In this collective action, a group of 2,519 EMTs and paramedics allege that their employer, the City of New York, willfully violated the Fair Labor Standards Act by requiring them to perform work before and after their shifts without paying them for that work unless the plaintiffs specifically requested overtime compensation from the City. A jury agreed following a twelve-day trial, and the U.S. District Court for the Southern District of New York (Broderick, J.) entered a $17.78 million judgment against the City. The City now appeals, raising four arguments: (1) the jury’s liability verdict cannot stand because plaintiffs failed to request overtime pay for the work at issue; (2) the jury’s willfulness finding was not supported by the evidence; (3) due to an erroneous instruction, the jury failed to make a necessary factual finding regarding the calculation of damages; and (4) the district court incorrectly forbade the jury from considering whether one component of the plaintiffs’ post-shift work was de minimis and therefore noncompensable. The City accordingly asks that we reverse the jury’s verdict or remand for a new trial on damages. We decline to do so and instead AFFIRM in toto. DENNIS JACOBS, C.J. In this collective action, a group of EMTs and paramedics won a multimillion-dollar verdict against their employer, the City of New York, for unpaid overtime wages. The 2,519 plaintiffs alleged that the City required the plaintiffs to perform work tasks before and after their shifts but compensated for that time only if plaintiffs requested overtime pay. After a twelve-day trial, the jury agreed and found that the City’s failure to pay for work it required was a willful violation of the Fair Labor Standards Act (“FLSA”). The City’s principal argument on appeal is that it cannot be held liable for the unpaid overtime because it affords an opportunity to report overtime work and, since the plaintiffs failed to report the work at issue, the City did not know that any plaintiff was being short-changed. But an employer must pay for all work it knows about or requires, even if the employee does not specifically request compensation for it. Whether an employee reports overtime work will often be relevant to an employer’s knowledge of the work — but allowing, or even requiring, an employee to report overtime work does not absolve employers of the obligation to compensate for work they suffer or permit. Moreover, we now hold that whether an employer knows an employee is not being paid is irrelevant to FLSA liability. If the employer suffers or permits the work — either by requiring it, knowing about it, or failing to exercise reasonable diligence to discover it — then it must compensate the employee, even if the employee failed to report the work and even if the employer did not know that the employee was working unpaid. And because the record supports the jury’s finding that the City had a policy or practice of requiring plaintiffs to perform work before and after their shifts, we uphold the jury’s verdict that the City violated the FLSA by not compensating them for that work. Nor do the City’s other arguments require reversal or vacatur. First, the jury’s willfulness finding is adequately supported by evidence that the City knew the plaintiffs were performing unreported extra-shift work yet took insufficient action to remedy the situation or to confirm its assumption that it was in compliance with the law. Next, the City complains that the district court should have — but did not — instruct the jury that plaintiffs had to show that 100 percent of the time included in plaintiffs’ damages calculation was FLSAcompensable. We conclude that this was not fatal to the verdict, since making such a showing would have been impossible in this case, and plaintiffs put forward an adequate approximation that showed the amount of their uncompensated work as a matter of just and reasonable inference. Finally, the City was not entitled to have the jury determine whether one certain component of the plaintiffs’ post-shift work was de minimis — an issue decided against the City at summary judgment. The de minimis inquiry generally applies to the claimed work as a whole, not to each task the employer requires. An employer may not avoid FLSA liability by segmenting extra-shift work into small tasks that may separately be deemed de minimis. And here, remand is inappropriate because no reasonable jury could have found for the City under any permissible framing of the de minimis inquiry. I On this appeal from a jury verdict, the facts are derived from trial testimony with all inferences drawn in favor of the plaintiffs. Plaintiffs are 2,519 Emergency Medical Technicians (“EMTs”) and paramedics1 employed by the Emergency Medical Services (“EMS”) division of the New York City Fire Department (“FDNY”) and thus ultimately by the City of New York (collectively, “the City”).2 As emergency responders, plaintiffs provide time-sensitive, potentially life-saving medical care in myriad emergency situations, including acute illness, drug overdoses, accidents, and shootings. Plaintiffs work eighthour shifts during which they are on call. Though based out of stationhouses throughout New York City, plaintiffs spend their shifts waiting in an ambulance at a designated location away from the stationhouse. An ambulance crew can receive a call at any time during the eight-hour shift. Preparation is needed before EMTs can set out with an ambulance. In order to respond to calls effectively and safely, each EMT has a set of personal protective equipment (“PPE”), including helmet, gloves, pants, coat, and a respirator. Before an EMT can log on to her ambulance, she must retrieve this PPE from her locker and inspect it to make sure it is in order. The same goes for gear, including a radio, radio holster, stethoscope, shears, and a duty belt. An EMT also carries a “Technician’s Bag” with additional first aid materials, which (like the other equipment) must be retrieved and inspected. Finally, once the outgoing shift has returned with the ambulance, EMTs must perform a thorough inspection of the vehicle before being able to log on as available to respond to a call. There is a similar sequence at the end of a shift. After returning to the station, plaintiffs: exchange certain equipment with the oncoming shift, noting the exchange in a logbook; inform the oncoming shift of pertinent information, such as hospital capacity, special events in the city, or issues with the ambulance; and secure and store their PPE and personal gear in the appropriate lockers. The City’s electronic timekeeping and payroll system, CityTime, utilizes a “pay to schedule” approach. J.A. 2920. Under that system, employees are automatically paid only for time during their shift, not for time at the station performing work before or afterward. CityTime registers presence at the station to the minute using scanners located by the entrance of each stationhouse. So, if an EMT scans in to CityTime ten minutes before the shift and scans out ten minutes after it ends, she will automatically be paid for the eight hours during the shift but not for the ten-minute intervals before and after, chunks of time the parties call “slivers.” See, e.g., id. Per City policy, an EMT who performs work during a sliver must submit an overtime request in order to be paid.3 Plaintiffs regularly requested overtime pay, including sometimes for pre-shift work. J.A. 2459. But plaintiffs did not request overtime pay on 99 percent of the occasions they scanned in before their shifts.4 J.A. 2410. In February 2013, plaintiffs brought this lawsuit, alleging that the City willfully violated the FLSA through a policy or practice of requiring work without payment unless pay was requested. They sought backpay for their unpaid extra-shift work. The City’s motion to dismiss the complaint was denied. Perry v. City of New York, No. 13-cv-1015, 2013 WL 6641893 (S.D.N.Y. Dec. 17, 2013). Following discovery and supplemental pleading, the parties cross-moved for summary judgment; the district court granted each motion in (small) part but left for the jury the key questions of the City’s policy regarding extra-shift work and the willfulness of any violation. See Perry v. City of New York, No. 13-cv-1015, 2018 WL 1474401 (S.D.N.Y. Mar. 26, 2018) (“Perry I”). The district court then certified plaintiffs’ suit as a “collective action.” Perry v. City of New York, No. 13-cv-1015, 2019 WL 1146581 (S.D.N.Y. Mar. 13, 2019) (“Perry II”); see 29 U.S.C. §216(b) (permitting collective suits “by any one or more employees for and in behalf of himself or themselves and other employees similarly situated”). The plaintiffs were similarly situated, the district court ruled, “with regard to [the allegation that] Defendants have a policy or practice requiring EMTs/Paramedics to conduct pre-and post-shift work, and that they are not compensated for all of this work.” Perry II, 2019 WL 1146581, at *8. In a twelve-day trial held six-and-a-half years after the complaint was initially filed, the jury heard testimony from thirteen plaintiffs, multiple FDNY supervisors and administrators, a city legal official, and competing experts on damages. The jury found the City liable for the unpaid overtime and found that its violation of the FLSA was willful. Based on that verdict, the district court ultimately entered a judgment against the City for $17,780,063, allocated as follows: backpay in the amount of $7,238,513; plus the same amount in liquidated damages;5 and $3,303,037 in attorneys’ fees. Final Judgment, Perry v. City of New York, No. 13-cv-1015 (S.D.N.Y. Feb. 5, 2020), Dkt. No. 313. The City then moved for judgment as a matter of law or for a new trial pursuant to Federal Rule of Civil Procedure 50(b), which the district court denied in August 2021. See Perry v. City of New York, 552 F. Supp. 3d 433 (S.D.N.Y. 2021) (“Perry III”). The City timely appealed. II The City’s principal attack is on the jury’s finding of an FLSA violation. In addition to disputing the existence of a policy requiring extra-shift work, the City asks this Court to adopt a rule of law that would require plaintiffs to show both that the employer required or knew about the overtime work and that it also knew that the workers would not be paid for it. In effect, the City argues that it does not have to compensate for required overtime work unless employees report the work and request pay. That appealing proposition is not the law. An employer violates the FLSA when it does not pay overtime wages for work it “suffers or permits,” that is, work it requires, knows about, or should have known about. Whether the employer also knows that the employee will not be paid is irrelevant to FLSA liability. Employers may, of course, require employees to report overtime work, and an employee’s failure to do so will in many circumstances allow the employer to disclaim the knowledge that triggers FLSA obligations. But an employer that nonetheless requires, knows about, or should know about work must compensate the worker, regardless of whether pay is requested and regardless of whether the employer knows the worker will not be paid. In light of the FLSA’s standards and the evidence adduced at trial, we decline to overturn the verdict. Though there were facts supporting both sides — there always are — plaintiffs presented enough evidence of the City’s policy or practice of requiring overtime work to support the jury’s verdict. A The Fair Labor Standards Act imposes minimum-wage and maximumhour requirements on certain U.S. employers. See 29 U.S.C. §§201-219. Among the Act’s mandates is that “no employer shall employ any of his employees…for a workweek longer than forty hours unless such employee receives compensation for his employment in excess of the hours above specified at a rate not less than one and one-half times the regular rate at which he is employed.” 29 U.S.C. §207(a)(1).6 That is, employees get time-and-a-half pay for “employment” in excess of 40 hours per workweek. Section 207 is “aimed not only at raising wages but also at limiting hours. In other words, these provisions were designed to remedy the ‘evil of overwork’ by ensuring workers were adequately compensated for long hours, as well as by applying financial pressure on employers to reduce overtime.” Chao v. Gotham Registry, Inc., 514 F.3d 280, 285 (2d Cir. 2008) (internal citation omitted) (quoting Overnight Motor Transp. Co. v. Missel, 316 U.S. 572, 578 (1942)). Employees may bring a civil suit against their employer to recover unpaid overtime wages. See 29 U.S.C. §216(b). The FLSA’s broad definition of “employ[ment]” “ includes to suffer or permit [an employee] to work.” 29 U.S.C. §203(g).7 So defined, “employment” includes work the employer requires; it also encompasses work the employer actually knows about (“actual knowledge”), as well as work it should have known about through the exercise of reasonable diligence (“constructive knowledge”).8 See 29 C.F.R. §785.11 (“Work not requested but suffered or permitted is work time.”); Kuebel v. Black & Decker Inc., 643 F.3d 352, 361 (2d Cir. 2011) (requiring “that the employer had actual or constructive knowledge of th[e claimed] work” for FLSA liability). This definition of “employment,” in tandem with §207′s mandate, produces the rule: an employer must pay overtime wages for work it (a) requires, (b) knows about, or (c) should have known about through the exercise of reasonable diligence. Some employers require employees to report any overtime work in order to receive compensation. This case asks us to consider what happens when an employee in such a system works overtime without reporting it. Employee reporting is obviously relevant to an employer’s knowledge of work: “While an employer must pay for work it suffers or permits, an employer cannot suffer or permit an employee to perform services about which the employer knows nothing.” Holzapfel v. Town of Newburgh, 145 F.3d 516, 524 (2d Cir. 1998) (internal citation omitted). It is also relevant for constructive knowledge: “establishing a reasonable process for an employee to report…work time” is one way to exercise reasonable diligence, so an employer with such a system will not ordinarily be chargeable with constructive knowledge of unreported work. Allen v. City of Chicago, 865 F.3d 936, 938 (7th Cir. 2017).9 But we have long recognized that an employee’s failure to report work the employer in fact knew about or required does not protect the employer from FLSA liability. In Caserta v. Home Lines Agency, Inc., 273 F.2d 943 (2d Cir. 1959), the defendant-employer “delegated to [the plaintiff] the duty of keeping time sheets showing the hours that he worked,” paying him “for all time and overtime shown on the time sheets” but “ke[eping] no independent record of his time.” 273 F.2d at 945. The plaintiff had failed to record certain required tasks in the erroneous belief that they were not compensable work, though he had complained about this work such that the employer “knew that plaintiff was working additional overtime not reflected in his time sheets.” Id. (quoting Caserta v. Home Lines Agency, Inc., 172 F. Supp. 409, 413 (S.D.N.Y. 1959)). The employer objected that the plaintiff was “barred…from claiming overtime pay for time not shown on his time sheets” and thus could not recover for the unreported work. Id. at 944. Writing for the Court, Judge Friendly rejected this argument as “inconsistent with both the language and the policy of the Fair Labor Standards Act,” id. at 946, holding that the employer could not escape liability for work it knew about (and indeed seems to have required10) by reason of the plaintiff’s failure to record the time. “The obligation [to comply with §207] is the employer’s and it is absolute,” we explained: “[The employer] cannot discharge it by attempting to transfer his statutory burdens of accurate record keeping and of appropriate payment[] to the employee.” Id. (internal citation omitted). In any wage-and-hour regulatory scheme, somebody must bear ultimate responsibility for recording time worked and for ensuring that payment is made. From Caserta on, we have recognized that the FLSA places the payment obligation on employers, a congressional choice consistent with the desire to “remedy the ‘evil of overwork’” and to “apply[] financial pressure on employers to reduce overtime,” Chao, 514 F.3d at 285. See, e.g., Foster v. City of New York, Nos. 14-cv-4142, 14-cv-9220, 2017 WL 11591568, at *20 (S.D.N.Y. Sept. 30, 2017) (Gardephe, J.) (“[T]he FLSA imposes on employers — and not on employees — the obligation to ensure that employee time records accurately reflect hours worked….”). If it wishes to avoid the expense of overtime work it requires, the employer must prevent the work from occurring. See 29 C.F.R. §785.13; Chao, 514 F.3d at 288. The principle — that failure to claim overtime does not discharge the duty to pay if the employer was on notice of the work — is well-settled in this Court and in other Circuits, as shown in the margin.11 It is instructive to consider the Tenth Circuit’s decision in Aguilar v. Management & Training Corp., 948 F.3d 1270 (10th Cir. 2020), which arose from facts similar to this case. The employer knew or required that its employees arrive early to (among other things) receive a pre-shift briefing and collect and inspect their gear; but it did not compensate for all of that time unless the employee submitted a request. See 948 F.3d at 1274-75. The Tenth Circuit acknowledged that “if an employer does not know that an employee is doing certain work, then the employer is not required to pay the employee for that work” — but it held that “if the employer is aware of the work and therefore ‘suffer[s] or permit[s]‘ the work, it must pay the employee.” Id. at 1286 (alterations in original). The employer “‘cannot stand idly by and allow [plaintiffs] to perform overtime work without proper compensation, even if’ the [plaintiffs] did not claim overtime compensation….” Id. at 1287 (quoting Fairchild v. All Am. Check Cashing, Inc., 815 F.3d 959, 964 (5th Cir. 2016)). The City seems to argue for a different rule: that an employer cannot be held liable for unpaid, unreported overtime work unless it knew that the employee would not be paid, even if it required or knew about the work. That is, the City argues that the decisive issue is the employer’s knowledge of nonpayment rather than its knowledge of the work itself. For support, the City cites White v. Baptist Memorial Health Care Corp., 699 F.3d 869 (6th Cir. 2012), in which a nurse alleged that she had not been compensated for time she worked during (unpaid) meal breaks. (Such a claim is conceptually identical to one for unreported overtime work, and it was analyzed as such. Id. at 873.) A divided panel of the Sixth Circuit ruled for the employer: although “White occasionally told her supervisors that she was not getting her meal breaks,” the court reasoned, “she never told her supervisors that she was not being compensated for missing her meal breaks.” Id. at 876. “Accordingly,” the court explained, “there is no way [the employer] should have known she was not being compensated for missing her meal breaks.” Id. (emphasis added); see also id. at 873 (“[T]he issue is whether [the employer] knew or had reason to know it was not compensating White for working during her meal breaks.”); Craig v. Bridges Bros. Trucking LLC, 823 F.3d 382, 389 (6th Cir. 2016) (summarizing White as requiring knowledge of non-payment). In addition, there appears to be division among the district courts of this Circuit regarding the relevance of knowledge of non-payment.12 We hold that knowledge of non-payment is irrelevant to FLSA liability.13 Our reasons are as follows. First, we consult the text of the FLSA. A worker is entitled to overtime pay for certain hours he is “employ[ed],” 29 U.S.C. §207(a)(1), which the statute defines to mean performing work his employer “suffer[s] or permit[s]” (i.e., requires, knows about, or should know about), 29 U.S.C. §203(g). Our decision in Caserta and our subsequent iterations of its rule in Holzapfel and Kuebel embody that straightforward understanding of the FLSA. To hold that §207 is violated only when the employer also knows that the employee will not be paid would add an extra-statutory precondition for liability. Second, employees cannot waive the protections of the FLSA. See, e.g., Tony & Susan Alamo Found. v. Sec’y of Lab., 471 U.S. 290, 302 (1985) (“[T]he purposes of the Act require that it be applied even to those who would decline its protections.”). Yet the City’s rule is tantamount to saying “that an employee may waive FLSA protections by not reporting time her employer knows about….” White, 699 F.3d at 881 (Moore, J., dissenting); see also Chao, 514 F.3d at 290 (“[O]nce it is established that an employer has knowledge of a worker’s overtime activities…liability does not turn on whether the employee agreed to work overtime voluntarily or under duress.”). We recognized in Caserta that “acts that would normally have controlling legal significance are overcome by Congressional policy,” such that “[a]n agreement…not to claim overtime pay for the work here in question would be no defense to [the employee] later demanding it.” 273 F.2d at 946. Third, the City’s rule would collapse the “significant distinction between ordinary violations and willful violations” of the FLSA. McLaughlin v. Richland Shoe Co., 486 U.S. 128, 132 (1988). As discussed infra, “[a]n employer willfully violates the FLSA when it either knew or showed reckless disregard for the matter of whether its conduct was prohibited by the Act.” Kuebel, 643 F.3d at 366 (internal quotation marks and citation omitted). That is, an employer violates the FLSA willfully when it knows that it is not paying an employee for compensable work; but if an employer is not liable unless it knows that its employees are not being paid (as the City contends), then every violation of §207 would automatically be willful, notwithstanding Congress’s “obvious” intent to distinguish between these two kinds of violations, McLaughlin, 486 U.S. at 132-33 (rejecting a rule that would “virtually obliterate[] any distinction between willful and nonwillful violations”). Nothing we say here calls into question the lawfulness or usefulness of systems requiring employees to report their overtime work. As explained above, such a system will in effect preclude liability in many situations.14 But it will not exempt an employer who required, knew about, or should have known about the claimed work. Accordingly, liability here (like in all FLSA overtime cases) is determined by whether the City suffered or permitted plaintiffs’ pre-and postshift work. We turn to that question next. B The City argues that the evidence at trial cannot support the jury’s finding that the City “ha[d] a policy or practice of suffering or permitting the Plaintiffs to perform work before [and after] their shift without pay, in violation of the Fair Labor Standards Act[.]“15 J.A. 3279. “We review de novo the district court’s decision on a motion for judgment as a matter of law, applying the same standard that is required of the district court.” Ojeda v. Metro. Transp. Auth., 41 F.4th 56, 63 (2d Cir. 2022) (quoting Smalls v. Collins, 10 F.4th 117, 131 (2d Cir. 2021)). That is, the Court must “find[] that a reasonable jury would not have a legally sufficient evidentiary basis to find for the [non-movant].” Fed. R. Civ. P. 50(a)(1). In doing so, we “consider[] the evidence in the light most favorable to the non-moving party and giv[e] that party the benefit of all reasonable inferences that the jury might have drawn in that party’s favor.” Syntel Sterling Best Shores Mauritius Ltd. v. The TriZetto Grp., Inc., 68 F.4th 792, 800 (2d Cir. 2023) (“TriZetto”) (quoting Triolo v. Nassau Cnty., 24 F.4th 98, 105 (2d Cir. 2022)). “We affirm the denial of a Rule 50(b) motion ‘unless there is such a complete absence of evidence supporting the verdict that the jury’s findings could only have been the result of sheer surmise and conjecture, or the evidence in favor of the movant is so overwhelming that reasonable and fair minded persons could not arrive at a verdict against it.’” Id. (quoting Ashley v. City of New York, 992 F.3d 128, 138-39 (2d Cir. 2021)). The burden on the movant “is ‘particularly heavy’ where, as here, the ‘jury has deliberated in the case and actually returned its verdict.’” Triolo, 24 F.4th at 105 (quoting Cross v. N.Y.C. Transit Auth., 417 F.3d 241, 248 (2d Cir. 2005)). The City has not sustained its “particularly heavy” burden here: a reasonable jury could — as it did — find that the City maintained a policy or practice of requiring EMTs and paramedics to perform compensable extra-shift work. And since plaintiffs were not compensated for that “employment,” the jury properly found that the FLSA was violated. Plaintiffs are required to be in their ambulance and ready to respond to calls “as soon as possible after their scheduled shift starts.” J.A. 2343; accord J.A. 2246, 2273-74. Plaintiffs testified to being held to targets of five, J.A. 1676; seven, J.A. 1836; and ten minutes after the scheduled start of their shift, J.A. 1471. Their performance evaluations corroborate the expectation that plaintiffs be ready to respond to calls as soon as possible. See, e.g., J.A. 4943, 4995, 5001, 5046. But as described above, there are preliminaries before an EMT can be ready to answer a call. One preliminary is a thorough inspection of the ambulance, which involves “inspect[ing] the patient compartment,” “check[ing] the outside and mechanical/safety features of the ambulance,” “check[ing] and verif[ying] [that all patient equipment is] in working condition,” and securing “all equipment in the rear of the vehicle…for safety.” J.A. 4848-50. Since the outgoing tour has to return the ambulance to the station before it can be inspected, those checks usually have to take place at the very beginning of the paid shift: tellingly, this inspection procedure is set out in an EMS training document entitled “The First Five Minutes of Your Tour.” J.A. 4844. In order to do the ambulance inspection at the beginning of the shift, the other preliminaries must be completed before the shift. Recognizing this, FDNY regulations require that “[m]embers of the Bureau of EMS…[r]eport punctually for duty as scheduled, in proper uniform, with all issued equipment and ready for duty, unless properly excused.” J.A. 4822-23 (2012 regulation) (emphasis added); accord J.A. 4832 (identical 2018 regulation). And the testimony of multiple witnesses, both line EMTs and supervisors, was that an EMT is not “ready for duty” unless he has collected and inspected the PPE, retrieved and checked the tech bag, and otherwise prepared the equipment for the upcoming tour.16 Those inspections, like the required ambulance check, are by no means cursory. An “EMS Academy” training document from 2016 explained that each item of PPE “should be taken out of the gear bag and inspected for contamination, fabric or material damage, thread or seam damage and damage to the reflective trim.” J.A. 4841. A similar document reminded EMTs to “check their own equipment” in addition to the ambulance: “[PPE] gear should be taken out of their respective bags, inspected and verified to be in a clean and functional condition.” J.A. 4850. In line with these training documents, one plaintiff attested to the expectation that an EMT is “responsible for making sure that [PPE] is in good working order before [he] actually need[s] to use it…. [b]y actually physically taking it out of the bag, holding it in [his] hand and examining it…. [j]ust prior to the start of [his] tour.” J.A. 1274. The jury could thus conclude that, in order to comply with the written requirement that EMTs and paramedics be “ready for duty” at the start of each shift, and to meet the expectation that they log on to a fully-inspected ambulance as soon as possible after their shift starts, plaintiffs must arrive early to prepare their equipment and PPE. See J.A. 1695-96 (“Q: [I]s it your understanding that at the start of your paid shift, you are expected to have all your equipment and be ready for duty? A: Yes. Q: Can you comply with that order when you clock in right at the start of your paid shift? A: No.”). Performance evaluations corroborate this aspect of the plaintiffs’ job: several plaintiffs were commended for coming to work early in order to prepare for their shift. See, e.g., J.A. 4943, 5046; see also J.A. 1446 (testimony of supervisor: Q: “[D]oes the FDNY look favorably upon individuals who arrive prior to the start of their shift?…A: Yes.”). And multiple plaintiffs testified that, in order to be an effective emergency responder, they had to start work early.17 The City argues that pre-shift work could not have been “required” because many plaintiffs arrived right at the start of their shifts and were never reprimanded for doing so. In the City’s view, this “proves, quite simply, that they were not required to arrive early, nor to do any work before their shifts started.” Appellants’ Br. at 40. While this is of course relevant, it does not compel a jury finding, nor does it in itself refute the countervailing evidence set out above. Given the nature of the plaintiffs’ jobs and the expectations for their performance, the response that plaintiffs didn’t have to come in early is not compelling. Cf. Kosakow v. New Rochelle Radiology Assocs., P.C., 274 F.3d 706, 718 (2d Cir. 2001) (rejecting argument that pre-shift preparatory tasks were not compensable work because plaintiff was “never expressly ‘required’ to arrive prior to [her shift] by [her] supervisors”: “If the proper performance of their job required the preparatory work to be completed [pre-shift], this time should have been counted, regardless of whether anybody specifically instructed them to arrive early.”). In sum, there is trial evidence that the City created a job that requires significant preparation to be done properly; it then also required that plaintiffs be ready to go as soon as possible after the start of their shift. The City issued formal regulations and enforced informal expectations to that effect. Testimony showed that it was essentially impossible to comply with the City’s demands without starting work before the paid shift. Considering the evidence in the light most favorable to the plaintiffs, the jury had a legally sufficient evidentiary basis for concluding that the City maintained a policy or practice of “requiring” plaintiffs to perform compensable, pre-shift work for which they were not automatically paid. Similar reasoning supports the verdict with respect to post-shift work. Plaintiffs adduced evidence of a de facto requirement that plaintiffs perform work after their shifts. To wrap up their days, EMTs must exchange equipment with the oncoming tour; provide that tour with pertinent information; and store PPE and personal gear, re-inspecting anything used during the shift. Nearly every trial witness testified that EMTs regularly perform these tasks at the end of their shifts, and the City does not contest that they are necessary parts of the job. But these wind-down tasks cannot be performed during the paid shift because EMTs must keep their equipment with them and ready to go for the whole of the shift. As one EMS lieutenant testified, “the FDNY requires [EMTs] to maintain [their] equipment…until the minute their shift ends” because “at any moment in time [they] can get called for an assignment,” including “one minute to the hour that [they] get off.” J.A. 1459-60; accord J.A. 2235 (Q: “[T]he outgoing tour [is] not permitted to hand off any of that equipment until after their tour has ended, correct? A: That’s true. Q: And that’s because they might get a call up to the second before their shift ends, correct? A: That is true.”). The City does not meaningfully contest any of these premises, and the trial record amply supports them. It thus follows (and the jury could reasonably conclude) that EMTs must perform work tasks after their paid shift. See J.A. 2258-59 (conceding that “[i]t’s highly unlikely” that an EMT would be able to comply with the expectation to keep equipment until the end of the tour without working after the end of the shift). As with pre-shift work, the City constructed a position that required work after the employee’s shift in order to be performed properly. * * * Combining the jury’s factfinding with the legal regime outlined in Part II.A, things come into focus. The jury reasonably concluded that the City had a policy or practice of requiring plaintiffs to report to their stations early in order to prepare for their shift and to stay late to perform necessary wrap-up tasks. That is, the jury found that the City “employed” plaintiffs when they performed the pre-and post-shift work at issue. Yet the City did not automatically pay them for this time, even though the FLSA places the burden of ensuring compensation for such employment on the employer. For the reasons explained above, that is a violation of the FLSA, regardless of whether plaintiffs requested payment for their work and regardless of whether the City knew that plaintiffs were working unpaid. We therefore affirm the district court’s denial of the City’s motion for judgment as a matter of law and decline to set aside the jury’s liability verdict. III The jury further found that the City’s FLSA violation was “willful.” “An employer willfully violates the FLSA when it either knew or showed reckless disregard for the matter of whether its conduct was prohibited by the Act.” Kuebel v. Black & Decker Inc., 643 F.3d 352, 366 (2d Cir. 2011) (quoting Young v. Cooper Cameron Corp., 586 F.3d 201, 207 (2d Cir. 2009)). In a suit for unpaid overtime wages, willfulness means that the employer knew that it had failed to properly pay its employees or was reckless with regard to that failure. So although knowledge of non-payment is irrelevant to liability, it is the sine qua non of willfulness. “Reckless disregard can be shown through ‘action entailing an unjustifiably high risk of harm that is either known or so obvious that it should be known.’” Mumby v. Pure Energy Servs. (USA), Inc., 636 F.3d 1266, 1270 (10th Cir. 2011) (quoting Safeco Ins. Co. of America v. Burr, 551 U.S. 47, 68 (2007)). An employer can evince reckless disregard for the unlawfulness of its conduct even when it “may not have had actual knowledge of the violative practices.” Herman v. RSR Sec. Servs. Ltd., 172 F.3d 132, 141 (2d Cir. 1999). Still, “if an employer acts unreasonably, but not recklessly, in determining its legal obligation, its action should not be considered willful.” Whiteside v. Hover-Davis, Inc., 995 F.3d 315, 324 (2d Cir. 2021) (quoting Reich v. Waldbaum, Inc., 52 F.3d 35, 39 (2d Cir. 1995)). The “operative inquiry focuses on the employer’s diligence in the face of a statutory obligation.” Mumby, 636 F.3d at 1270. The City challenges the jury’s willfulness finding. Here too, it bears the “particularly heavy” burden of showing that “a reasonable jury would not have a legally sufficient evidentiary basis” for its finding. Fed. R. Civ. P. 50(a)(1). The line between “unreasonable” and “reckless” conduct is for the jury to draw — appellate courts are rightfully hesitant to overturn a jury’s delineation. E.g., Chesher v. Neyer, 477 F.3d 784, 804 (6th Cir. 2007) (“[S]uch distinctions [between negligence and recklessness] are almost always left for a jury — rather than appellate judges — to decide.”). Our decision in Herman v. RSR Security Services is instructive. There, the defendant company (“RSR”) had underpaid its employees in various ways, and Murray Portnoy, the company’s chairman, challenged the finding that his conduct was willful. See 172 F.3d at 138, 141-42. Portnoy was not directly involved with payroll, but he knew that Michael Stern, the individual running that department, “had conducted his earlier business activities in an illegal manner,” id. at 142; Portnoy had also discovered that Stern had wrongly designated some RSR employees as independent contractors on tax filings, a separate violation, id. Portnoy nonetheless took Stern’s word for it that their company was otherwise obeying the law and “made no effort to ascertain [his company's] compliance with the FLSA” even though he “could easily have inquired into the pay rates for the [employees].” Id. Employing a highly deferential standard of review (as we do here), id. at 139, we affirmed the willfulness finding. Although Portnoy had “repeatedly checked with [Stern] in order to ensure that RSR was complying with the law,” the factfinder could find that, given his notice of Stern’s past unlawful behavior, Portnoy’s reliance “on information from [Stern] in this context was reckless.” Id. at 142. Herman thus suggests that failure to confirm an assumption of compliance when the defendant has reason to suspect otherwise can support a finding of willfulness on a theory of reckless disregard. Here, the jury found that plaintiffs “prove[d] by a preponderance of the evidence that the Defendants willfully violated the Fair Labor Standards Act,” J.A. 3281, a finding with two important consequences. First, it “extend[ed] the statute of limitations period from two to three years,” Young, 586 F.3d at 207; see 29 U.S.C. §255(a), so plaintiffs’ recovery period started in 2010 instead of 2011 — a difference of approximately a million minutes worked and $250,000 in backpay, J.A. 2421-23. Second, in this case the jury’s willfulness determination impacted liquidated damages. Successful FLSA plaintiffs receive liquidated damages equal to the amount of backpay, see supra n.5, but the district court may reduce the liquidated damages award “if the employer shows to the satisfaction of the court that the act or omission giving rise to such action was in good faith….” 29 U.S.C. §260. Judge Broderick concluded that he lacked discretion to reduce the award in light of the jury’s willfulness verdict, reasoning that an employer acting in knowing or reckless disregard of its legal obligations cannot also be acting in good faith. See Perry v. City of New York, 2019 WL 7047327, at *2-4 (S.D.N.Y. Dec. 23, 2019). (That conclusion has not been challenged on appeal.) The jury’s willfulness verdict therefore resulted in a full liquidated damages award of $7.2 million. Thus, significant stakes ride on whether the jury’s willfulness finding is supported by the evidence. We conclude that it is. First, the jury could find that the City was aware that the FLSA required it to ensure compensation for all the overtime work it knew about, regardless of whether the employee reported the work or requested compensation. Georgia Pestana, a senior labor lawyer for the City, testified that she told FDNY that supervisors have a duty to ensure that employees whom they know to be working overtime receive compensation, even if the employee did not submit an overtime request. See J.A. 2113-15 (“Q: So it would be an employer’s responsibility, the supervisor’s responsibility to ensure that the employees get paid for work about which the employer is aware; correct? A: Correct…. Q: This is the advice that you gave to EMS over and over, correct?…A: I gave it to EMS.”). The evidence also supports a finding that the City knew that some required extra-shift work was not being compensated. When CityTime was being designed in 2005, more than a thousand EMTs and paramedics asked that the new system accommodate the extra-shift work they perform, thereby putting the City on notice that such work was occurring and that at least some of it was going uncompensated. E.g., J.A. 1827, 2076, 2110; see also Perry III, 552 F. Supp. 3d at 442 (noting that the City had received “complaints from thousands of [its] employees” in 2005 regarding “potential FLSA violations”). Then, in 2008, the City prepared a draft order forbidding EMS employees from performing extrashift work “unless the work has been approved by a [supervisor] and such time has been accurately recorded on their timesheets.” J.A. 5140. The jury could reasonably infer from the City’s drafting of this order that it knew that extra-shift work was being performed, that it was compensable, and that EMTs and paramedics were not reliably reporting (nor being paid for) such work. See Perry III, 552 F. Supp. 3d at 442. It was not necessary to demonstrate that the City knew that the particular slivers now at issue were going unpaid: knowledge that some compensable, extra-shift work was not being paid suffices to put the City on notice. Despite its apparent knowledge of non-payment and of its FLSA obligations, the City did little to fix the problem. The 2008 draft order remained a draft — it was not until six years later that the FDNY actually promulgated an order prohibiting unreported pre-shift work. See J.A. 2387-90 (testimony to that effect); J.A. 4914 (order as issued in 2014). As the district court observed, “[i]t is…reasonable for a jury to infer that…Defendants evaluated the problem and made a conscious decision to refrain from implementing the order, either to avoid paying Plaintiffs additional money or because the policy would force Defendants to change the way they managed EMTs and paramedics.” Perry III, 552 F. Supp. 3d at 442. Nor did the City ensure that FDNY leadership understood their obligation to provide overtime compensation. The FDNY’s assistant commissioner for budget and finance did not recall ever being told “that [the City] must pay for work that it knows or should know is being performed even if there is no overtime request put in for that work.” J.A. 2064-65. The same was true of the director of payroll, timekeeping, and compliance services, J.A. 1903-05, who instead operated on the belief that “[t]he responsibility falls on the employee” to request overtime even if a supervisor knew that the employee was working, J.A. 1900. Two senior FDNY supervisors testified that plaintiffs were not permitted to request overtime for such pre-shift work as checking PPE and equipment. Deputy Chief Norman Ortiz, a twenty-five-year veteran of the FDNY (twelve years as a supervisor) and one of 20 deputy chiefs in the EMS Bureau, was of the view that “there is no way for the EMTs or paramedics to ask for overtime for checking their PPE prior to the start of their shift” and that FDNY “[does not] allow people to request overtime for things like checking equipment prior to the start of the shift.” J.A. 1843-44. Such time, Deputy Chief Ortiz agreed, is “not requestable.” Id. at 1844. Lieutenant Jose Gonzalez, a supervisor of fourteen years, likewise agreed that “EMTs and paramedics could not submit [overtime] requests for checking their PPE prior to the start of their shift” or “for checking their tech bag prior to their shift,” though he also testified that he prefers that those tasks be performed before the shift begins. J.A. 1515. The trial testimony reflected a corresponding understanding among EMTs and paramedics.18 Taken together, this testimony could easily lead a jury to conclude that the City was reckless with respect to ensuring that EMTs were paid for their extra-shift work. The City disclaims knowledge that plaintiffs were not being paid for at least some extra-shift work and instead contends that it “reasonably believed that when plaintiffs worked compensable non-shift hours, they would submit overtime requests.” Appellants’ Br. at 55. In support, the City observes that EMTs submitting their weekly timecards must check a box certifying “that I have requested compensation for any time that I worked in excess of my scheduled hours and that any time outside my scheduled hours…for which I have not requested compensation[] was time not worked.” E.g., J.A. 3905; id. at 1473-74 (testimony to that effect). These certifications surely constitute evidence relevant to the City’s knowledge of non-payment; but in light of the countervailing evidence, the jury was not required to accept the City’s claim. Cf. Aguilar v. Mgmt. & Training Corp., 948 F.3d 1270, 1287 (10th Cir. 2020) (rejecting employer’s argument that it did not know about extra-shift work because employees had “signed an acknowledgement form included with each paycheck stating that they submitted [a request] for any overtime work conducted before or after their shift” when evidence demonstrated that it required and/or had actual knowledge of the work). Even accepting arguendo that the City assumed plaintiffs were requesting overtime whenever appropriate, City officials acknowledged at trial that they took no steps to confirm that assumption. CityTime records every minute EMTs are at their stations, but the City never audited or otherwise monitored the uncompensated minutes logged in CityTime to ascertain whether overtime work was being paid. See J.A. 1473-74, 2075-76. Not unlike the defendant in Herman who credited false assurances of compliance, the City (at best) assumed without confirmation that overtime compensation was being claimed and paid notwithstanding reason to suspect otherwise. The City also argues that it could not have violated the FLSA willfully because CityTime was designed in consultation with attorneys who advised the City that “it could require employees to make use of a readily available timekeeping system to request overtime.” Appellants’ Br. at 55; see J.A. 2920, 2924-25, 2934 (testimony regarding lawyers’ advice). We disagree. Attorney approval of a practice can demonstrate that an FLSA violation was not willful,19 but the City’s liability here does not arise solely from the attorney-approved design of CityTime. The violation was that the City required work which it then failed to pay for. And the testimony of Georgia Pestana demonstrates that the City was never advised that it could refuse compensation for unreported work it required or knew about. In sum, the jury could have reasonably concluded that (1) there was a citywide policy or practice requiring extra-shift work; (2) the City knew that it was responsible for paying its employees for extra-shift work regardless of whether they reported it; (3) the City knew that EMTs and paramedics were not reporting (and therefore not being paid for) some extra-shift work being performed; but (4) the City failed for six years to prohibit unreported overtime after it saw the need to do so; and (5) FDNY’s senior managers were neither made aware of their obligations with respect to unreported overtime, nor were they even unanimous in believing that overtime was available for the pre-shift work at issue. At minimum, the jury could have found that the City failed to take reasonable steps to confirm its assumption that it was complying with the FLSA despite having reason to believe that many plaintiffs were not reporting their work. A jury could easily have found that the City’s conduct was merely unwise or unreasonable. But this jury received accurate instructions and concluded that it amounted to recklessness. We owe proper deference to that assessment. Since our examination of the record does not reveal evidence favoring the City “so overwhelming that reasonable and fair minded persons could not arrive at a verdict against it,” TriZetto, 68 F.4th at 800 (citation omitted), we affirm the district court’s denial of the City’s Rule 50(b) motion and decline to set aside the jury’s willfulness verdict. IV As to damages, the City argues that the district court, in answering a question from the jury about the verdict sheet, effectively instructed that the jurors need not decide an essential factual premise underlying plaintiffs’ damages calculation. This error, the City contends, requires that we set aside the judgment and remand for a new trial on damages. The City’s argument has force; but for the reasons explained below, we do not agree that the alleged error necessitates retrial. To understand the City’s argument one must first understand how plaintiffs — and then the district court — calculated damages. CityTime records to the minute how much time each EMT spends at the station; intervals logged in CityTime outside of the compensated shift are known as “slivers.” Plaintiffs’ damages calculation was built around this readily-accessible source of historical data, but they recognized that the CityTime slivers had to be modified in several ways to avoid duplicative or excessive compensation. Most obviously, some plaintiffs had already requested and received overtime pay for certain slivers, so paid slivers were subtracted from the damages calculation. Likewise, the FLSA only entitles plaintiffs to overtime for employment in excess of forty hours per workweek. So even though their collective bargaining agreement entitles plaintiffs to compensation for all extrashift work, see supra n.3, the damages calculation reflects pay only for hours worked in excess of forty. Plaintiffs also rounded the slivers to the nearest fifteen minutes: if a plaintiff clocked in seven minutes or fewer before the shift, the sliver would round down to zero; if a plaintiff clocked in between eight and twenty-two minutes before the shift, the sliver would round to fifteen minutes. See J.A. 2406-09, 2467-68. (The same rules apply after the shift as well.) This rounding rule conforms to the practice that ordinarily governs EMTs’ overtime requests per their collective bargaining agreement. In that way, the damages award would approximate what a plaintiff would have received if overtime had been requested in the ordinary course. Finally, plaintiffs capped their claims at fifteen minutes per plaintiff per sliver: even if a plaintiff scanned out “32 minutes after the paid end time…. it would be just 15 minutes of post-shift time going into the calculation.” J.A. 2409 (testimony of plaintiffs’ damages expert). This cap reflects that, if an EMT arrives early for coffee and the newspaper before doing ten minutes of pre-shift work, the amount of time spent at the station will overstate the time spent working. The fifteen-minute limit obviates this potential discrepancy while ensuring compensation for the ten minutes of work performed. The City does not contest these limits on plaintiffs’ damages calculation. Instead, it argues that CityTime logs presence at the station whereas a given plaintiff is entitled to compensation only for time spent working. Thus, according to the City, plaintiffs had to prove (and the jury had to find) that each sliver in the damages award represented 100 percent compensable work time. There are two ways plaintiffs could make that showing: either by demonstrating that each sliver was occupied only by compensable work, or by showing that plaintiffs performed a work task immediately upon scanning into CityTime, thereby triggering the “continuous workday” during which even time not spent actively working is FLSA-compensable.20 Plaintiffs pursued the latter method, arguing that EMTs and paramedics would receive a pre-shift briefing from their lieutenant — concededly a work task — immediately after scanning in, thus triggering the continuous workday doctrine. The jury’s consideration of damages was assisted by a special verdict sheet offering two alternative methods of calculation. First, the jury was asked: Did the Plaintiffs prove by a preponderance of the evidence that the CityTime system accurately captures the unpaid [extra]-shift work minutes at issue in this case (i.e., up to 15 minutes of [extra]-shift work)? J.A. 3280. If not, the jury was to estimate “[o]n average, how many unpaid [extra]-shift minutes, per Plaintiff, per shift, should be counted as work time for which the Plaintiffs should have been paid[.]” Id. Thus, the jury could either use the modified CityTime slivers, or figure it out for itself. See J.A. 3260 (jury charge evincing this understanding). During deliberations, the jury sought “clarification” regarding the first option: “Is [the verdict sheet] asking whether the CityTime system documentation of preshift work time should all be considered work time? Or is the question asking if the preshift work time in question is accurately captured in CityTime?” J.A. 3271-72. This question probed the ambiguity in the word “captured” — did the jury have to find that the slivers constituted the work time or only included it? The City advocated the former, but the district court endorsed the latter. So instructed, the jury agreed that the work minutes were “accurately captured” by CityTime, basing the damages calculation on the modified slivers. The City argues that this instruction was reversible error. By telling the jury to determine only whether the slivers included the compensable work time, the City contends that the district court relieved plaintiffs of having to prove that all of the time recorded in CityTime was compensable. We conclude that the district court’s answer to the jury’s question does not fatally undermine the damages award. At the outset, we of course agree that an FLSA plaintiff is not entitled to backpay for non-compensable time. And it is of course a plaintiff’s burden to demonstrate entitlement to the relief sought. But the law of damages does not require the impossible. When a plaintiff has demonstrated the existence of harm, but the situation is “of such a nature as to preclude the ascertainment of the amount of damages with certainty,” then “it would be a perversion of fundamental principles of justice to deny all relief to the injured person, and thereby relieve the wrongdoer from making any amend for his acts.” Story Parchment Co. v. Paterson Parchment Paper Co., 282 U.S. 555, 563 (1931) (emphasis added). The district court gave the jury a similar instruction here: “[T]he law does not require plaintiffs to prove the amount of their losses with mathematical precision, but only with as much definiteness and accuracy as the circumstances permit.” J.A. 3258; see, e.g., Hydro Invs., Inc. v. Trafalgar Power Inc., 227 F.3d 8, 19 (2d Cir. 2000). This principle is familiar to FLSA litigation. The Supreme Court held in 1946 that when an employer’s records are “inaccurate or inadequate” and thereby prevent an employee from “prov[ing] the precise extent of uncompensated work,” it is enough “if [the employee] 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.” Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680, 687 (1946). So long as “[t]he uncertainty lies only in the amount of damages arising from the statutory violation,” “[i]t is enough…if there is a basis for a reasonable inference as to the extent of the damages.” Id. at 688. This rule is rooted in fundamental fairness: “[T]he employer, having received the benefits of such work, cannot object to the payment for the work on the most accurate basis possible under the circumstances.” Id.; see also Perma Rsch. & Dev. v. Singer Co., 542 F.2d 111, 116 (2d Cir. 1976) (“[S]ince [the defendant] produced the damage, it must bear the uncertainty of proof.” (citing Autowest Inc. v. Peugeot, Inc., 434 F.2d 556, 565 (2d Cir. 1970))). The jury was instructed on this principle as well. See J.A. 3260-61. Here, plaintiffs had no way to prove the exact number of compensable-yetuncompensated minutes worked. Though many plaintiffs testified regarding how often their “continuous workday” began right when they scanned in, they simply could not prove that every sliver for every plaintiff was exclusively compensable time. Nor could CityTime do this job: though neither unlawful nor unreasonable, CityTime was “inadequate” for the task of “prov[ing] the precise extent of uncompensated work,” Anderson, 328 U.S. at 687. The “solution” to this problem of proof, as the Supreme Court recognized in Anderson, “is not to penalize the employee by denying him any recovery[.]” Id.; see also Story Parchment, 282 U.S. at 563. Speculation is of course impermissible. Story Parchment, 282 U.S. at 563 (“ [D]amages may not be determined by mere speculation or guess.”). But so long as plaintiffs “produce[] sufficient evidence to show the amount and extent of that work as a matter of just and reasonable inference” and “there is a basis for a reasonable inference as to the extent of the damages,” Anderson, 328 U.S. at 687-88, the jury may rely on a reasonable approximation. So, the fact that the jury was instructed to determine only whether the compensable work time is included in the slivers does not invalidate the jury’s award so long as it could have concluded that plaintiffs put forth a sufficiently accurate approximation of the amount of their damages. Such a conclusion is warranted here: given the obstacles to making a more precise showing, plaintiffs’ damages calculation was close enough, even without a finding that every sliver was completely compensable. Several considerations lend confidence to the general accuracy of plaintiffs’ damages calculation. Plaintiffs adduced evidence that it was common for EMTs and paramedics to receive a pre-shift briefing immediately after scanning into CityTime, thereby triggering the continuous workday doctrine. Indeed, one plaintiff testified that he spoke to his lieutenant right after scanning in 90 percent of the time. J.A. 1419. Every time a plaintiff received a pre-shift briefing immediately after scanning into CityTime, the resulting sliver was entirely compensable time, as the City agrees. See Appellants’ Br. at 42-43. So although plaintiffs could hardly demonstrate that every plaintiff always received a preshift briefing immediately after scanning in — and the City introduced evidence tending to disprove such an absolute claim21 — there was support in the record for the conclusion that most of the slivers were completely compensable. Moreover, plaintiffs’ modifications to the CityTime data minimized discrepancies between the sliver totals and the true (and unknowable) amount of FLSA-compensable time. As explained above, plaintiffs excluded time for which they had already been paid; they claimed compensation only for time in excess of 40 hours per week; and they rounded the slivers according to the rule which ordinarily governs EMTs’ overtime requests. Perhaps most importantly, they limited each claim to fifteen minutes per sliver to reflect the usual duration of the required extra-shift work while ensuring that plaintiffs are not credited for time spent at leisure. Once one accounts for the rounding rule and the fifteen-minute cap, a relatively small number of slivers would be over-counted. In summary: the City claims that the district court erred by relieving plaintiffs of the obligation to prove that 100 percent of the claimed time was compensable. But on the evidence in this case, such a showing would be impossible, and to require it unreasonable. See Anderson, 328 U.S. at 687 (“The remedial nature of this statute and the great public policy which it embodies…militate against making that burden [of proving that the employee performed uncompensated work] an impossible hurdle….”). Therefore, we do not think the verdict or the resulting judgment invalid because the jury was not instructed to make such a finding. “[H]aving received the benefits of” plaintiffs’ required extra-shift work, the City “cannot object to the payment for the work on the most accurate basis possible under the circumstances.” Id. at 688. The jury could reasonably conclude that plaintiffs’ damages calculation, given its various limitations, was indeed the “most accurate basis possible under the circumstances” (and certainly that it “showed the amount and extent of that work as a matter of just and reasonable inference”). Id. at 687-88. The judgment based on that verdict may therefore stand, and we decline to remand for a new trial. V The City’s final ground for appeal is that the district court erred by instructing the jury that the EMT-plaintiffs’ post-work equipment exchanges were not “de minimis.” See J.A. 3254-57 (jury instruction in question). The district court had made a corresponding ruling at summary judgment — that “the time spent on exchanging equipment or narcotics is not de minimis” as a matter of law, Perry I, 2018 WL 1474401, at *4 n.5 — and declined the City’s request to reconsider that ruling at trial. The City argues that the district court’s summary judgment ruling was unsupported by the record, that it was entitled to present this argument to the jury, and that we need to remand for a new trial. A Plaintiffs first argue that this issue is not properly before us, contending that a party may not appeal the denial of summary judgment once there has been a trial on the merits. See Ortiz v. Jordan, 562 U.S. 180, 183-84 (2011); see also Schaefer v. State Ins. Fund, 207 F.3d 139, 142 (2d Cir. 2000) (stating same rule). But the Ortiz rule applies to a denial of summary judgment: appellate courts lack jurisdiction over such an order because it “retains its interlocutory character as simply a step along the route to final judgment.” Ortiz, 562 U.S. at 184. Ortiz explains that, after a trial on the merits, a defendant may not appeal a summary judgment denial that was based on the need for a trier of fact to resolve factual disputes. But here, the district court took an issue away from the jury and decided it as a matter of law, which is more closely analogous to a (partial) grant of summary judgment. “[E]arlier summary dispositions merge in the judgment and are reviewable” on “appeal from a final judgment concluding the action.” Gold v. New York Life Ins. Co., 730 F.3d 137, 144 (2d Cir. 2013) (quoting West v. Goodyear Tire & Rubber Co., 167 F.3d 776, 781 (2d Cir. 1999)); see also Dupree v. Younger, 143 S. Ct. 1382, 1389 (2023) (recognizing the “‘general rule’ [that interlocutory rulings] merge into the final judgment, at which point they are reviewable on appeal” (quoting Quackenbush v. Allstate Ins. Co., 517 U.S. 706, 712 (1996))). For instance, in Abrams v. Department of Public Safety, 764 F.3d 244 (2d Cir. 2014), the district court granted summary judgment to the defendant on all but one claim. After going to trial on the remaining claim (and losing), the plaintiff appealed both the jury verdict and the district court’s summary judgment decision. 764 F.3d at 247. We had jurisdiction over the entirety of the appeal — and we exercised it, vacating the district court’s prior partial grant of summary judgment with respect to two claims. Id. at 257. We see nothing in Ortiz to abrogate this longstanding rule. More generally, the Ortiz rule is fundamentally a “preservation requirement.” Dupree, 143 S. Ct. at 1387; accord id. at 1389 (“Ortiz holds[] that a party must raise a sufficiency-of-the-evidence claim in a post-trial motion to preserve it for appeal.”); see also Appellees’ Br. at 52 (“A motion for summary judgment does not preserve an issue for appellate review of a final judgment entered after trial.” (citing Ortiz, 562 U.S. at 184)). And there can be no reasonable argument that the City has somehow forfeited or failed to preserve the claim at issue. It asked the district court to reconsider its previous ruling in light of the trial evidence and to send the de minimis question to the jury. J.A. 3096-101. And it argued post-trial that the district court erred by instructing the jury consistently with its summary judgment ruling. Defs.’ Mem. of Law. in Supp. Mot. for J. as a Matter of L. at 32-35, Perry III, 552 F. Supp. 3d 433 (S.D.N.Y. 2021) (No. 13-cv-1015), Dkt. No. 319. It is unclear what more plaintiffs would have had the City do to preserve this issue. Cf. Omega SA v. 375 Canal, LLC, 984 F.3d 244, 252 n.6 (2d Cir. 2021) (noting that even when the Ortiz rule applies, a party may preserve an issue for appellate review by “challeng[ing] the jury instructions that raise the same legal question”).22 For these reasons, we may review the district court’s ruling that the EMTplaintiffs’ post-shift equipment exchange was not de minimis.23 B On the merits, we decline to order a new trial. The City argues that one of the EMTs’ post-shift tasks (exchanging equipment with the oncoming tour) was de minimis. We need not decide whether the district court was correct to disagree at summary judgment, because the City’s argument (and, derivatively, the district court’s resolution) rests on the faulty premise that each of the plaintiffs’ post-shift tasks should be considered separately for purposes of the de minimis inquiry. However, an employer may not disaggregate required work into constituent tasks and then avoid FLSA liability because each piece is de minimis. And because plaintiffs would have been entitled to a ruling as a matter of law on a properly framed de minimis question, remand is inappropriate. The FLSA requires compensation for all work an employer suffers or permits; but what counts as compensable time must account for “the realities of the industrial world.” Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680, 692 (1946). As Anderson explained, “[w]hen the matter in issue concerns only a few seconds or minutes of work beyond the scheduled working hours, such trifles may be disregarded.” Id. “It is only when an employee is required to give up a substantial measure of his time and effort that compensable working time is involved”; the FLSA is not concerned with “[s]plit-second absurdities.” Id. Whether “the matter in issue” is de minimis, however, depends on what the matter in issue is. By singling out a specific task — EMTs’ post-shift equipment exchange — the City implicitly argues that the “matter in issue” is each identifiable task the plaintiffs perform before or after their shifts. Such an approach is incompatible with the statute and with our cases. The de minimis doctrine is a limited, judicially-created exception to the FLSA’s fundamental rule that employees must be paid for their work. The focus is on avoiding “[s]plit-second absurdities” and claims for “trifles.” Anderson, 328 U.S. at 692. The Department of Labor has warned against potential abuse of the de minimis doctrine: although “insubstantial or insignificant periods of time…may be disregarded,” “[a]n employer may not arbitrarily fail to count as hours worked any part, however small, of the employee’s fixed or regular working time or practically ascertainable period of time he is regularly required to spend on duties assigned to him.” 29 C.F.R. §785.47. So, once an employee “is required to give up a substantial measure of his time and effort,” “compensable working time is involved” and the claim is not de minimis. Anderson, 328 U.S. at 692. The City’s approach would make the de minimis doctrine a loophole that compromises the FLSA’s mission “to guarantee[] compensation for all work or employment engaged in by employees covered by the Act.” Tenn. Coal, Iron & R.R. Co. v. Muscoda Loc. No. 123, 321 U.S. 590, 602 (1944). “[T]o give the de minimis rule too broad a reach would contradict congressional intent by denying proper effect to a statute that is ‘remedial and humanitarian in purpose.’” Perez v. Mountaire Farms, Inc., 650 F.3d 350, 378 (4th Cir. 2011) (Wilkinson, J., concurring) (quoting Tenn. Coal, 321 U.S. at 597). Our view finds support in the cases. In Reich v. New York City Transit Authority, 45 F.3d 646 (2d Cir. 1995), we considered whether transit police officers were entitled to overtime pay for their commutes because they were required to transport their canine partners to and from work. After holding that only time spent actively caring for the dog — discipline, cleaning up, water stops, etc. — qualified as compensable work, 45 F.3d at 650-52, we considered whether that time was de minimis. In doing so, we asked whether the K-9 tasks were de minimis taken together, not separately: the “matter in issue” was “the time spent by the handlers engaged in active duties during the commute.” Id. at 652. Although we assessed each task’s frequency and duration,24 our holding was that “dog-care duties during the commute” in total were de minimis. Id. at 652-53. The Fourth Circuit rejected the City’s approach even more directly in Perez v. Mountaire Farms, in which the defendant asked the court to “evaluate each task or group of tasks separately to determine if the time period is de minimis.” 650 F.3d at 373. The Fourth Circuit declined: In applying the de minimis rule, we consider the aggregate amount of time for which the employees are otherwise legally entitled to compensation…. Adopting [the defendant's] approach would undermine the purpose of the FLSA by allowing employers to parcel work into small groups of tasks that, when viewed separately, always would be considered de minimis. Id. (internal citation omitted). We entirely agree. To be sure, when a heterogeneous group of plaintiffs bring FLSA claims based on multifarious required tasks and the defendant argues for application of the de minimis doctrine, it may be appropriate to subdivide the claimed work in some way. The “rule” is one of common sense, informed by the general principles we have set out above. Here, although the district court’s partial grant of summary judgment was premised on the erroneous piecemeal approach advanced by the City, remand is unnecessary. Even under the narrowest permissible definition of the matter in issue — EMTs’ post-shift work — plaintiffs would have been entitled to a ruling as a matter of law that such work is not de minimis. (We draw the line there because it is favorable to the City and in keeping with this case’s consistent pre-vs. postshift distinction.) When assessing whether otherwise-compensable time should be considered de minimis, we consider: “(1) the practical administrative difficulty of recording additional time; (2) the size of the claim in the aggregate; and (3) whether the claimants performed the work on a regular basis.” Singh v. City of New York, 524 F.3d 361, 371 (2d Cir. 2008) (citing Reich, 45 F.3d at 652). The first and third factors seem more important in this case, as they correlate with whether the time in question “cannot as a practical administrative matter be precisely recorded” and thus whether “the failure to count such time is due to considerations justified by industrial realities.” 29 C.F.R. §785.47; see also Anderson, 328 U.S. at 692. While the amount of time at issue is relevant, the fact that a task is of short duration does not necessarily render it noncompensable if the employer could still easily record and pay for it. See 29 C.F.R. §785.47. Here, each factor weighs against deeming plaintiffs’ post-shift work de minimis. First, post-shift work was very easy to record: CityTime already does, recording to the minute each post-shift sliver an EMT or paramedic spends at the station. Recording any and all work plaintiffs perform post-shift is easier than accounting for the sporadic dog-care activities the Reich plaintiffs might have needed to do; it is also far easier than tracking how much additional time inspectors spent commuting due to a requirement that they carry work materials to and from home, work we deemed de minimis in Singh, 524 F.3d at 370-72. See also Peterson v. Nelnet Diversified Sols., LLC, 15 F.4th 1033, 1045 (10th Cir. 2021) (finding that this factor favored plaintiffs because “[defendant] already monitors the actual time that the [plaintiffs] devote to booting up their computers and launching software [i.e., the uncompensated work at issue]“). Second, the size of the claim favors plaintiffs. The City focuses exclusively on how much time the claimed work takes per day, but the proper inquiry is the amount of time claimed “in the aggregate.” Singh, 524 F.3d at 371 (emphasis added). This of course includes the time spent per day, but as the Ninth Circuit explained when creating this three-factor inquiry, “[c]ourts have granted relief for claims that might have been minimal on a daily basis but, when aggregated, amounted to a substantial claim.” Lindow v. United States, 738 F.2d 1057, 1063 (9th Cir. 1984). That makes sense: an employee made to work a few minutes late every day for months has been “required to give up a substantial measure of his time and effort,” Anderson, 328 U.S. at 692, just as has an employee made to work several extra hours in a single day. “We would promote capricious and unfair results, for example, by compensating one worker $50 for one week’s work while denying the same relief to another worker who has earned $1 a week for 50 weeks.” Lindow, 738 F.2d at 1063 (citing Addison v. Huron Stevedoring Corp., 204 F.2d 88, 95 (2d Cir. 1953)). And in this case, the recovery period spans years. Even a handful of minutes on most days adds up to thousands of minutes of required post-shift work for each plaintiff.25 Finally, for many of the reasons explained in Part II, plaintiffs’ post-shift work occurred regularly — the tasks had to be performed every day. The City posits a few situations in which these tasks could be performed during, rather than after, a shift, but the question is whether the work is regular and predictable versus sporadic and occasional. The frequency and regularity with which plaintiffs performed post-shift work contrasts with the Reich plaintiffs’ unpredictable and irregular instances of K-9 care. For these reasons, no jury could conclude that EMT-plaintiffs’ post-shift work is de minimis and thus non-compensable. We will not therefore remand for a new trial on that question. * * * For the reasons set forth above, we reject each of the City’s arguments. The judgment is AFFIRMED.
