The district court granted a defense motion for summary judgment in this action for violations of the Equal Pay Act, 29 U.S.C. § 206(d). The questions presented on appeal are: (1) whether the plaintiff met her burden of showing that the defendant paid male employees more than it paid her for substantially equal work, and (2), if so, whether the defendant met its burden of showing that factors other than sex account for the disparity in pay.
Upon de novo review, we conclude, as did the district court, that a rational jury could find the plaintiffs work “equal” to that of her male colleagues. The plaintiffs *104job required performance of both higher-skilled and lower-skilled tasks. The fact that she was assigned more of the lower-skilled work than male employees with the same job title does not necessarily render her work unequal to theirs. It is clear, moreover, that the plaintiff was paid less than most male employees in the same job.
We also conclude, as did the district court, that the record compels a finding that sex played no part in the difference between the plaintiffs rate of pay and the rates at which male employees were paid. Unrebutted evidence shows that the defendant sets starting salaries on the basis of experience and “market conditions” and that it awards annual raises pursuant to a merit system. In our view, the record does not support an inference that sex is a factor in these compensation decisions. Accordingly, we shall affirm the grant of summary judgment.
I
The plaintiff, Lisa Ambrose, was hired by the defendant, Summit Polymers, Inc., in 1992. Summit is in the business of engineering, designing, and manufacturing molded plastic parts for automobiles. Ms. Ambrose, who had no prior automotive experience, started as a “detailer” at an annual salary of $20,800; in that role she assisted Summit’s designers by preparing drawings and finishing off designs.
After about a year Ms. Ambrose began to be assigned design work of her own, and for several years the majority of her work consisted of designing. She was given the title “designer” in November of 1996, by which time Summit’s design department had grown to include 10 employees. Her job classification did not change after 1996.
Although she had received substantial pay raises in 1994, 1995, and 1996, Ms. Ambrose was still among the three lowest earners in the department during those years. She received a 10 percent raise in January of 1997, but her salary remained lower than that of seven male designers.
Summit eliminated its “detailer” job classification at or about the time Ms. Ambrose was classified as a “designer.” Thereafter, designers did both design work and lower-skilled detail work. The amount of design work assigned to Ms. Ambrose decreased, and by late 1997 she was doing mostly detail work. From that time until her resignation in 2003, most— but not all — of Ms. Ambrose’s work was detail work.
Sometime in 1997 Ms. Ambrose learned that Bül Funk, a male designer whose job performance she viewed as comparable to her own, was making about $45,000 annually. Her own pay was only $35,200. She raised the issue with her supervisor, Mark Roodbeen, who said that he was aware of the disparity and was trying to correct it, but that the funds available for annual raises were limited. In 1998 or 1999, Ms. Ambrose began to work under a new supervisor, Brian Boyer. She broached the issue of the salary disparity with Mr. Boyer as well, and his responses were similar to Mr. Roodbeen’s.
As of January, 2003, the annual salaries in Summit’s design department (which had by then grown to 18 employees) ranged from $35,000 to $66,505. Ms. Ambrose’s salary was $44,388. Thirteen male designers were paid more, three male designers were paid less, and the only other female designer was paid about the same as Ms. Ambrose.
Ms. Ambrose resigned effective April 11, 2003. Four months later she sued Summit in the circuit court of Kalamazoo County, Michigan. Ms. Ambrose claimed that Summit had violated the Equal Pay Act by paying her at rates less than those of male *105employees who did the same or similar work. She also asserted a state-law claim of employment discrimination based on sex. Summit removed the case to federal district court.
Summit subsequently moved for summary judgment on both claims.' The district court granted the motion as to the equal-pay claim. Although Ms. Ambrose had succeeded in establishing a prima facie case, the court held, Summit had “met its heavy burden of proving that factors other than sex were the basis for [the] wage differential.” The court remanded the state-law claim to state court. Ms. Ambrose filed a timely appeal from the final judgment.
II
The Equal Pay Act prohibits a covered employer from
“paying wages to employees in [a particular] establishment at a rate less then the rate at which [it] pays wages to employees of the opposite sex in such establishment for equal work on jobs the performance of which requires equal skill, effort, and responsibility, and which are performed under similar working conditions, except where such payment is made pursuant to (i) a seniority system; (ii) a merit system; (iii) a system which measures earnings by quantity or quality of production; or (iv) a differential based on any other factor other than sex....” 29 U.S.C. § 206(d)(1).
A private right of action for violations of the Act was created by 29 U.S.C. § 216(b).
To establish a prima facie case under the Equal Pay Act, a plaintiff must show that the defendant paid higher wages to employees of the opposite sex for substantially equal work. See Coming Glass Works v. Brennan, 417 U.S. 188, 195, 94 S.Ct. 2223, 41 L.Ed.2d 1 (1974); Buntin v. Breathitt County Board of Education, 134 F.3d 796, 799 (6th Cir.1998). If the plaintiff succeeds in establishing a prima facie case, the defendant must show that the difference in pay is justified under one of the four affirmative defenses set forth in the Act. See Coming Glass Works, 417 U.S. at 196-97, 94 S.Ct. 2223; Buntin, 134 F.3d at 799. The defendant will be entitled to judgment as a matter of law if it establishes a defense “so clearly that no rational jury could [find] to the contrary.” Buntin, 134 F.3d at 800 (internal quotation marks omitted); see Kovacevich v. Kent State University, 224 F.3d 806, 827 (6th Cir.2000) (“There must be no genuine issue as to whether the difference in pay is due to a factor other than sex.”).
A
Our first task in evaluating the issue of a prima facie violation here is to determine whether there were male employees whose work was equal to Ms. Ambrose’s. “Equal work does not require that the jobs be identical, but only that there exist substantial equality of skill, effort, responsibility, and working conditions.” Buntin, 134 F.3d at 799 (internal quotation marks omitted). Substantial equality must be determined through “an overall comparison of the work, not its individual segments.” Id. (internal quotation marks omitted). On the record before us, we think a rational jury could find that Ms. Ambrose’s work was “equal” to the work of Summit’s male designers.
Summit points out that Ms. Ambrose, by her own admission, was mainly doing detail work from late 1997 through the end of her employment, while male designers were “doing primarily design work.” But Ms. Ambrose did not stop designing altogether — she handled three design projects in the years 2000 through 2003. Because *106the same skill, effort, and responsibility-needed to perform design projects were still required of Ms. Ambrose, albeit less frequently than they were required of male designers who received more such projects, the requirements of her job can be regarded as equal to the requirements of the other designers’ jobs. See 29 C.F.R. § 1620.15(a) (“If an employee must have essentially the same skill in order to perform either of two jobs, the jobs will qualify under the [Equal Pay Act] as jobs the performance of which requires equal skill, even though the employee in one of the jobs may not exercise the required skill as frequently or during as much of his or her working time as the employee in the other job”).
Equality of job requirements would bring Ms. Ambrose within the protection of the Act even if male designers were never assigned detail work. See id. § 1620.14(b) (“[W]here employees of the opposite sex are employed in jobs which are equal in the levels of skill, effort, and responsibility required for their performance ... the fact that some of the duties assigned to employees of one sex require less skill than the employee must have for the job as a whole does not warrant any conclusion that the jobs are outside the purview of the equal pay standard”). But the evidence here, as we have said, is that all designers were expected to do detail work as well as design work. Indeed, Mr. Roodbeen testified that Ms. Ambrose’s work was “similar” to that of the other designers. (One of the issues in the case that has been remanded to the state court, as we understand it, is whether Summit discriminated illegally in its assignment of design projects. That issue is not before us here.)
Summit contends that Ms. Ambrose was in fact less skilled, and had less education and experience, than the higher-paid male designers. Be that as it may, equality of work, not equality of workers, is what matters in establishing a prima facie case under the Act. See Miranda v. B & B Cash Grocery Store, Inc., 975 F.2d 1518, 1533 (11th Cir.1992) (“at this stage the jobs and not the employees are compared”); Hein v. Oregon College of Education, 718 F.2d 910, 914 (9th Cir.1983) (“The only comparison of skills that should be made in a prima facie case is a comparison of skills required by a job”). Because there is no showing here that the designers’ jobs were not performed under similar working conditions, the question is simply whether performance of Ms. Ambrose’s job “require[d] equal skill, effort, and responsibility” in comparison to performance of the male designers’ jobs. 29 U.S.C. § 206(d)(1). For the reasons stated above, we believe that a jury could answer this question in the affirmative.
Our next task is to determine whether Ms. Ambrose has made a sufficient showing of inequality in pay. It is undisputed that between August of 2000 and April of 2003, Summit paid three male designers annual salaries that were less than Ms. Ambrose’s annual salary, while 15 mail designers were paid more. (Pay disparities before August of 2000 are not actionable; a suit under the Equal Pay Act must be brought within a maximum of three years from a violation of the Act. See 29 U.S.C. § 255(a); Gandy v. Sullivan County, Tennessee, 24 F.3d 861, 863-65 (6th Cir.1994). Ms. Ambrose sued Summit in August of 2003.)
There is caselaw suggesting that a plaintiff may not ignore lower-earning employees of the opposite sex when seeking to show inequality in pay — ie., a plaintiff cannot establish a prima facie case by showing only that the highest-earning employees of the opposite sex were paid more. See Hein, 718 F.2d at 916-17, 918; Heymann *107v. Tetra Plastics Corp., 640 F.2d 115, 121-23 (8th Cir.1981). But under the methodology used in those cases — comparison of the plaintiffs pay to the average pay of opposite-sex employees doing equal work — the record shows that Ms. Ambrose was paid less than Summit’s male designers between August of 2000 and April of 2003.1 We conclude that Ms. Ambrose has met her burden of establishing a prima facie case.
B
Ms. Ambrose having presented a prima facie case, Summit is entitled to summary judgment only if it has so clearly established a statutory defense that there is no genuine issue as to whether Ms. Ambrose’s lower pay rates were based, in whole or in part, on her sex. See Balmer v. HCA Inc., 423 F.3d 606, 612 (6th Cir. 2005). Summit has asserted the second and fourth defenses enumerated in § 206(d)(1); it contends that it awarded annual raises pursuant to a merit system and that it set starting salaries on the basis of factors other than sex. Summit’s burden of establishing these defenses “is a heavy one.” Kovacevich, 224 F.3d at 826-27.
We turn first to the setting of starting salaries. Mr. Roodbeen testified that Summit determines new employees’ salaries by examining their experience, “market conditions,” and the salaries of recently hired designers. When asked why particular male designers had been hired at salaries higher than Ms. Ambrose’s, Mr. Roodbeen pointed to those designers’ skills and experience. Mr. Boyer testified similarly that starting salaries are based on “present salary and what the incoming employee is able to negotiate,” as well as on previous design experience and test scores.
Summit’s vice president of human resources, Gregory Goodman, attested in an affidavit that the company sometimes sets starting salaries “at the level necessary” to persuade employees with desirable experience to “leave their then-current employment.” Summit did exactly that, Mr. Goodman averred, when it hired 10 of the 15 male designers whose pay exceeded Ms. Ambrose’s in the years 2000 through 2003. Goodman denied that sex had played a part in the setting of those designers’ salaries.
Experience is a factor other than sex for purposes of the Equal Pay Act. Balmer, 423 F.3d at 612. “[A] new employee’s prior salary” also constitutes a factor other than sex “as long as the employer does not rely solely on prior salary to justify a pay disparity.” Id. (noting that sex-based inequality would be perpetuated if prior salary alone justified a difference in pay). Summit’s evidence shows that it considered not only prior salaries but also skills and experience that were of value to the company.
Ms. Ambrose’s efforts to discredit Summit’s evidence are, in our view, inadequate. First, we do not agree with her characterization of Mr. Goodman’s affidavit as “conjecture.” Speaking from his personal knowledge, Goodman identified 10 designers and described the experience that made each of them a desirable hire. He stated unequivocally that Summit paid what was required to recruit those employees. Summit was not required to bolster Goodman’s averments with evidence of the precise level of the employees’ prior salaries.
*108Second, we see no conflict between Mr. Goodman’s affidavit and the testimony of Messrs. Roodbeen and Boyer. All three men said that when it sets starting salaries, Summit considers both experience and the salary that an employee has earned (or could expect to earn) elsewhere.2
Further, we are not persuaded that consideration of sex as a factor in setting starting salaries can be inferred from Summit’s payroll records. Of the 13 male designers Summit hired between Ms. Ambrose’s re-classification and her resignation, four were paid less than Ambrose, eight were identified by Mr. Goodman as employees who were paid more to leave their previous employment, and one had education and experience that — according to Mr. Roodbeen’s uncontradicted testimony — -justified a higher salary. Taking the record as a whole, we do not think a reasonable jury could find that sex played a part in Summit’s determination of starting salaries.
But starting salaries are only one component of the pay Ms. Ambrose and her male colleagues received in the years 2000 through 2003. Their pay was also the product of annual raises. Summit is not entitled to summary judgment, therefore, if there is any genuine issue as to whether sex played a part in the awarding of raises.
Mr. Roodbeen testified that Summit awards annual raises on the basis of performance. A review process is conducted, he said, in accordance with instructions that typically establish a maximum total increase for each department. Mr. Goodman attested likewise that
“[e]ach department in the company, and within each department, each team, is budgeted a specific pool of funds for salary increases each year. The head of the team and the department head allocate the pay increases to employees, within the confines of the departmental allocation, on the basis of job performance of the employees.”
The criteria used in evaluating employees’ performance are, in Goodman’s words, “timeliness of design delivery, whether they complete projects within the budgeted number of hours, their design accuracy, their compliance with customer standards, and their ability to work with others on their team.”3 The record contains performance review forms that are consistent with this testimony.
In our view, Summit’s evidence shows that the company awards annual raises pursuant to a “merit system” — ie., “an organized and structured procedure whereby employees are evaluated systematically according to predetermined criteria.” Ryduchowski v. Port Authority of New York and New Jersey, 203 F.3d 135, 142-43 (2d Cir.) (internal quotation marks omitted), cert. denied, 530 U.S. 1276, 120 S.Ct. 2743, 147 L.Ed.2d 1007 (2000). Ms. Ambrose stresses that Summit never produced the “annual review instructions” mentioned by Mr. Roodbeen. But we have been given no reason to suppose that the *109instructions would contradict either Mr. Roodbeen’s testimony that raises are based on performance or Mr. Goodman’s averments that employees are evaluated according to predetermined criteria.
Ms. Ambrose also points to testimony that Mr. Roodbeen and other Summit managers sometimes rank the company’s designers during the annual review process “to see if people are in the right pay range” vis-a-vis one another. Mr. Rood-been testified that the rankings are based on designers’ ’’capabilities and work,” however, and these considerations are broadly consistent with the criteria enumerated by Mr. Goodman.
If Summit’s use of a ranking procedure suggests that factors other than the predetermined criteria might somehow have affected designers’ annual raises, we do not think a reasonable jury could find that sex was such a factor. In all but the last three years of her employment, Ms. Ambrose’s annual raise, expressed as a percentage of the previous year’s salary, equaled or exceeded the majority of her male colleagues’ raises.4 And Ms. Ambrose acknowledges that her skills “deteriorated ... dramatically” after 1999 — i e., in the time period during which she was not receiving challenging design projects, for the most part, and her raises were lower. Christine Miller, another female designer, fared about as well as Ms. Ambrose. In 1996, 1998, 1999, and 2000, Ms. Miller’s annual percentage raise equaled or exceeded those of the majority of the male designers. In 1997, 2001, and 2002, Miller’s raise equaled or exceeded the raises of 46 to 50 percent of the male designers.5
In sum, we think the record compels a finding that sex played no part in the difference between Ms. Ambrose’s rate of pay and the rates of male designers. Whatever the merits of her claims under state law (a matter as to which we intimate no opinion), Ms. Ambrose is not entitled to proceed further under the federal Equal Pay Act. The grant of summary judgment on the claim under that act is AFFIRMED.
MARTHA CRAIG DAUGHTREY, Circuit Judge, concurring in part and dissenting in part.. Summit paid its male designers an average of about $44,000 in 2000, $48,500 in 2001, $51,000 in 2002, and $52,500 in 2003. Ms. Ambrose was paid $40,250, $42,250, $43,518, and $44,388 during those same four years.
. Ms. Ambrose points to a performance evaluation for one of the 10 male designers identified by Mr. Goodman, wherein Summit manager Tony Crespo wrote that he hired the designer in question “for his attitude not his experience.” There being no evidence that Mr. Crespo was the sole decision-maker with respect to the designer’s hiring, we do not think this evidence undercuts Goodman's averment.
. The application of these criteria may be characterized as "subjective,” to a degree, but this is no bar to a finding that Summit's annual review procedure is merit-based. See Harrison-Pepper v. Miami University, 103 Fed.Appx. 596, 601 (6th Cir.2004). An element of subjectivity is inevitable in the operation of any pay system that is truly merit-based.
. In 2001, moreover, Ms. Ambrose's raise exceeded the raises of six male designers, while seven male designers received higher raises.
. Our dissenting colleague points out that Reiker Schultz, a male designer who was hired in 1999 with no prior designing experience, earned nearly as much as Ms. Miller did in 2003. The record shows that Mr. Schultz was given substantial mid-year raises in 2000 and 2001 to prevent him from leaving Summit for other employment. Mr. Goodman specifically denied that these raises were based on Mr. Schultz’ sex, and there is no contrary evidence in the record.