CARNES, Circuit Judge:

We deny Farm Stores Grocery, Inc.'s petition for rehearing but withdraw our previous opinion dated January 28, 2008, and published at 514 F.3d 1207, and substitute the following opinion in its place:

This appeal in a Fair Labor Standards Act, 29 U.S.C. §§ 201-19, case involves a number of issues arising under that statute. It also presents us with an interesting issue that is not FLSA-specific. Without objection from either party, the district court gave the jury an erroneous instruction on how to calculate damages. The jury compounded the error by returning a verdict for a larger amount of damages than the erroneous instruction would permit. Is the correct remedy a remittitur, reducing the damages down to the maximum amount that could have been awarded under the erroneous but unobjected to instruction, or a new trial with a proper instruction on calculating damages?

I.

Farm Stores operates a chain of 103 full-service, free-standing, drive-through grocery stores throughout Florida. Each store employs between three and six workers. One worker at each store is given the title "store manager" and is paid a weekly salary, unlike the others who are "sales associates" and are paid an hourly wage. Each store manager reports to a "district manager" who supervises between ten and twenty stores. Twenty-six of the twenty-eight plaintiffs are former store managers who were terminated during a company-wide reorganization in 2002.

After they were terminated for reasons unrelated to hours and wages, the store managers brought this FLSA lawsuit in federal district court. Their complaint sought back pay for overtime they claim Farm Stores owes them pursuant to 29 U.S.C. § 207(a)(1), which requires that employers pay time and a half for each hour an employee works beyond forty each week. Generally, employees may only recover up to two years of back pay under the FLSA's statute of limitations. 29 U.S.C. § 255(a). However, the complaint alleged that Farm Stores' violation of the FLSA was willful, which would entitle the plaintiffs to as much as one additional year of back pay.

Farm Stores' answer asserted that the store managers were exempt from the FLSA's overtime provision because they fell within the Act's "executive exemption." That exemption applies to employees who earn a salary of at least $250 a week, whose primary duty is management of a recognized department or subdivision, and who regularly direct two or more employees. 29 C.F.R. § 541.1(f) (2002). Farm Stores also asserted that it had not willfully violated the FLSA.

After the close of discovery, both Farm Stores and the store managers moved for summary judgment on the issue of whether the store managers were covered by the executive exemption. The district court denied both parties' motions, explaining that: "It is impossible to grant summary judgment for either party because the record does reveal very hotly disputed facts concerning what [the store managers] did in their day-to-day duties as `store managers,' and whether those activities bring them within the relevant exemptions as a matter of law." The case then proceeded to a jury trial.

The main issue at trial was whether the store managers fell within the executive exemption. Both sides presented evidence on the question of whether the store managers' primary duty was management. On the one hand, Farm Stores introduced evidence that the store managers interviewed, hired, trained, evaluated, and disciplined employees; maintained store inventory; and were relatively free from supervision of their daily activities. Additionally, its expert witness testified that, based on his review of the store managers' job descriptions, their primary duty was management.

On the other hand, the store managers testified that their primary duties were sales related, not managerial. They explained that they spent almost no time performing managerial tasks during most weeks, that they lacked real authority over their stores and employees, and that they were required to consult their district managers before making management decisions. They also testified that their hourly rate of pay, calculated by dividing their weekly salary by the total number of hours they worked each week, was essentially the same as the hourly rate of pay for sales associates. Their expert witness testified that, based on his review of affidavits from the store managers, as well as their job descriptions, their primary duty was not management.

On the issue of damages, each store manager also testified about the average number of hours he worked each week and the amount of his weekly salary from 1999 to 2002. Their counsel asked each store manager what his regular rate of pay was and asked most of them how they calculated it. The store managers were virtually unanimous in testifying that they calculated their regular rate of pay by dividing their weekly salary by the total number of hours they worked each week. This method is the one that counsel for the store managers used to explain and argue damages to the jury in his opening statement and closing argument. It is also the way the district court instructed the jury to calculate damages. Neither side asked for or objected to the instruction.

The jury returned a verdict finding that the store managers did not fall within the executive exemption and awarding damages to each individual store manager. The jury also found that Farm Stores' violation of the FLSA overtime provision was not willful, a finding which had the effect of limiting the store managers to a maximum of two years of back pay. The total amount of damages the jury awarded was $297,700.00.

Farm Stores filed two post-trial motions. In the first one it sought judgment as a matter of law or, in the alternative, a new trial, under Federal Rule of Civil Procedure 50 on the ground that the jury's finding on the executive exemption was not supported by the evidence. In the second motion Farm Stores sought a remittitur of the verdict because the damages the jury awarded exceeded the amount established by the evidence. The district court summarily denied both of those motions.

The store managers also filed a post-trial motion, seeking an award of liquidated damages equal to the amount of the jury award under 29 U.S.C. § 216. The district court granted that motion. This is Farm Stores' appeal.

II.

Farm Stores first contends that the district court erred in denying its Rule 50 motion because store managers, at least those who are assigned to separate locations, categorically fall within the executive exemption and therefore are not entitled to overtime under the FLSA. Alternatively, it contends that the evidence presented at trial is insufficient to sustain the jury's finding regarding that exemption. The store managers counter that the executive exemption is an inherently fact-based inquiry that turns on the individual circumstances of each case, and that they presented sufficient evidence for the jury reasonably to conclude that these specific store managers were not exempt employees.

Generally, the FLSA requires employers to pay their employees time and a half for all the work they do over forty hours a week, 29 U.S.C. § 207(a)(1), but that requirement does not apply to "any employee employed in a bona fide executive . . . capacity," id. § 213(a)(1). The Department of Labor has promulgated regulations defining who is an "executive" for purposes of the FLSA. Under the applicable regulation, an employee is an "executive" if he earns at least $250.00 per week and: (1) is paid on a salary basis; (2) manages, as his primary duty, a recognized department or subdivision; and (3) regularly directs two or more employees. 29 C.F.R. § 541.1(f) (2002).

Before trial, the parties stipulated that the store managers were paid a salary of at least $250.00 per week. After hearing the evidence, the jury found that the store managers "customarily and regularly directed the work of two (2) or more employees at his or her store," and the store managers do not contest that finding. Accordingly, the overtime wages issue turns on whether the store managers' primary duty was management. The jury found that it was not, but Farm Stores argues that finding is not supported by the evidence.

The sufficiency of the evidence on that issue must be viewed against the applicable DOL regulation, which sets out five factors to use in determining whether an employee's primary duty is management. The factors are: (1) the amount of time spent performing managerial duties; (2) the relative importance of an employee's managerial and non-managerial duties; (3) the frequency with which an employee may exercise discretionary powers; (4) the employee's relative freedom from supervision; and (5) the relationship between the purportedly exempt employee's wages and the wages paid to other employees performing similar, non-exempt work. 29 C.F.R. § 541.103 (2002).

In the face of the regulation, Farm Stores insists that the store managers are exempt employees "as a matter of law." What we understand that to mean is that any employee who has the responsibility over a free-standing business location, as these plaintiffs did, must come within the executive exemption. Farm Stores cites a number of decisions where courts concluded that managers of free-standing business units fell within the executive exemption, but in each of those cases the court reached that conclusion only after thoroughly reviewing the specifics of each employee's duties. None of the courts simply slapped on a talismanic phrase.

Take, for example, Baldwin v. Trailer Inns, Inc., 266 F.3d 1104 (9th Cir. 2001), which affirmed the district court's grant of summary judgment in favor of an employer, concluding that the managers in question fell within the executive exemption. Id. at 1117. In analyzing the primary duty requirement, the Baldwin court discussed in detail the evidence concerning the amount of time the managers spent performing managerial duties, the importance of those duties, the managers' independence and discretionary authority, and the difference between the managers' salaries and the wages paid to other employees. Id. at 1113-16. The Baldwin opinion illustrates the necessarily fact-intensive nature of the primary duty inquiry. That particularized approach is consistent with the DOL regulation, which provides that the "determination of whether an employee has management as his primary duty must be based on all the facts in a particular case." 29 C.F.R. § 541.103 (2002). When it comes to deciding whether an employee is an executive within the meaning of the FLSA, the answer is in the details. We reject Farm Stores' categorical approach.

Farm Stores' fallback argument is that even if the exemption issue is one to be decided on the particulars of the evidence, in this case there was not enough evidence to support the jury's finding that the store managers were not exempt employees. This argument primarily focuses on the evidence from which the jury could have found in Farm Stores' favor on this issue. We agree with Farm Stores that it presented abundant documentary evidence and testimony at trial indicating that the store managers' primary duty was management. We would affirm a jury verdict in that direction, but that is not what we have.

Where an issue turns on the particular facts and circumstances of a case, it is not unusual for there to be evidence on both sides of the question, with the result hanging in the balance. The result reached must be left intact if there is evidence from which the decision maker, the jury in this instance, reasonably could have resolved the matter the way it did. See Pickett v. Tyson Fresh Meats, Inc., 420 F.3d 1272, 1278 (11th Cir. 2005); Collado v. United Parcel Serv., Co., 419 F.3d 1143, 1149 (11th Cir. 2005); Bogle v. Orange County Bd. of County Comm'rs, 162 F.3d 653, 659 (11th Cir. 1998). The issue is not whether the evidence was sufficient for Farm Stores to have won, but whether the evidence was sufficient for it to have lost. It was.

We won't recount all of the testimony and other evidence supporting the verdict, but a few examples will illustrate how the jury reasonably could have determined that the store managers' primary duty was not management. One store manager testified that in a typical week his time was allocated as follows: "[A]n average 30 percent as regards cleaning, 50 percent customer attention or service, 10 percent in merchandise receiving, another 10 percent approximately in stacking up the racks, but the main point was customer sales." From his testimony a jury reasonably could conclude that at least some of the store managers spent no time performing managerial duties during most weeks. While other managers conceded that they spent some time each week performing managerial tasks, all of them insisted it was not the majority of their work, and more than one testified to spending only about ten percent of their time on management-related duties.

The jury also heard testimony that the store managers lacked authority and discretion over their respective stores and employees. Some of the store managers testified that their stores were actually run by district managers. They told how employee evaluation forms that they were required to sign were actually filled out by district managers, and how they could not make any management decisions without first seeking permission from their district manager. Several store managers testified that their hourly rate of pay (computed by dividing their weekly salary by the number of hours they worked each week) was comparable to, and in one case actually less than, those of the sales associates or assistant managers at their stores.

Finally, the store managers' expert witness, after reviewing their job descriptions and affidavits, concluded with "a reasonable degree of professional certainty" that Farm Stores could not claim the store managers as exempt under the executive exemption because "their primary duty was that of being engaged in sales and sales related activities," not management.

Considering the evidence as a whole, and viewing it in the light most favorable to the verdict, there was enough for the jury reasonably to find, as it explicitly did, that the primary duty of the store managers was not management. The district court did not err in denying Farm Stores' post-verdict motion for judgment as a matter of law, or in the alternative, a new trial on this ground.

III.

The next issue arises from the district court's denial of Farm Stores' motion for a remittitur. The court instructed the jury that if it found that Farm Stores had violated the FLSA overtime provision, it must award back pay damages in the amount of the unpaid overtime. The court's instruction about how those damages should be calculated included this key passage:

The employee's "regular rate" during a particular week is the basis for calculating any overtime pay due to the employee for that week. The "regular rate" for a week is determined by dividing all of the hours worked into the total wages paid for those hours. The overtime rate, then, would be one-half of that rate and would be owing for each hour in excess of 40 hours worked during the work week.

Neither party objected to any part of that instruction, including the second sentence of it, which as we shall see, is problematic.

Farm Stores' chief argument for a remittitur is that, even viewing the evidence in the light most favorable to the store managers, the damages the jury calculated are nearly twice the amount the evidence supports if the formula laid out in the damages instruction is applied. If the jury had calculated each store manager's regular rate of pay, in the manner it was instructed, Farm Stores insists that the jury would have returned an aggregate damages award of no more than $158,281.66, instead of the $297,700.00 it did.

A.

As a general rule, "a remittitur order reducing a jury's award to the outer limit of the proof is the appropriate remedy where the jury's damage award exceeds the amount established by the evidence." Goldstein v. Manhattan Indus., Inc., 758 F.2d 1435, 1448 (11th Cir. 1985); see also Frederick v. Kirby Tankships, Inc., 205 F.3d 1277, 1284 (11th Cir. 2000) ("The rule in this circuit states that where a jury's determination of liability was not the product of undue passion or prejudice, we can order a remittitur to the maximum award the evidence can support.").

To determine whether the jury's award is within the range dictated by the evidence, we view the evidence in the light most favorable to the store managers and apply the district court's instructions, which were not objected to by either party. After independently reviewing the testimony and evidence presented at trial, we have no doubt that the jury's damages verdict far exceeds the maximum amount that could have been awarded based on the evidence and the instructions.

The jury's award to Miguel Lavastida illustrates the errors. Taking the evidence in the light most favorable to him, it established that he worked an average of 60 hours per week and that his weekly salary was $458. Both parties agree that, if Lavastida is owed any overtime, he is entitled to 78 weeks of it. Applying the formula from the district court's instruction, Lavastida's weekly salary of $458 should be divided by 60, which is the number of hours he actually worked. This yields a regular hourly rate of $7.63. His overtime rate would be $3.82, half his regular hourly rate. Because Lavastida worked 60 hours a week, he is entitled to 20 hours of overtime per week. And $3.82 multiplied by 20 is $76.40, which is the amount of overtime pay he would be entitled to each week. Therefore, Lavastida's total award should have been $5,959.20 ($76.40 multiplied by 78). The jury, however, awarded him $11,700, almost twice the amount he was entitled to under the jury instruction.

The jury's award to Claribel Altamirano provides another example. She testified that she worked on average 52 to 60 hours a week for a weekly salary of $430. The parties agree that if Altamirano is owed any overtime, she is entitled to 82 weeks of it. Applying the formula contained in the district court's instruction, Altamirano's weekly salary of $430 should be divided by some number of hours between 52 and 60. That would yield an hourly rate between $7.17 and $8.26 per hour. Her overtime rate would be between $3.59 and $4.13, half her regular hourly rate. She would be entitled to between 12 and 20 hours of overtime per week. The amount awarded to her in back pay, therefore, should have been between $4,063.92 and $5,887.60. The jury, however, returned a verdict of $12,350, which is more than three times the low-end of the possible range and more than double the high-end of it. We have tried to reproduce the jury's damages number by performing all of the calculations ourselves, but have not been able to do so.

The fact that the verdict exceeds the maximum amount under the calculation theories on which the case was tried is also evidenced by the closing argument of the store managers' own counsel. After discussing Farm Stores' liability and the method for calculating damages, he proceeded to list the amounts he thought the store managers were entitled to receive based on his view of what the evidence would support. The total of the damages awards that he advocated for his clients, the store managers, was only $197,310.79 — approximately $100,000 less than the $297,700 awarded by the jury. Thus the jury's total damages verdict substantially exceeds the amount that the store managers' own attorney thought they were entitled to receive.

The store managers have not even attempted to defend the jury's damages verdict to this Court using the instructions the jury was given. Instead, they contend that we should apply the applicable DOL interpretative bulletin, which contained a different formula for calculating the regular rate of pay. We will have more to say about that bulletin later. For the time being, it is enough to say that even assuming that we can apply the bulletin's formula, which was not applied at trial, the damages verdict still cannot stand. For nine of the twenty-six store managers, the jury's verdict exceeds by $1,000 or more their own calculations — contained in their brief to us — of what they are entitled to receive under the formula in the interpretative bulletin. To give the most extreme example, for one of the store managers, their calculations under that formula would result in an award of $307.50. The jury, however, returned a verdict of $11,700 for that manager — an award thirty-eight times greater than the managers themselves say should result from applying the bulletin's formula.

We recognize, of course, that the jury had to make approximations and estimates in order to arrive at a damages verdict because Farm Stores did not maintain payroll records documenting the number of hours each store manager worked. The jury was required, however, to operate within the bounds of the evidence presented at trial and to calculate damages using the formula they were instructed to apply. Each store manager testified to the average number of hours he worked each week and to his salary; sometimes they testified to ranges, such as 52 to 60 hours per week. In that and other ways the evidence gave the jury some flexibility in coming up with a dollar amount of damages for each store manager. Unfortunately, the jury's damages verdict ended up well outside the boundaries of the evidence. Whether the error resulted from disregarding the evidence, a mathematical mistake, confusion, or some other reason, it is still error. As our predecessor Court observed, we cannot "permit damage speculation where the formula for calculation is articulable and definable. Flexibility beyond the range of the evidence will not be tolerated." Jamison Co. v. Westvaco Corp., 526 F.2d 922, 936 (5th Cir. 1976). The damages award must be set aside.

B.

If the district court's jury instruction on calculating damages had accurately reflected the law, we could stop here. We would reverse the order denying Farm Stores' motion for a remittitur and remand the case with instructions that the district court reduce the award to the maximum amount established by the evidence. See Goldstein, 758 F.2d at 1448; see also Frederick, 205 F.3d at 1284. We cannot do that in this case, however, because of the erroneous instruction.

The district court instructed the jury that it should use the total number of hours the store managers worked in order to calculate their regular rate of pay. That is not the law. An interpretative bulletin promulgated by the DOL provides the correct method for calculating an employee's regular rate of pay. That interpretative bulletin specifies that:

If the employee is employed solely on a weekly salary basis, his regular hourly rate of pay, on which time and a half must be paid, is computed by dividing the salary by the number of hours which the salary is intended to compensate. If an employee is hired at a salary of $182.70 and if it is understood that this salary is compensation for a regular workweek of 35 hours, the employee's regular rate of pay is $182.70 divided by 35 hours, or $5.22 an hour, and when he works overtime he is entitled to receive $5.22 for each of the first 40 hours and $7.83 (one and one-half times $5.22) for each hour thereafter. If an employee is hired at a salary of $220.80 for a 40-hour week his regular rate is $5.52 an hour.

29 C.F.R. § 778.113(a) (2002). In the situation here, where the employee is paid solely on a weekly salary basis, the number of hours the employee's pay is intended to compensate — not necessarily the number of hours he actually works — is the divisor.

Farm Stores argues that we should instruct the district court on remand to order a remittitur notwithstanding the error in the jury instructions, because the store managers' weekly salary was intended to cover all of the hours they worked. If that is so, then the error in the instructions would have no effect, because the divisor and the result will be the same, anyway. The flaw in this argument is that it requires us to assume as a fact that the weekly salary was intended to compensate the store managers for however many hours they actually worked. The evidence at trial does not compel that finding, but conflicts.

The evidence on Farm Stores' side of this issue included the testimony of its chief operating officer that the company intended for its store managers to work as many hours as needed to fulfill their responsibilities, and for their salary to compensate them for as many hours as that took. On the other side, when asked how many hours store managers were required to work, one district manager testified: "48 hours." There was also testimony that the company required the store managers to put 40 hours on the time sheets they filled out each week. The number of hours the store managers' salaries were intended to compensate is an unresolved factual issue.

We are also unpersuaded by the store managers' argument that despite the error in the district court's jury instruction, and the fact that the jury was never asked to address the issue, we should find that it decided that the store managers' weekly salary was intended to compensate them for a forty-hour work week. The argument is that even though the court got the law wrong in its instructions, and the store managers' own attorney got the law wrong in his opening statement and closing argument, the jury somehow divined that they were to follow the DOL interpretative bulletin's approach even though it was never mentioned to them. We will not presume that the jury knows more about the law than the judge and lawyers do, especially where virtually all of the testimony was framed in terms of the erroneous standard, and all of the store managers' arguments reinforced it.

C.

Insisting that the jury's damages verdict, even if it cannot be explained under the district court's instruction, is consistent with the evidence and the DOL's interpretive bulletin, the store managers argue that we should apply the "right for the wrong reasons" principle and affirm the district court's order. At oral argument, counsel for the store managers directed us to a line of Florida cases applying that principle, which is also known by the delightful title of the "tipsy coachman" doctrine, to the review of judicial rulings and decisions. See Applegate v. Barnett Bank of Tallahassee, 377 So.2d 1150, 1152 (Fla. 1979) ("The written final judgment by the trial court could well be wrong in its reasoning, but the decision of the trial court is primarily what matters, not the reasoning used. Even when based on erroneous reasoning, a conclusion or decision of a trial court will generally be affirmed if the evidence or an alternative theory supports it."); see also Muhammad v. State, 782 So.2d 343, 359 (Fla. 2001) ("[T]he trial court's ruling on an evidentiary matter will be affirmed even if the trial court ruled for the wrong reasons, as long as the evidence or an alternative theory supports the ruling."). We are all for rules that promote judicial economy and efficiency, but even assuming that the doctrine can be applied to jury verdicts, it cannot save the one in this case. The record in this case reminds us less of a tipsy coachman arriving at the right destination than of a blind one who ends up at the wrong place.

As we have explained this is not a case where the jury, confused by an erroneous instruction, still drove the chariot to the right home. Nine of the twenty-six awards exceeded by $1,000 or more the maximum amount that the evidence would support if the correct rule of law had been applied. Where one-third of the awards are greater than even the store managers' best take on the evidence, the result cannot fairly be characterized as correct.

Attempting to push wrong closer to right, the store managers resourcefully point to the testimony of one of them stating that he actually worked an extra five hours each week beyond those he spent working at the store. That extra time, he explained, was spent making bank deposit runs. However, that was the testimony of only one of the store managers. The others did not say that the total hours they estimated working excluded any of the time they spent performing duties for the company. In any event, adding five hours to every store manager's work week would only reduce the degree of error in the jury's damages verdict, not remove it entirely. As counsel for the store managers conceded at oral argument, we cannot "affirmatively tell what [the jury's] reasoning was." What we do know is that it did not faithfully apply the district court's instruction — erroneous as it was — to the evidence. The chariot wound up at the wrong house. We can neither chart its course nor let stand its destination.

D.

Having concluded that the jury's verdict exceeds the outer bounds of the evidence under the district court's erroneous instruction, and under the correct rule of law as well, we arrive at the question we posed at the beginning of this opinion: Should we direct the district court to order a remittitur based on the incorrect jury instruction, thereby reducing the damages award to the maximum that could have been awarded under the erroneous instruction that was given without objection, or should we order a new trial on damages where the correct standard would be applied?

We have found no decision by this Court directly answering that question, but our decisions in cases where district courts have given unclear or confusing jury instructions provide some guidance. Consider, for example, Overseas Private Investment Corp. v. Metropolitan Dade County, 47 F.3d 1111 (11th Cir. 1995), where a jury found two defendants liable under breach of contract, negligence, and strict liability theories for supplying toxic fertilizer to a farming operation. Id. at 1115-16. Although the same damages should have been awarded under each theory, the jury returned a different damages verdict on the strict liability theory. Id. at 1115. After reviewing the verdicts, special interrogatories, and jury instructions, we concluded that "the jury instructions . . . were confusing and unclear, resulting in confusing damage awards." Id. at 1116. As is the case here, though, "the liability issues were properly and clearly decided by the jury." Id. Where liability was properly decided by the factfinder but an erroneous damages verdict was returned, we held that "the remedy . . . [was] to remand the case to the district court for a new trial on the amount of damages only." Id.

Similarly, in King v. Exxon Co., 618 F.2d 1111 (5th Cir. 1980), we concluded that the district court gave the jury an incorrect instruction for determining the period for which the plaintiff was entitled to damages under an employment contract. Id. at 1118-19. Because the correct instruction would have required the jury to determine an inherently fact-based question that it had not decided, we refused to order a remittitur. Id. at 1119. Instead, we remanded for a new trial on the damages question, directing the district court to properly instruct the jury about how to calculate the damages period. Id.

In both OPIC and King, we concluded that a remittitur was not an appropriate way to remedy the erroneous damages verdict because the jury's calculations had come after erroneous instructions. As in those cases, the jury's verdict in this one is excessive when measured against the evidence presented at trial — here it is excessive when measured against the erroneous instruction and against the correct rule of law. Whatever the cause of the damages verdict in this case, it is confusing and erroneous. The best way to remedy it is to start over on damages, give the jury the correct legal standard to apply, and hope for the best.

Because of the contemporaneous objection rule, we probably would have affirmed the judgment in this case if the only error were in the instruction — if the jury had correctly applied the erroneous standard to which no one objected. The twist here is that the jury did not correctly apply the wrong standard, but instead compounded the error. The jury in effect took the error to a different level, one beyond the level of error insulated by the contemporaneous objection rule. There was no failure to object to the erroneous application of the erroneous standard. Farm Stores promptly and properly raised that second level error in its motion for a remittitur and is entitled to have that issue decided, as we are doing.

We do not believe that our review of this second level error issue rewards either party for failing to object to the erroneous jury instruction, which was the first level error. Farm Stores came to us asking that we order a remittitur under the erroneous standard; we are not doing that. The store managers asked us to affirm the damages award that was returned; we are not doing that either. Both parties would have been better off if they had recognized the error in the jury instruction and objected to it, so that the damages part of the case could have been decided under the correct standard to begin with.

On remand, when it conducts a new trial on the issue of damages, the court should instruct the jury according to the regular rate of pay standard contained in the applicable DOL interpretive bulletin. See 29 C.F.R. § 778.113(a). Although we leave the final decision on this detail to the discretion of the experienced district court judge, he might wish to propound specific interrogatories to the jury that include these questions as to each store manager: the weekly salary, the number of hours that salary was intended to compensate, and the number of hours actually worked. If the district court decides to use special interrogatories, it would be free to phrase the questions and include any additional ones it believes relevant to deciding the amount of damages.

IV.

There is one issue left for us to decide. Farm Stores contends that the district court erred by awarding the store managers liquidated or "double" damages under 29 U.S.C. § 216(b). We review the district court's ultimate decision to award liquidated damages only for an abuse of discretion. See 29 U.S.C. § 260 ("[A district] court may, in its sound discretion, award no liquidated damages or award any amount thereof not to exceed the amount specified in section 216 of [the FLSA]."); Dybach v. Fla. Dep't of Corr., 942 F.2d 1562, 1566 (11th Cir. 1991) (explaining that whether an employer acted in good faith and with a reasonable belief that it was in compliance with the FLSA is a mixed question of law and fact, but that once an employer establishes these elements, "the district court's refusal to award liquidated damages is reviewed for abuse of discretion" (citations omitted)).

Under the FLSA a district court generally must award a plaintiff liquidated damages that are equal in amount to actual damages. The statute provides: "Any employer who violates the provisions of [the FLSA] . . . shall be liable to the . . . employees affected in the amount of their . . . unpaid overtime compensation . . . and in an additional equal amount as liquidated damages." 29 U.S.C. § 216(b). The Portal to Portal Act, 29 U.S.C. §§ 251-62, which amended the FLSA, provides a safe harbor for an employer who can establish that it acted in good faith and under the reasonable belief that it was in compliance with the FLSA. Id. § 260. The safe harbor provision provides:

In any action . . . to recover . . . unpaid overtime compensation, or liquidated damages, under the [FLSA] . . ., if the employer shows to the satisfaction of the court that the act or omission giving rise to such action was in good faith and that he had reasonable grounds for believing that his act or omission was not a violation . . . the court may, in its sound discretion, award no liquidated damages. . . .

Id. An employer who violates the FLSA's overtime provision carries the burden of proving its entitlement to the safe harbor. Joiner v. City of Macon, 814 F.2d 1537, 1539 (11th Cir. 1987). To satisfy the good faith requirement, an employer must show that it acted with both objective and subjective good faith. Dybach, 942 F.2d at 1566-67.

Farm Stores advances several arguments about why the district court abused its discretion in awarding liquidated damages. First, it contends that the court held it to an excessively high evidentiary standard. According to Farm Stores, to sail within the safe harbor all an employer must do is demonstrate good faith; it need not furnish documentary evidence of its good faith. Second, Farm Stores suggests that the district court order imposed a safe harbor precondition that employers consult with the DOL before declining to pay overtime, and asserts that there is no basis for requiring that.

We do not share Farm Stores' view of the district court's order. The court did not read a documentary evidence or DOL consultation requirement into the safe harbor. Instead, it reasoned that the absence of documentary evidence or testimony that Farm Stores consulted the DOL or an FLSA expert before or during the period of the violation weighed against a finding of objective good faith. That is self evident from the fact that documentary evidence or consultation with the DOL or an outside expert would have weighed in Farm Stores' favor. The court did not indicate that the only evidence that would persuade it of Farm Stores' good faith was documentary evidence or proof of consultation. The court's examples were not intended as an exhaustive list.

Farm Stores also contends that under the evidence that was presented, the district court erred by finding that it had not shown objective good faith. Farm Stores did present an expert (it had not consulted before the litigation began) who gave his opinion that the store managers fit within the executive exemption to the overtime requirement. That expert's testimony, however, was not uncontested. The store managers presented their own expert who reached the opposite conclusion. The district court was not required to credit Farm Stores' expert, even to the point of establishing a good faith difference of opinion. The district court heard the testimony of the experts, as well as that of the President of Farm Stores, on the issue of good faith, and it was in a much better position than we are to decide this fact-intensive issue and exercise its discretion accordingly

Finally, Farm Stores contends that in deciding the liquidated damages issue, the district court was barred from finding that the company had not acted in good faith by the jury's prior finding on the statute of limitations issue that it had not willfully violated the Act. This argument is two-fold: the absence of willfulness is inconsistent with the absence of good faith; and, where there is evidence on both sides of a factual issue, the court cannot make findings inconsistent with those of the jury.

The parties cite the decisions of three circuits showing a clear split over whether the standards for finding willfulness and for finding the absence of good faith are the same so that the jury's finding on the former issue controls the judge's finding on the latter one. Compare Brinkman v. Dep't of Corr., 21 F.3d 370, 373 (10th Cir. 1994) ("The same willfulness standard for the statute of limitations issue applies to the liquidated damages issue." (citation omitted)), with Broadus v. O.K. Indus., Inc., 226 F.3d 937, 944 (8th Cir. 2000) (noting that the "jury's decision on willfulness is distinct from the district judge's decision to award liquidated damages" (citation omitted)), and Fowler v. Land Mgmt. Groupe, Inc., 978 F.2d 158, 162 (4th Cir. 1992) ("[T]he explicit language of [the safe harbor provision] expressly vest[s] discretion to award liquidated damages in the hands of the trial judge. We do not believe that, in light of this clear delegation of authority, Congressional intent would be effectuated by a scheme in which, in every case, the trial court's discretion to award liquidated damages would be completely constrained by the jury's determination on `willfulness' for purposes of the statute of limitations.").

Interesting as the issue is, we don't have to pick a side in the circuit split in order to decide this appeal. The underlying concern of the Tenth Circuit in Brinkman, which affirmed a district court's finding that good faith was absent because the jury had found willfulness, was fostering consistency between the factfindings of judges and juries. Brinkman, 21 F.3d at 372-73. That concern is the same whether it is articulated as a Seventh Amendment problem, see id. ("We have held that when fact issues central to a claim are decided by a jury upon evidence that would justify its conclusion, the Seventh Amendment right to a jury trial prohibits the district court from reaching a contrary conclusion."), or as a collateral estoppel problem. However it is articulated, that concern is not implicated here.

We assume for present purposes that the absence of willfulness is equivalent to the presence of good faith. Even with that assumption, in the present case the jury's finding and the court's finding are not necessarily inconsistent. The jury's actual finding was that the store managers had failed to carry their burden of proving that Farm Stores had willfully violated the FLSA. The interrogatory it returned, accurately phrased in light of the law, stated that the store managers did not prove "by a preponderance of the evidence that [Farm Stores] knew it was violating the [FLSA] or showed reckless disregard as to whether it was violating the [FLSA] by treating its store managers as exempt for purposes of overtime." The court's actual finding, in light of the statute, is that Farm Stores did not carry its burden of proving that it had acted in good faith.

The reconciliation point is the burden of proof, and more specifically, the differences in its placement. For the willfulness issue on which the statute of limitations turns, the burden is on the employee; for the good faith issue on which liquidated damages turns, the burden is on the employer. Because the burden of proof is placed differently, a finding that willfulness was not present may co-exist peacefully with a finding that good faith was not present. The result varies with the burden of proof, provided that a factfinder could conclude that the evidence on the issue is evenly balanced. See 18 Charles Alan Wright, Arthur R. Miller Edward H. Cooper, Federal Practice and Procedure § 4422 (2d ed. 2002) (explaining that where the burden of proof varies, collateral estoppel does not necessarily apply because "it is possible that the initial determination rested on a conclusion that the evidence was in exact equipoise"); see also Steelmet, Inc. v. Caribe Towing Corp., 747 F.2d 689, 694 (11th Cir. 1984) (surveying decisions of other courts, academic commentary, and the Restatement (Second) of Judgments and concluding that "the change in allocation of the burden of proof makes a difference" as to the applicability of the doctrine of collateral estoppel).

In deciding whether a rational factfinder could view the evidence on the willfulness/good faith issues as evenly balanced, we bear in mind that it is the factfinder's function to resolve credibility questions, and we must view the evidence through the curative lens of resolving all conflicts in favor of the judgment. In this instance, that means resolving all conflicts toward equipoise. With that in mind, we conclude that acting rationally, a factfinder could have found the evidence evenly balanced.

Our analysis and the result we reach does not conflict with the Tenth Circuit's Brinkman decision. The conclusion in that case was that a finding by the judge that the employer had proven it acted in good faith would have been inconsistent with the jury's finding that the employees had proven the employer acted willfully. Brinkman, 21 F.3d at 372-73. In other words, a finding that a positive condition (good faith) had been proven would have been inconsistent with finding that the negative of the same condition (willfulness) had also been proven. We already have assumed, after all, that good faith and willfulness cannot coexist. By contrast, what we have here is the failure of Farm Stores to prove a positive condition (good faith). That is not inconsistent with the store managers' failure to prove the negative of that same condition (willfulness). It is not inconsistent, because if the evidentiary needle is pointing straight up, neither side has carried its burden of proof, nothing has been proven.

For these reasons, we reject Farm Stores' contrary argument that the judge's finding that it lacked good faith cannot stand.

V.

The district court's order denying Farm Stores' motion for judgment as a matter of law is AFFIRMED in all respects, except that the part of the judgments awarding damages in favor of the twenty-six store manager plaintiffs are REVERSED, and the case is REMANDED for a new trial on damages consistent with this opinion.