Plaintiff-Appellant Lorenzo Baker appeals the district court’s order granting summary judgment to Defendant-Appel-lee, Randstad North America, on his employment discrimination claim. Baker contends that he successfully established a prima facie case of race discrimination and that a genuine issue of material fact exists as to whether Randstad’s articulated reason for terminating him was a pretext for race discrimination.

I. FACTS AND PROCEEDINGS

Randstad is a staffing company that places individuals in temporary and permanent positions in a variety of sectors, including light industrial and office support. Randstad entered the Houston market in 2000 and hired Baker as a Business Development Manager (“BDM”) that spring. BDMs are responsible for gener *316 ating new business for Randstad. Rand-stad requires each BDM to make at least 75 contacts per week with potential clients and have ten meetings per week with decision makers at potential clients’ businesses. Randstad also expects BDMs to secure new accounts on a consistent, weekly basis. Once the BDM closes the initial deal, Randstad agents assume responsibility for client relationship and work with the client to fill personnel vacancies as they occur. Baker’s initial supervisor was Regional Market Manager Alyson Blake. In the fall of 2000, Ron Griffin replaced her.

Within his first two months at Randstad, Baker secured Reliant Energy as a client. The Reliant account generated substantial revenue, and Baker received special recognition for being the top revenue producing BDM in the Houston market as a result of it. Ultimately, however, Randstad had to drop the account in early 2001 because it was not profitable. Specifically, it cost Randstad money to place employees with Reliant because of various workers’ compensation claims that arose out of Reliant placements.

Securing the Reliant account turned out to be the high point in Baker’s tenure at Randstad. Notably, although Baker typically made a qualifying number of contacts each week, 1 he admittedly never met the requirement of ten meetings per week with the potential clients’ decision makers. In the two months before he was fired, Baker averaged approximately four meetings per week. Baker also failed to secure new accounts on a consistent basis. During his 11 months with Randstad, Baker secured only nine new client accounts; Randstad expected Baker to secure more than three times that many new accounts in the same period.

In the fall of 2000, Randstad entered a period of financial decline and laid off 250 employees nationwide. In December, Randstad instructed the Regional Market Managers to terminate the employees who were their lowest performers. Although Baker was near the bottom of the list on the basis of his activity levels, his supervisor, Griffin, decided to keep Baker on the staff and monitor his progress over the following three months. Ultimately, Randstad laid off eight employees in the Houston market in December. The only BDM who Griffin selected for layoff at this time was Heather Barladge, a white female. Griffin selected Barladge because she was unable to meet her activity requirements. Of the total number of employees that Randstad laid off in December, more than half were white.

At the end of the first quarter of 2001, Randstad again initiated a nationwide reduction in force, this time eliminating approximately 200 employees. As with the December 2000 layoffs, Randstad instructed its Regional Market Managers to lay off their lowest performers. In evaluating the BDMs under his supervision, Griffin placed special emphasis on their abilities to secure new accounts. If a BDM was un-derperforming in that area, Griffin would evaluate the BDM’s activity levels, i.e., the number of contacts that the BDM made each week and the number of client meetings that the BDM had each week. Griffin also considered productivity, which includes revenue, gross margin, and profitability. By April of 2001, Baker had failed to improve his performance, and he was the lowest performing BDM in the Houston market. Accordingly, Griffin selected him for termination. Baker was the only BDM in the Houston market that Rand-stad laid off at that time.

*317 Baker filed discrimination charges against Randstad with the EEOC. The EEOC repeatedly requested documentation from Randstad regarding the activity levels of the retained BDMs. Randstad, however, produced only the activity reports for Baker and Melissa Tennison, a Hispanic employee, for the two-month period preceding Baker’s termination. These reports show that Tennison secured more accounts and had a much higher activity level than Baker during that period. After investigation, the EEOC issued Baker a right to sue letter. He then sued Randstad under Title VII and 42 U.S.C. § 1981, alleging that Randstad terminated him because he is black.

Randstad filed a motion for summary judgment, asserting that there is no evidence that its articulated nondiseriminato-ry reason for terminating Baker was a pretext for race discrimination. The district court granted summary judgment to Randstad on two bases. First, the district court ruled that Baker had failed to establish a prima facie case of race discrimination because he submitted no evidence that he was replaced by someone outside the protected group. Second, the district court concluded that there was insufficient evidence to create a genuine issue of material fact as to whether Randstad’s articulated reason for terminating Baker was pretextual. Baker appeals the district court’s grant of Randstad’s summary judgement motion, arguing that (1) he established a prima facie case of race discrimination, and (2) an issue of fact exists as to whether Randstad’s reason for terminating him is a pretext for race discrimination.

We have jurisdiction over Baker’s appeal of the district court’s judgment under 28 U.S.C. §§ 1381 and 1291.

II. ANALYSIS

A. Standard of Review

We review the district court’s grant of summary judgment in favor of Randstad de novo. 2 We shall affirm the district court when there is no genuine issue of material fact and the moving party is entitled to summary judgment as a matter of law. 3 We consider the evidence in the light most favorable to the non-movant, but he must point to evidence that shows that there is a genuine issue of fact for trial. 4

B. Legal Standard

Both Title VII and § 1981 prohibit employers from taking adverse employment actions against employees on the basis of race. 5 As there is no direct evidence of discrimination in this case, we evaluate Baker’s claims under the burden-shifting framework of McDonnell Douglas Corp. v. Green. 6 First, Baker must establish a pri-ma facie case of race discrimination. This means that Baker must demonstrate that (1) he is a member of a protected class, (2) he was qualified for the job, (3) he suffered an adverse employment action, and (4) similarly situated employees outside the protected group were treated more favor *318 ably than he. 7 Once Baker establishes a prima facie case of race discrimination, the burden shifts to Randstad to articulate a legitimate, non-discriminatory reason for terminating Baker. If Randstad does so, Baker has the ultimate burden of proving that Randstad’s articulated reason is a pretext for race discrimination. 8

On appeal, Randstad has waived any argument that Baker did not establish his prima facie case. 9 Accordingly, we address only the issue whether there was sufficient evidence to establish that Rand-stad’s articulated reason for terminating Barker was a pretext for race discrimination. Baker contends that (1) the absence of documentary evidence regarding the activity levels of most of the other BDMs who were not fired when he was, invokes the spoliation doctrine and alone establishes an issue of fact as to whether Rand-stad’s articulated reason for terminating Baker was pretextual; (2) the evidence shows that Baker was a better employee than both Melissa Tennison, a female Hispanic BDM, and Stephen Horton, a white male BDM; and (3) Randstad engaged in a pattern of laying off black employees.

C. Misplaced Documentary Evidence and Spoliation

Baker asserts that Randstad’s failure to produce full documentation in support of its decision to terminate Baker invokes the spoliation doctrine and justifies a presumption that the withheld evidence would have been unfavorable to Randstad. Thus, argues Baker, this non-production alone raises an issue of fact as to whether Randstad’s reason for terminating him was pretextual and exempts him from an adverse summary judgment. Randstad counters that it simply misplaced the relevant documents.

At the outset, Baker cites no authority for the proposition that misplaced or un-produced documents establish a per se presumption of pretext in an employment discrimination case. Furthermore, spoliation is a specific doctrine that requires the party invoking it to show, inter alia, that his adversary destroyed or misplaced the evidence in bad faith. 10 Here, Baker neither specifically alleged that Randstad misplaced the documents in bad faith nor moved the district court to determine whether Randstad misplaced the documents in bad faith. 11 In fact, Baker never even moved the district court to compel Randstad to produce the missing documents. As Baker bears the ultimate burden of proving pretext, the employer’s destruction or withholding of the missing documents could very well have damaged him. 12 Nevertheless, the *319 summary judgment record on appeal is devoid of evidence addressing whether Randstad misplaced, withheld, or destroyed the documents in bad faith. Therefore, our hands are tied: As there is no evidence of bad faith, Baker’s spoliation argument fails.

D. Other Evidence

Baker contends that, even absent the aforesaid records, he presented sufficient evidence to create a genuine issue of material fact as to whether Randstad’s reason for terminating him was pretextual. Specifically, Baker points to evidence that he was the highest revenue generator for Randstad in the Houston market. Baker also offers evidence that other BDMs whom Randstad did not fire never secured any accounts at all.

1. Baker’s Revenue

Randstad does not dispute that Baker was the highest revenue generator in the Houston market. This was because of the Reliant account, which Baker secured at the beginning of his employment with Randstad. Significantly, however, revenue was neither the only nor the most important factor that Griffin considered in deciding whom to fire. Instead, Griffin focused on both productivity and activity levels. Griffin explained that a high revenue generator would be a proper candidate for termination if he was “living off an account [he] had closed six months before” and his activity levels were not up to par. As noted, Griffin reviewed the BDMs’ activity reports and, in so doing, paid special attention to the BDMs’ activity levels and to new accounts that the BDMs generated. Furthermore, Griffin focused on the months immediately preceding the April 2001 reduction in force.

Griffin chose to lay off Baker because he failed to bring in new business on a consistent, weekly basis and his activity levels were the lowest in the region: Baker was never able to secure ten meetings with decision makers in any week. Moreover, despite Baker’s revenue generation, his overall productivity levels are not stellar under Griffin’s analysis. This is because Baker was continuing to “live off’ the Reliant account. Notably, Baker closed the Reliant account more than eight months before his termination, and that particular account had negative profitability despite its gross revenue generation because of the high costs and expenses that Randstad incurred in staffing Reliant. Ultimately, Baker’s argument regarding Griffin’s evaluation of his performance in light of his high revenue generation does not demonstrate that Randstad fired Baker because of his race. 13 That argument is insufficient to impugn Randstad’s articulated, nondiscriminatory reason for terminating Baker.

2. Tennison’s Performance

Randstad did produce Tennisoris activity reports for the two-month period preceding Baker’s termination. These reports show that she met or exceeded Randstad’s requirements for weekly contacts and meetings and that she secured seven new accounts during that period. In comparison, Baker’s activity reports show that he never met the weekly meetings requirement and secured only two new accounts during that period.

Baker proffers the declaration of Kimberly Sanders to contradict Tennisoris record. Specifically, Sanders stated that Ten-nison had not sold a single account from *320 the time Randstad hired her through March of 2001. Construing Sanders’s testimony in the light most favorable to Baker, i.e., assuming that Randstad’s activity reports for Tennison are erroneous, still does not create an issue of fact. This is because, even if the reports are proved to be erroneous, it is undisputed that Griffin relied on them in making the reduction-in-force determinations, and Tennison’s purportedly erroneous activity reports nevertheless show her to be a more productive employee than Baker. “The question is not whether an employer made an erroneous decision; it is whether the decision was made with discriminatory motive. Even an incorrect belief that an employee’s performance is inadequate constitutes a legitimate, nondiscriminatory reason.” 14 Accordingly, Sanders’s statement fails to raise an issue of fact as to whether Rand-stad’s articulated reason for terminating Baker was a pretext for race discrimination.

3. Horton’s Record

Baker contends that Randstad should have laid off Horton instead of him because Horton had not secured any accounts from the time he was hired by Randstad in March of 2001 until the time that Baker was terminated in April of 2001. Assuming this to be true, the evidence in the record is undisputed that (1) Baker secured no new accounts during the same period, (2) Baker failed to fulfill his activity requirements, and (3) unlike Baker, Horton met his activity requirements in the weeks preceding Baker’s termination. Accordingly, Baker’s testimony that Horton failed to secure any new accounts before his termination is insufficient to raise an issue of fact as to whether Randstad’s articulated reason for terminating Baker was a pretext for race discrimination.

E. Pattern of Terminating Black Employees

Finally, Baker insists that Randstad engaged in a pattern of terminating black employees, and that this pattern raises an issue of fact as to whether Randstad’s articulated reason for terminating Baker is pretextual. Although statistical evidence can be probative of pretext, it is extraordinarily rare that raw numbers can insulate a plaintiff from summary judgment. 15 Indeed, “[t]he probative value of statistical evidence ultimately depends on all the surrounding facts, circumstances, and other evidence of discrimination.” 16

For openers, Baker fails to show that a distinct pattern of laying off black employees relative to white employees even existed. In the first round of lay offs late in 2000, Randstad terminated eight employees in the Houston market: Five were white; three were black. Of the BDMs terminated in this initial round of layoffs, only one in the Houston market was selected by Griffin, and she was white. Baker also asserts that Randstad terminated four black BDMs (including himself) and six black agents in the Houston market in 2001. There is also evidence that Rand-stad laid off at least one white BDM in the Houston market in 2001 for low activity levels, but there is no evidence in the record as to the total number of white employees Randstad laid off at that time. *321 We cannot conclude from the evidence before us that Randstad engaged in a pattern of targeting black employees for termination.

Moreover, even if Baker’s numbers had shown a possible pattern, his argument still fails. This is because statistical presentations that do not include analyses of the facts surrounding the circumstances of the individual employees at issue are “impotent, without more, to rebut” an employer’s articulated, nondiscriminatory reason for terminating an employee. 17 Significantly, Baker concedes that nothing more than his subjective belief that black employees were laid off in disproportionate numbers supports his pattern-of-discrimination argument. Baker admittedly offers no evidence or analysis apart from the bare numbers to indicate that Randstad engaged in a pattern of terminating black employees because of their race. 18 Accordingly, Baker’s pattem-of-discrimination argument fails. In this regard, we note in passing that the undisputed evidence in the record before us shows that Randstad had legitimate, nondiscriminatory reasons for terminating the black BDMs in 2001. Specifically, Randstad fired two black BDMs for violating company policy, laid off one black BDM for poor performance, and the fourth black BDM voluntarily resigned.

III. CONCLUSION

Baker fails to point to any summary judgment evidence sufficient to create an issue of material fact whether Randstad’s articulated reason for terminating him was a pretext for race discrimination. Absent that, Randstad is entitled to summary judgment as a matter of law. The district court’s grant of Randstad’s motion for summary judgment is AFFIRMED.