POSNER, Circuit Judge.

The three appellants are the named plaintiffs in a class action suit (actually a set of such suits) to enforce the Fair Labor Standards Act and parallel state laws governing minimum wages and overtime. Only the supplemental state law claims were brought as class actions, all under Rule 23(b)(3) of the Federal Rules of Civil Procedure; the claim under the FLSA is a "collective action" governed by section 16(b) of that Act, 29 U.S.C. § 216(b), which allows an employee to bring suit on behalf of all "similarly situated" employees. The relief sought, both under the collective action and the class actions, is damages.

The only difference of moment between the two types of action is that in a collective action the members of the class (of the "collective") must opt into the suit to be bound by the judgment or settlement in it, while in a class action governed by Rule 23(b)(3) (a class action seeking damages) they must opt out not to be bound. That difference can have consequences, the obvious one being the need to protect the right of Rule 23(b)(3) class members to opt out. But none of the consequences bears on this case. Indeed, despite the difference between a collective action and a class action and the absence from the collective-action section of the Fair Labor Standards Act of the kind of detailed procedural provisions found in Rule 23, see, e.g., O'Brien v. Ed Donnelly Enterprises, Inc., 575 F.3d 567, 584 (6th Cir.2009), there isn't a good reason to have different standards for the certification of the two different types of action, and the case law has largely merged the standards, though with some terminological differences. See Alvarez v. City of Chicago, 605 F.3d 445, 449 (7th Cir.2010); Thiessen v. General Electric Capital Corp., 267 F.3d 1095, 1105 (10th Cir.2001); Shushan v. University of Colorado, 132 F.R.D. 263, 265-68 (D.Colo. 1990); but see O'Brien v. Ed Donnelly Enterprises, Inc., supra, 575 F.3d at 584-86; Grayson v. K Mart Corp., 79 F.3d 1086, 1096 (11th Cir.1996); see generally 7B Charles Alan Wright, Arthur R. Miller & Mary K. Kane, Federal Practice and Procedure § 1807, pp. 477-85 (3d ed.2005). Simplification is desirable in law, especially in the present context, because joining a collective action and a class action or actions in one suit, as in this case, is both common and, we have held, permissible. Ervin v. OS Restaurant Services, Inc., 632 F.3d 971, 974 (7th Cir.2011).

It is true that one function of the procedural provisions in Rule 23 is to protect the rights of unnamed class members, who need such protection because unless they are permitted to and do opt out of the class they will be bound by the judgment or settlement. In contrast, collective actions bind only opt-ins. But the provisions of Rule 23 are intended to promote efficiency as well, American Pipe & Construction Co. v. Utah, 414 U.S. 538, 555-56, 94 S.Ct. 756, 38 L.Ed.2d 713 (1974); Committee Note to Subdivision (b)(3) of 1966 Amendment to Rule 23, and in that regard are as relevant to collective actions as to class actions. And so we can, with no distortion of our analysis, treat the entire set of suits before us as if it were a single class action.

The district judge certified several subclasses but later decertified all of them, leaving the case to proceed as an individual lawsuit by the three plaintiffs — who then settled with the defendants, and so the suit was dismissed. But the settlement reserved the plaintiffs' right to appeal the decertification, and they have appealed, and in a ruling last August, reported at 688 F.3d 872, we rejected the defendants' motion to dismiss the appeal. The defendants had argued that the settling plaintiffs had suffered no injury as a result of the denial of certification and so the federal judiciary had lost jurisdiction of the case. We rejected their argument. The plaintiffs had sought an incentive award (also known as an "enhancement fee") in the district court for their services as class representatives, and the award was contingent on certification of the class. We held that the prospect of the award if the class was certified gave them a tangible financial stake in getting the denial of class certification revoked, and so entitled them to appeal that denial. We now decide the appeal.

The class (remember that we're treating the FLSA "collective" and the Rule 23 classes as a single class) consists of 2341 technicians employed by defendant DirectSat to install and repair home satellite dishes (we can ignore the other defendant, DirectSat's parent). These technicians are more like independent contractors than employees; they spend the work day installing and repairing satellite equipment at customers' homes and are paid on a piece-rate basis — so many dollars per job — rather than being paid a fixed hourly wage. Nevertheless they are within the Act's broad definition of (covered) employees. See 29 U.S.C. § 203(e), (g); United States v. Rosenwasser, 323 U.S. 360, 361-64, 65 S.Ct. 295, 89 L.Ed. 301 (1945); Secretary of Labor v. Lauritzen, 835 F.2d 1529, 1536-38 (7th Cir.1987); Brock v. Superior Care, Inc., 840 F.2d 1054, 1060 (2d Cir.1988). They therefore are entitled to be paid the federal minimum wage for every hour they work, and 1.5 times their regular hourly wage for every hour they work in excess of 40 hours a week, although the parties disagree over how to calculate that regular hourly wage. See 29 C.F.R. § 778.318.

The suit alleges that management compelled the technicians to do work for which they were not compensated at all, and also to work more than 40 hours a week without being paid overtime for the additional hours. For purposes of deciding the appeal we assume (of course without deciding) that at a trial the plaintiffs could prove that DirectSat's policies violated the Fair Labor Standards Act in these ways. Nevertheless the district judge decertified the class (more precisely, the subclasses into which she had divided the class) when it became apparent that the trial plan submitted by the plaintiffs was infeasible. The plaintiffs then settled. So unless we reverse the decertification order, the case is dead.

There would have been no problem had the plaintiffs been seeking just injunctive or declaratory relief, because then the only issue would have been whether DirectSat had acted unlawfully. But the plaintiffs didn't seek either form of relief (their lawyer says that DirectSat has desisted from its principal violation, though that would not be a defense to a declaratory or injunctive action, United States v. W.T. Grant Co., 345 U.S. 629, 632-33, 73 S.Ct. 894, 97 L.Ed. 1303 (1953); Friends of the Earth, Inc. v. Laidlaw Environmental Services (TOC), Inc., 528 U.S. 167, 189-90, 120 S.Ct. 693, 145 L.Ed.2d 610 (2000)) — only damages. And to determine damages would, it turns out, require 2341 separate evidentiary hearings, which might swamp the Western District of Wisconsin with its two district judges. For it's not as if each technician worked from 8 a.m. to 5 p.m. and was forbidden to take a lunch break and so worked a 45-hour week (unless he missed one or more days because of illness or some other reason) but was paid no overtime. Then each technician's damages could be computed effortlessly, mechanically, from the number of days he worked each week and his hourly wage. And when "it appear[s] that the calculation of monetary relief will be mechanical, formulaic, a task not for a trier of fact but for a computer program, so that there is no need for notice ..., the district court can award that relief without terminating the class action and leaving the class members to their own devices." Johnson v. Meriter Health Services Employee Retirement Plan, 702 F.3d 364, 372 (7th Cir.2012). Nothing like that is possible here.

The suit charges that DirectSat concealed its violations by forbidding the technicians to record time spent on certain tasks, such as calling customers, filling out paperwork, and picking up tools from one of the company's warehouses. Remember that the technicians are paid on a piece-rate system, which implies — since workers differ in their effort and efficiency — that some, maybe many, of the technicians may not work more than 40 hours a week and may even work fewer hours; others may work more than 40 hours a week. Variance would also result from different technicians' doing different tasks, since it's contended that the employer told them not to report time spent on some of those tasks, though — further complicating the problem of proof — some of them reported that time anyway.

The plaintiffs proposed to get around the problem of variance by presenting testimony at trial from 42 "representative" members of the class. Class counsel has not explained in his briefs, and was unable to explain to us at the oral argument though pressed repeatedly, how these "representatives" were chosen — whether for example they were volunteers, or perhaps selected by class counsel after extensive interviews and hand picked to magnify the damages sought by the class. There is no suggestion that sampling methods used in statistical analysis were employed to create a random sample of class members to be the witnesses, or more precisely random samples, each one composed of victims of a particular type of alleged violation.

And even if the 42, though not a random sample, turned out by pure happenstance to be representative in the sense that the number of hours they worked per week on average when they should have been paid (or paid more) but were not was equal to the average number of hours of the entire class, this would not enable the damages of any members of the class other than the 42 to be calculated. To extrapolate from the experience of the 42 to that of the 2341 would require that all 2341 have done roughly the same amount of work, including the same amount of overtime work, and had been paid the same wage. See Reich v. Southern Maryland Hospital, Inc., 43 F.3d 949, 952 (4th Cir.1995); Secretary of Labor v. DeSisto, 929 F.2d 789, 793 (1st Cir.1991). No one thinks there was such uniformity. And if for example the average number of overtime hours per class member per week was 5, then awarding 5 × 1.5 × hourly wage to a class member who had only 1 hour of overtime would confer a windfall on him, while awarding the same amount of damages to a class member who had 10 hours of overtime would (assuming the same hourly wage) undercompensate him by half. The differences would not be trivial, because the technicians' average hourly rate was about $15.

That's just the beginning of the difficulty of computing the damages of each class member. Consider the complication created by the piece-rate system. When one is paid by the job rather than by the hour, the hourly wage varies from job to job and worker to worker. Suppose the same job, for which the piece rate is $500, is completed by one worker in 30 hours (and it is the only job he does that week) and by another in 60 hours (also in a week). The hourly wage of the first worker, $16.67, will be well above the minimum wage, and as he has no overtime either, he has no damages. The hourly wage of the second worker, $8.33, is also above the minimum wage, but he is entitled to 1.5 times that amount for 20 of his 60 hours and thus to a total wage of $583.10 (($8.33 × 40) + ($8.33 × 1.5 × 20)), making his damages $83.10 if Direct-Sat paid him only $500 for the job. The plaintiffs have not indicated how their method of "representative" proof would enable these workers to be separated when it came time to calculate damages.

Consider the further complication presented by a worker who underreported his time, but did so, DirectSat offers to prove, not under pressure by DirectSat but because he wanted to impress the company with his efficiency in the hope of obtaining a promotion or maybe a better job elsewhere — or just to avoid being laid off. The plaintiffs have not explained how their representative proof would distinguish such benign underreporting from unlawful conduct by DirectSat.

A further complication is that the technicians have no records of the amount of time they worked but didn't report on their time sheets. They are not like lawyers, who record every bit of work they do for a client, in 6-minute segments. The plaintiffs claim that their records are incomplete because DirectSat told them not to report all their time. But this if true does not excuse them from having to establish the amount of the unreported time. The "representative" proof they have submitted does not do this. The unreported time for each employee could be reconstructed from memory, inferred from the particulars of the jobs the technicians did, or estimated in other ways — any method that enables the trier of fact to draw a "just and reasonable inference" concerning the amount of time the employee had worked would suffice. Urnikis-Negro v. American Family Property Services, 616 F.3d 665, 669 and n. 2 (7th Cir.2010); Brown v. Family Dollar Stores of Indiana, LP, 534 F.3d 593, 595 (7th Cir. 2008); see Anderson v. Mt. Clemens Pottery Co., 328 U.S. 680, 687, 66 S.Ct. 1187, 90 L.Ed. 1515 (1946). But what can't support an inference about the work time of thousands of workers is evidence of the experience of a small, unrepresentative sample of them.

With no genuinely representative evidence having been suggested by class counsel, 2341 separate hearings loomed even if the district judge bifurcated the proceedings — that is, scheduled separate, successive trials on liability and damages, a common practice in complex litigation. Bifurcation would not eliminate variance in damages across class members, but once liability is established damages claims can usually be settled with the aid of a special master, and trials thus avoided, Carnegie v. Household International, Inc., 376 F.3d 656, 661 (7th Cir.2004); In re Visa Check/MasterMoney Antitrust Litigation, 280 F.3d 124, 140-41 (2d Cir.2001), as in such cases as Jenkins v. Raymark Industries, Inc., 782 F.2d 468, 471 (5th Cir.1986), and New York City Asbestos Litigation, 142 F.R.D. 60 (E. & S.D.N.Y.1992).

The plaintiffs claimed that DirectSat's management had committed three separate types of violation of the Fair Labor Standards Act and its state-law counter-parts. The district judge proposed therefore not only to bifurcate liability and damages but also, for purposes of determining liability, to divide the class into three subclasses (actually more, because of the different state laws applicable to the class actions, as distinct from the collective action, which of course was governed by federal law; but we'll ignore this wrinkle), corresponding to the three types of violation:

The judge asked the plaintiffs' lawyer to propose a specific plan for litigating the case within the framework she had established. This was a reasonable request, Vega v. T-Mobile USA, Inc., 564 F.3d 1256, 1278-79 and n. 20 (11th Cir.2009); Zinser v. Accufix Research Institute, Inc., 253 F.3d 1180, 1189 (9th Cir.2001); 2 William B. Rubenstein, Newberg on Class Actions § 4:79, pp. 309-17 (5th ed.2012), given the difficulty of trying a class action.

The plaintiffs responded rather truculently by opposing bifurcation and subclasses and refusing to suggest a feasible alternative, including a feasible method of determining damages. Eventually they expressed grudging acquiescence in dividing the class into subclasses but insisted that all the technicians were in all the subclasses and that their unrepresentative "representative" witnesses could therefore testify about the violations challenged by all three subclasses. Yet they acknowledged that it would "be difficult for Plaintiffs to provide an objective framework for identifying each class member within the current class definitions without making individualized findings of liability."

They continue on appeal to labor under the misapprehension that testimony by 42 unrepresentative "representative" witnesses, supplemented by other kinds of evidence that they have been unable to specify, would enable a rational determination of each class member's damages. They must think that like most class action suits this one would not be tried — that if we ordered a class or classes certified, DirectSat would settle. That may be a realistic conjecture, but class counsel cannot be permitted to force settlement by refusing to agree to a reasonable method of trial should settlement negotiations fail. Essentially they asked the district judge to embark on a shapeless, freewheeling trial that would combine liability and damages and would be virtually evidence-free so far as damages were concerned.

Although each class member claims to have lost several thousand dollars as a result of DirectSat's alleged violations, that isn't enough to finance a modern federal lawsuit; and in such a case, where it is class treatment or nothing, the district court must carefully explore the possible ways of overcoming problems in calculating individual damages. Yet there may be no way if for example there are millions of class members each harmed to a different extent (and many not harmed at all). See, e.g., In re New Motor Vehicles Canadian Export Antitrust Litigation, 522 F.3d 6, 28 (1st Cir.2008). In that event the suit may have to be limited to injunctive relief, or nonlitigation remedies explored. This is not so extreme a case; but if class counsel is incapable of proposing a feasible litigation plan though asked to do so, the judge's duty is at an end. And that's what happened.

The plaintiffs, or rather (to be realistic) class counsel, have overlooked a promising alternative to class action treatment in a case such as this. That is to complain to the Department of Labor, which enforces the Fair Labor Standards Act and can obtain in a suit under the Act the same monetary relief for the class members that they could obtain in a class action suit were one feasible. See 29 U.S.C. § 216(c), (e); 2 The Fair Labor Standards Act § 17.V.E, p. 17-117 (Ellen C. Kearns ed., 2d ed.2010). The Department has brought large numbers of FLSA enforcement actions, and is aided in doing so by being able to supplement employees' testimony with the results of its own investigation of the employer. Laurent Badoux, "Trends in Wage and Hour Litigation Over Unpaid Work Time and the Precautions Employers Should Take" 1-2, www.adp.com/workforce-management/docs/whitepaper/trendsinwageandhourlitigation_05292012.pdf; Mark Landler & Steven Greenhouse, "Solis Stepping Down as Labor Secretary," New York Times, Jan. 9, 2013, www.nytimes.com/2013/01/10/business/solis-stepping-down-as-labor-secretary.html (both visited Jan. 11, 2013); Solis v. Best Miracle Corp., 709 F.Supp.2d 843, 852 (C.D.Cal.2010). Badoux reports that "in fiscal year 2011, for instance, the [Department of Labor] recovered $225 million in back wages for employees, up 28% from fiscal year 2010, and the largest amount collected in a single fiscal year in [the Wage and Hour Division's] history." We were unable at the oral argument to elicit from class counsel an explanation for why he had not considered this route.

Finally, the plaintiffs challenge a ruling by the district court barring them from raising certain additional claims. Since they settled with the defendants — their sole interest in this appeal being, as we said at the outset, the chance of an incentive award if the class is certified-and we have decided that the district court's decertification ruling was correct, their objection to the judge's substantive ruling is moot.

AFFIRMED.