that the record showed that the supervisor had actual knowledge or, at a minimum, constructive knowledge that the employees were working, but not reporting, overtime hours.[24] We further stated that “[t]he company cannot disclaim knowledge when certain segments of its management squelched truthful responses.”[25] Thus, based on the representative testimony of a de facto policy of underreporting time, we affirmed the district court’s finding that the employer violated the FLSA’s overtime requirements.[26] So too here. All testifying employees stated that their lead supervisor, Cobian, either said or implied that they shouldn’t record pre-and post-shift time. So even though Five Star’s manual instructed employees to record all of their time, the record shows that the de facto policy was that they shouldn’t. Although the sample size here was arguably small (6 of 53 employees—11% of the relevant employees), Five Star points to no authority saying 11% is insufficient for extrapolation purposes. And more importantly, Five Star failed to negate the inferences raised by the 11% of employees who testified. We find no error in the magistrate judge’s holding that Five Star is liable for unpaid overtime for all 53 construction employees. B Five Star also disputes the district court’s damages award of $121,687.37 in back wages. All of Five Star’s arguments concern the variations in the employees’ schedules and alleged overgeneralizations by Alba, the DOL investigator. This is how Alba arrived at the number the district court adopted: Alba made initial calculations based on employee interviews, as Five Star didn’t provide her with any time sheets until almost two years into the investigation. Once she had the timesheets, Alba reviewed all of them for the two-year investigative period.[27] Because the time records were incomplete, Alba relied on her employee interviews and the testimony at trial to calculate the amount of unpaid time employees worked. During regular day shifts, employees had to arrive sometime between 6:30 am and 6:45 am to get ready for the day’s work. Alba took a “conservative approach” and estimated that, on average, all construction employees worked for 15 uncompensated minutes before their shifts officially started. For post-shift work, which only applied to foremen who had to drive the company truck back to Five Star at 3:30 pm, Alba calculated an average of 30 minutes per day. Alba explained that this was also a conservative estimate because some employees told her that the post-shift drive time could take up to an hour. So in total, she added 15 minutes a day for laborers and 45 minutes a day for foremen. Alba added these averages to each employee’s weekly timesheets. Following the FLSA’s overtime requirement, Alba calculated damages for the weeks when employees exceeded 40 hours before or after she added the average pre-and post-shift time.[28] Alba didn’t calculate damages for four weeks of the year to account for vacations and holidays. Alba made other adjustments. If the timesheets showed that an employee worked only at the ranch or the shop for the day, Alba did not add uncompensated time since there would be no pre-7 am loading or post-3:30 pm driving. But when the timesheets showed that the employee worked at the ranch or shop for only part of the day, Alba added the pre-and post-shift averages because it was impossible to tell from the timesheets whether the employee started or ended the day at the ranch, the shop, or the jobsite. Overall, Alba’s final calculations were higher than what she initially estimated, but she presented the lesser amount in an effort to settle the case. After excluding four employees who were owed less than $20 for the entire timeframe, Alba offered the amount that the district court adopted: $120,417.62. Five Star contends that these calculations failed to account for variations in the employees’ schedules. For example, Five Star states that when employees were working the night shift, at the ranch, or in the shop, they wouldn’t have the pre-work loading time and post-work driving time. The employees that testified at trial said they spent anywhere from 2.5% to 30% of their time on the night shift. Five Star also argues that Alba didn’t account for all of the employees’ vacation time, as crews had at least one full day off for six different weeks in the year, which doesn’t include days off for personal reasons. To substantiate these schedule variations, Five Star provided the district court with a summary chart showing, among other things, which employees worked night shifts, out of town, or at the shop or ranch. Mr. Palacios created the chart based off his memory of different work projects. The district court found this chart unreliable because “Five Star’s timesheets simply do not allow for the retrospective analysis its president proffers.” We agree. In short, Five Star mainly contests that the damages award was an approximated number. But that’s what Mt. Clemens allows when, as here, FLSA-required time records are incomplete. IV Five Star fails to show that the district court committed any error concerning its finding of FLSA liability or calculation of damages. We thus AFFIRM.
