Elizabeth CASTRO, et al., Plaintiffs-Appellants, v. DEVRY UNIVERSITY, INC., Defendant-Appellee, 786 F.3d 559

HAMILTON, Circuit Judge. Plaintiffs Elizabeth Castro, LaTonya Brooks, and Michael Florez sued their former employer, defendant DeVry University, Inc., under Title VII of the Civil Rights Act of 1964. Plaintiffs allege that DeVry retaliated against them 'by terminating their employment for complaining about their supervisor’s racially and ethnically derogatory remarks. DeVry transferred the supervisor about three months after plaintiffs complained. After that time, he neither supervised plaintiffs nor participated in any of the termination decisions.

Plaintiffs were discharged at different times, from ten to thirty months after their complaint, and the evidence concerning their individual circumstances and job performance varies. The district court granted

summary judgment to DeVry on all three retaliation claims. . We affirm the district court’s decision on the claims by Castro and Brooks, but we reverse its decision on the claim by Florez. Castro was terminated thirty months after the complaint because of poor performance over a sustained period. Brooks was terminated fifteen months after the complaint because of multiple instances of dishonesty and inconsistent performance. Neither Castro nor Brooks has raised a genuine issue of material fact on whether these reasons were pretexts for retaliation.

Florez, however, has raised a genuine issue of material fact about retaliatory motive. He was terminated ten months after the complaint for two stated reasons: inconsistent performance and his “volatile behavior.” On appeal, DeVry has conceded that Florez’s performance did not justify his termination. Florez has also offered evidence that DeVry’s “volatile behavior” explanation was a pretext for retaliation. First, he has presented evidence that his managers did not honestly believe he had behaved unprofessionally.

Second, he has offered evidence that DeVry falsely told the Equal Employment Opportunity Commission that his manager — who made the key recommendation for’ his firing— did not know about the complaint when in fact she did know. Third, an email recommending his termination referred specifically to his complaint about the supervisor’s remarks. Although a reasonable jury would not be compelled to find retaliation on this record, such a finding would be permissible. DeVry was not entitled to

summary judgment on Florez’s claim. I. Procedural Background Plaintiffs filed suit under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq., asserting two claims: (1) they were subjected to a racially and ethnically hostile work environment, and

(2) DeVry terminated their employment in retaliation for complaining about their supervisor’s racially and ethnically derogatory remarks. The parties consented to the jurisdiction of the magistrate judge under 28 U.S.C. § 686(c). After more than a year of discovery, DeVry moved for

summary judgment on all claims. Plaintiffs conceded that their hostile work environment claims should be dismissed but argued that they had raised genuine issues of material fact on their respective retaliation claims. The district court disagreed, granting

summary judgment to DeVry on all three retaliation claims. Castro v. DeVry University, Inc., 941 F.Supp.2d 965 (N.D.Ill.2013). We review de novo the grant of

summary judgment, examining the record in the light most favorable to the non-moving parties and drawing all reasonable inferences in their favor. E.g., Carter v. Chicago State University, 778 F.3d 651, 657 (7th Cir.2015).

Summary judgment is appropriate only where there are no genuine issues of material fact and the moving party is entitled to judgment as a matter of law. Fed.R.Civ.P. 56(a). II. The Law of Retaliation &

Summary Judgment Title VII prohibits employers from retaliating against employees who engage in activity protected by the statute. 42 U.S.C. § 2000e-3(a). We have often said there are two ways plaintiffs may prove their claims, which we have labeled the “direct” and “indirect” methods of proof. E.g., Silverman v. Board of Education of City of Chicago, 637 F.3d 729, 740 (7th Cir.2011).

But over the past several years, we have questioned the utility of the distinctions between them, recognizing that both methods of proof converge on the same fundamental question: could a reasonable trier of fact infer retaliation or discrimination, as the case may be? See, e.g., Bass v. Joliet Public School District No. 86, 746 F.3d 835, 840 (7th Cir.2014); Perez v. Thorntons, Inc., 731 F.3d 699, 703 (7th Cir.2013); Hitchcock v. Angel Corps, Inc., 718 F.3d 733, 737 (7th Cir.2013); Naficy v. Illinois Dep’t of Human Services, 697 F.3d 504, 514 (7th Cir.2012); Harper v. C.R.

England, Inc., 687 F.3d 297, 313-14 (7th Cir.2012); Coleman v. Donahoe, 667 F.3d 835, 862-63 (7th Cir.2012) (Wood, J., concurring). Plaintiffs proceed under only the direct method of proof. Under this method, plaintiffs must offer evidence of three elements: (1) they engaged in protected activity, (2) they suffered adverse employment actions, and (3) there was a causal connection between the protected activity and the adverse employment actions. E.g., Greengrass v. Int’l Monetary Systems Ltd., 776 F.3d 481, 485 (7th Cir.2015).

Whether we apply this method formally or just cut to the chase and ask the fundamental question directly — could a reasonable trier of fact infer retaliation? — makes no difference. The first two elements are satisfied. Plaintiffs complained to Human Resources manager Alana Hurt on April 16, 2007 that their supervisor, Phil Giambone, often used racially and ethnically derogatory language in the workplace. Whether Giambone’s comments went so far as to violate Title VII does not matter.

Plaintiffs sincerely and reasonably believed they were complaining about conduct prohibited by Title VII, which is all that is required to establish protected activity. E.g., Magyar v. St. Joseph Regional Medical Center, 544 F.3d 766, 771 (7th Cir.2008). The plaintiffs were all terminated at various times after the complaint. A termination is of course a materially adverse employment action. E.g., Nichols v. Southern Illinois University-Edwardsville, 510 F.3d 772, 780 (7th Cir.2007).

The question is whether plaintiffs have offered sufficient evidence to create a genuine issue of material fact on whether their complaint caused their terminations. To establish this causal link, plaintiffs can rely on direct or circumstantial evidence. E.g., Harper, 687 F.3d at 307. Plaintiffs do not claim they have any direct evidence of DeVry’s retaliation — i.e., there is no admission from a DeVry agent that it discharged the plaintiffs because they complained. Plaintiffs rely on circumstantial evidence.

Circumstantial evidence suffices if “a convincing mosaic of circumstantial evidence” would permit a reasonable trier of fact to infer retaliation by the employer. Rhodes v. Illinois Dep’t of Transportation, 359 F.3d 498, 504 (7th Cir.2004) (internal quotation marks omitted). In retaliation cases, we have recognized three categories of circumstantial evidence available to a plaintiff using the “convincing mosaic approach.” Coleman, 667 F.3d at 862.

These categories include (1) evidence of suspicious timing, (2) evidence that similarly situated employees were treated differently, and (3) evidence that the employer’s proffered reason for the adverse employment action was pretextual. Id. at 860, citing Silverman, 637 F.3d at 734 (suspicious timing); Volovsek v. Wisconsin Dep’t of Agriculture, Trade & Consumer Protection, 344 F.3d 680, 689 (7th Cir.2003) (disparate treatment); Dickerson v. Board of Trustees of Community College District No. 522, 657 F.3d 595, 601 (7th Cir.2011) (pretext).

Each category of circumstantial evidence can suffice by itself to preclude

summary judgment, depending on its strength 'in relation to the other evidence, but plaintiffs may also use them together. Coleman, 667 F.3d at 862; Troupe v. May Dep’t Stores Co., 20 F.3d 734, 736 (7th Cir.1994). Here, plaintiffs rely on both suspicious timing and pretext. Suspicious timing can sometimes raise an inference of a causal connection, but temporal proximity alone is “rarely sufficient” to establish causation. O’Leary v. Accretive Health, Inc., 657 F.3d 625, 635 (7th Cir.2011).

We have rejected any bright-line rule about how close the events must be to establish causation, but in cases where there is “corroborating evidence of retaliatory motive,” an “interval of a few weeks or even months may provide probative evidence of the required causal nexus.” Coleman, 667 F.3d at 861. We have also noted, though, that the mere passage of time “does not conclusively bar an inference of retaliation.” Malin v. Hospira, Inc., 762 F.3d 552, 560 (7th Cir.2014) (reversing

summary judgment for employer where evidence showed patient retaliation over period of several years). To show pretext, an employee “must present evidence suggesting that the employer is dissembling.” O’Leary, 657 F.3d at 635; see also Naik v. Boehringer Ingelheim Pharmaceuticals, Inc., 627 F.3d 596, 601 (7th Cir.2010). “The question is not whether the employer’s stated reason was inaccurate or unfair, but whether the employer honestly believed the reason it has offered to explain the discharge.” O’Leary, 657 F.3d at 635.

To meet this burden, the employee “must ‘identify such weaknesses, implausibilities, inconsistencies, or contradictions’ ” in the employer’s proffered reason “ ‘that a reasonable person could find [it] unworthy of credence.’ ” Coleman, 667 F.3d at 852 (alteration in original), quoting Boumehdi v. Plastag Holdings, LLC, 489 F.3d 781, 792 (7th Cir.2007). To survive

summary judgment, plaintiffs’ mosaic of circumstantial evidence must be strong enough to permit a reasonable trier of fact to find that DeVry terminated each plaintiff because he or she complained about Giambone’s remarks. See University of Texas Southwestern Medical Center v. Nassar, 570 U.S.-, 133 S.Ct. 2517, 2533, 186 L.Ed.2d 503 (2013). Each plaintiff relies on a different assortment of evidence to establish a causal link between the protected activity and his or her firing. We begin by setting out the facts common to all plaintiffs.

Next we analyze the evidence of suspicious timing upon which all three plaintiffs rely. We conclude that this evidence does not by itself create a genuine issue of material fact on causation for any of the plaintiffs. We then analyze the remaining evidence offered by each plaintiff to show that De-Vry’s stated reasons for each termination were pretextual. III. Facts Common to All Plaintiffs Defendant DeVry University, Inc. is a for-profit school with several campuses in the Chicago area. Plaintiffs Elizabeth Castro, LaTonya Brooks, and Michael Flo-rez are former admissions officers of De-Vry.

Castro and Florez are Mexiean-American; Brooks is African-American. In early 2007 Castro was an assistant director of admissions for DeVry. Her immediate supervisor was Phil Giambone, then the director of admissions for the Chicago campus. Florez and Brooks were academic advisors who reported to Castro. DeVry’s enrollment model is based on recruiting prospective students for classes that begin every other month — in January, March, May, July, September, and November.

Admissions officers at DeVry are responsible for following up on “leads” (potentially interested students), interviewing applicants, and otherwise assisting prospective students with the enrollment process. The goal for admissions officers is to enroll new students. On April 14, 2007, Florez approached Giambone and requested time off on May 5. Giambone responded by asking if Florez was going to “get drunk with your people” on Cinco de Mayo. Florez reported the remark to Castro. Castro, along with Flo-rez and Brooks, then decided to approach Human Resources about Giambone’s behavior. A.

Plaintiffs’April 16, 2007 Complaint to Human Resources Plaintiffs met with HR manager Alana Hurt on April 16, 2007. They discussed their concerns about Giambone’s racially and ethnically derogatory remarks. Other academic advisors corroborated plaintiffs’ complaints about Giambone. Shortly after the meeting, Hurt called HR director Deb Maher to tell her about the complaint. Maher relayed the complaint to Christine Hierl, the dean of enrollment management for the Chicago area. B. Giambone’s Conduct After the April 16 Meeting Ten minutes after Castro returned from the meeting with HR, Giambone came to her office.

He was “furious” and asked Castro whether she had “anything to tell him.” He then told Castro that she needed to do “phone work” with him. Castro had not done this type of work since she had been an academic advisor, eight years earlier. Several days after the April 16 meeting, Giambone began steering “leads” about prospective students away from Castro’s team toward another team. Giambone also told Brooks and Florez not to “hang out” or otherwise associate with Castro. Then, in early June 2007, Giambone told Florez that Castro was “stealing” his students, an accusation Castro denies. C.

The April 28, 2007 Sales Meeting About two weeks after plaintiffs’ complaint to HR, DeVry conducted a sales meeting in its Chicago office to address concerns about low enrollment numbers for the July 2007 admissions class. At the meeting Castro complained about how the “leads” were being distributed in the Chicago office. Christine Hierl jumped in and “verbally attacked” Castro. Hierl later sent Castro a memo stating that her negativity was hurting morale in the office.

Immediately after the meeting, plaintiffs assert, Castro entered a stairwell and heard Hierl say to Giambone: “There’s no way we’re going to let a bunch of wetbacks run this office.” Giambone responded with laughter. After hearing the “wetbacks” comment, Castro filled out a transfer request for any opening for a director or assistant director of admissions at DeVry. Giambone approved the request. Hierl told Castro that she could be transferred anywhere she wanted if she waited until school started for the July 2007 admissions class. D.

The July 2007 Reorganization In July 2007 DeVry transferred Giam-bone from his position and made him a high school manager, where he would supervise admissions presentations to high school students. From that point forward, Giambone did not supervise Castro, Brooks, or Florez. There is no evidence that he participated in any of the termination decisions. Kathaleen Berry, who had been the director of admissions for DeVry’s Addison and Tinley Park campuses in the Chicago area, took over Giambone’s position.

At Berry’s request, Julie Strauss, an assistant director of admissions for the Tinley Park campus, began supervising some of the Chicago-based admissions officers, including Brooks and Florez. As part of the reorganization, DeVry also demoted Castro to senior academic advisor and transferred her to the Addison campus. At the Addison campus, Castro reported to assistant director of admissions Casey Tobin, who in turn reported to Berry in Chicago. Despite her demotion, Castro’s compensation was not reduced. When Berry replaced Giambone in the Chicago office, she met with the admissions officers.

She told them that if they had a problem with another employee, they needed to handle it “in-house.” She warned her staff not to go “running off to HR.” After the July 2007 meeting, Strauss reinforced this message, telling Florez and Brooks: “Kathy Berry is different. Don’t go to HR. If you go to HR, the people that have went to HR no longer work here.” Sometime after the meeting, Giambone told Florez: “You see what happens to traitors like Liz [Castro].... I told you if you want to be someone in this company,, you need to be loyal.” Many months passed before any of the plaintiffs were terminated.

Florez was discharged on February 21, 2008, ten months after the April 2007 complaint; Brooks on July 8, 2008, fifteen months after the complaint; and Castro on November 3, 2009, thirty months after the complaint. IV Plaintiffs’ Evidence of Suspicious Timing Plaintiffs acknowledge that delays of these lengths cannot themselves establish causation. Cf. Clark County School District v. Breeden, 532 U.S. 268, 273-74, 121 S.Ct. 1508, 149 L.Ed.2d 509 (2001) (per curiam) (gap of twenty months between protected activity and adverse employment action “suggests, by itself, no causality at all”); Everroad v.

Scott Truck Systems, Inc., 604 F.3d 471, 481 (7th Cir.2010) (one-year delay too long to establish causation in the absence of other evidence); Haywood v. Lucent Technologies, Inc., 323 F.3d 524, 532 (7th Cir.2003) (same); accord, Oest v. Illinois Dep’t of Corrections, 240 F.3d 605, 616 (7th Cir.2001) (“The inference of causation weakens as the time between the protected expression and the adverse action increases.... ”).

Plaintiffs argue that although the periods between their protected activity and their terminations were relatively long, a reasonable trier of fact could still infer causation based on suspicious timing because retaliatory acts started almost immediately after the April 2007 complaint. According to plaintiffs, the

summary judgment record would permit a reasonable finding that DeVry began retaliating against plaintiffs within weeks of their protected activity and that each termination was merely the culmination of a long campaign to punish them for complaining about Giambone. Plaintiffs identify four types of retaliatory acts to support this theory:

(1) Giambone came to Castro’s office ten minutes after the meeting with HR and assigned her phone work;

(2) Giambone began steering “leads” about prospective students away from Castro’s team toward another team;

(3) Giambone told Brooks and Florez not to associate with Castro and told Florez that Castro was “stealing” students from him; and

(4) Hierl’s offensive “wetbacks” comment to Giambone. None of this evidence establishes a genuine issue of material fact on the issue of causation. DeVry transferred Giambone from his position in July 2007. There is no evidence that he had any authority over plaintiffs after that time or that he had any input in the decisions to terminate them. Even if Giambone wanted to retaliate against plaintiffs, a reasonable trier of fact could not find that he influenced the termination decisions.

Hierl, on the other hand, did have input in the decision to terminate Brooks, but her offensive comment was no more than an isolated “stray remark.” See, e.g., Overly v. KeyBank Nat’l Ass’n, 662 F.3d 856, 865 (7th Cir.2011). Hierl’s offensive “wetbacks” epithet did not refer to Brooks (she is African-American, while the other plaintiffs are Mexican-American), was made approximately fourteenth months before Brooks was terminated, and did not refer to either the protected activity or the termination decision.

A reasonable trier of fact could not find based on this single offensive comment that Hierl intended to terminate Brooks because she had complained to HR about Giambone. See Sun v. Board of Trustees of University of Illinois, 473 F.3d 799, 813 (7th Cir.2007) (“stray remarks that are neither proximate nor related to the employment decision are insufficient to defeat

summary judgment”); cf. Perez v. Thorntons, Inc., 731 F.3d 699, 709-10 (7th Cir.2013) (contrasting tension between “stray remarks” cases and “common actor” cases involving racial, ethnic, or gender bias).

Plaintiffs counter that Giambone’s retaliatory intent should be imputed to Berry (the director of admissions), who did have input in the termination decisions, based on Berry’s comment that her staff should not go “running off to HR” and Strauss’s comment that Florez and Brooks should not go to HR because the people who “went to HR no longer work here.” Plaintiffs argue that a reasonable trier of fact could find based on these comments that Berry and Strauss had picked up the mantle and intended to retaliate against plaintiffs out of loyalty to Giambone. We disagree.

Construing the comments in the light most favorable to plaintiffs, the statements suggest that Berry valued loyalty among her subordinates and would have perceived a future complaint to HR as

an act of disloyalty, potentially worthy of punishment. This type of comment could under some circumstances give rise to an inference of retaliation. Employers cannot retaliate against employees who complain about violations of Title VII under the ruse that the employee was being “disloyal” or “insubordinate” by opposing the unlawful activity. But there are two problems with plaintiffs’ position on this record. First, neither comment referred either explicitly or implicitly to the protected activity at issue: plaintiffs’ complaint in April 2007 about Giambone. The statements looked forward, not backward.

They warned about making hypothetical complaints in the future, not plaintiffs’ past complaint about Giambone. After all, Brooks and Florez had gone to HR and were still working in the same office when the comments were made. Absent more context tying the comments to the protected activity, these generic, forward-looking remarks about loyalty would not permit a reasonable trier of fact to infer a causal connection between the April 2007 complaint and the terminations. Second, both comments were made long before any of the plaintiffs’ terminations.

The shortest gap was approximately seven months. (Florez was fired in February 2008.) Ambiguous comments so far removed from the adverse employment action are insufficient, without more, to defeat

summary judgment. See, e.g., Dass v. Chicago Board of Education, 675 F.3d 1060, 1072 (7th Cir.2012) (ambiguous comment made ten months before discharge was insufficient, without more, to defeat

summary judgment). Accordingly, a reasonable trier of fact could not find based on these ambiguous comments made at least seven months before the earliest termination that Berry retaliated against plaintiffs because of their complaint about Giambone. Thus, plaintiffs’ evidence of suspicious timing does not, by itself, raise a genuine issue of material fact on causation for any of plaintiffs’ claims. The timing does not conclusively bar inferences of retaliation, however, so we proceed to the remaining evidence for each of the plaintiffs. V.

Florez’s Evidence of Retaliatory Intent We begin with Florez, who was fired first, in February 2008. At that time DeV-ry offered two reasons for its decision: (1) inconsistent performance and (2) “volatile behavior.” On appeal, however, DeVry concedes that Florez’s performance was “adequate” and that his performance alone did not justify his termination. We focus on the second explanation. DeVry identifies two incidents of Flo-rez’s “volatile behavior”: (1) a conversation he had with Strauss in October 2007 and (2) a conversation he had with Berry in January 2008.

Florez disputes the factual basis of both managers’ descriptions of these events. The first incident occurred sometime in October 2007. Florez acknowledged during his deposition that there was a “blowup” between Strauss and him, but he testified tha,t it was Strauss who caused it. Strauss came by to observe Florez’s “phone work,” but he was not in his office. When Florez returned to his office, Strauss again tried to conduct the observation, but there were technical difficulties with the phone.

According to Strauss, Florez grew impatient and complained about being observed; he then yelled, slammed doors, and eventually left his office. Florez disputes Strauss’s version of the conversation in ways that call into doubt the honesty of Strauss’s account. He testified during his deposition — and we accept as true for purposes of

summary judgment — that he did not yell or slam doors and that he behaved professionally throughout the conversation. Moreover, Mara Leal (another academic advisor) corroborated Florez’s account of his conversation with Strauss. Leal testified during her deposition that it was Strauss, not Florez, who slammed a door and yelled. Leal also testified that after the incident between Strauss and Florez, she called Berry to ask if she could move offices. Berry responded that Leal needed to do her a “favor.” Berry asked Leal if she had heard or seen the incident between Florez and Strauss. Leal said yes.

Then Berry told Leal that if HR manager Hurt called her to ask whether she had observed the incident, Leal needed to say that she did not see anything. Leal agreed, and Berry told her that she could move offices later that night. The second incident occurred on January 31, 2008, when Berry herself tried to observe Florez’s phone work. Florez complained to Berry about a poor performance evaluation he had received from Strauss in October 2007, shortly after the reported “blow-up” with her.

Berry found Florez’s behavior inappropriate; she believed that if he had concerns about his evaluation, he should have addressed them three months earlier, not during the scheduled observation. Following the phone session, Berry sent a memo to Hurt in HR with a copy to Hierl (the dean of enrollment management). Berry’s memo said that Florez had been “strident and yelling” and had made threats about lawyers and complaints to the EEOC. She also criticized Florez for “continually rehashing” the October incident with Strauss and noted that HR had already investigated that earlier incident.

She described the conversation as a “45 minute diatribe” and said: “There were strong undertones of racism and lawsuits, and he even asked if I would like to talk to his lawyer and that he was recording all conversations in his office.” Florez disputes Berry’s version of the January 2008 confrontation, again in ways that call into question the honesty of her account. He admits that he complained about Strauss’s evaluation from October 2007: But he testified during his deposition — and again, we accept as true for purposes of

summary judgment — that he never yelled at Berry and never made any threats about lawyers or the EEOC. He also testified in a declaration opposing

summary judgment that his conversation with Berry lasted only five to ten minutes, that he did not “continually rehash” the October 2007 incident, and that he never claimed to be recording conversations in his office. DeVry argues we must disregard this evidence because Florez’s declaration was dated only March 2012, with the blank for the specific day left blank, and because it contradicted his deposition testimony. DeVry does not cite any authority for the proposition that a court must exclude a sworn declaration because it does not specify the day it was sworn.

While including a specific date is the better practice, we are not convinced that having left the specific date blank required exclusion. DeVry’s assertion that Florez’s declaration contradicted his deposition testimony is the sort of assertion often made in sum-. mary judgment practice. In this case, the assertion is not supported by the record.

At his deposition, Florez was shown Berry’s written description of the January 2008 incident and was asked whether it was “an accurate account of the interaction.” He answered that the written statement was “inaccurate,” and he highlighted several specific statements with which he disagreed. The questioning moved on to other subjects. The first problem for DeVry’s argument is that Florez was not asked whether he had described all the inaccuracies with the written statement.

Without that question having been asked and answered to ensure that his deposition testimony exhausted his memory of the subject, his later declaration identifying other inaccuracies simply did not contradict any specific testimony in his deposition. Cf. Flannery v. Recording Industry Ass’n of America, 354 F.3d 632, 638 (7th Cir.2004) (explaining that a “contradiction” exists only when the statements are “inherently inconsistent,” not when the later statement “merely clarifies an earlier statement which is ambiguous or confusing”).

At a more fundamental level, DeVry’s argument reads too broadly the cases it invokes. DeVry cites Beckel v. Wal-Mart Associates, Inc., where we said that affidavits “when offered to contradict the affi-ant’s deposition are so lacking in credibility as to be entitled to zero weight in

summary judgment proceedings unless the affiant gives a plausible explanation for the discrepancy.” 301 F.3d 621, 623 (7th Cir.2002). Beckel and cases like it involved contradictions so clear that the only reasonable inference was that the affidavit was a sham designed to thwart the purposes of

summary judgment. See Bank of Illinois v. Allied Signal Safety Restraint Systems, 75 F.3d 1162, 1168-69 (7th Cir.1996) (“If such contradictions were permitted ... ‘the very purpose of the

summary judgment motion — to weed out unfounded claims, specious denials, and sham defenses — would be severely undercut.’ ”), quoting Babrocky v. Jewel Food Co., 773 F.2d 857, 861 (7th Cir.1985). This principle must be applied with great care, though, because

summary judgment is not a tool for deciding questions of credibility. See Bank of Illinois, 75 F.3d at 1169-70 (collecting cases). Few honest witnesses testify at any length without at least occasional lapses of memory or needs for correction or clarification. Disregarding as a sham every correction of a memory failure or variation in a witness’s testimony requires “far too much from lay witnesses” and would usurp the trier of fact’s role in determining which portion of the testimony was most accurate and reliable. Tippens v. Celotex Corp., 805 F.2d 949, 953-54 (11th Cir.1986).

That’s why we have said an affidavit can be excluded as a sham only where the witness has given “clear answers to unambiguous questions which negate the existence of any genuine issue of material fact.” Bank of Illinois, 75 F.3d at 1170, quoting Van T. Junkins & Associates, Inc. v. U.S. Industries, Inc., 736 F.2d 656, 657 (11th Cir.1984). The differences between Florez’s deposition testimony and his declaration fall far short of this standard.

The district court concluded that a reasonable trier of fact could not find that DeVry’s “volatile behavior” explanation was pretextual because Florez only “quarrels with ... the details concerning the clashes” without undermining the honesty of DeVry’s stated belief. Castro, 941 F.Supp.2d at 984. We respectfully disagree. The differences between Florez’s testimony and Strauss’s and Berry’s accounts of the October 2007 and January 2008 conversations are so basic that they are not obviously the stuff of honest disagreement.

And this is not a case where a supervisor had to decide which of two conflicting stories to believe; Berry herself made the decisive recommendation to fire Florez, on advice from Strauss. They relied on their own accounts — not reports .from co-workers or third-parties — to justify Florez’s termination. Cf. Johnson v. Nordstrom, Inc., 260 F.3d 727, 732 (7th Cir.2001); Flores v. Preferred Technical Group, 182 F.3d 512, 515 (7th Cir.1999).

If, as Florez contends, Strauss and Berry are lying about these events, then a reasonable trier of fact could find that Strauss and Berry fabricated their reports to create a false reason for terminating him. Under these circumstances,

summary judgment is not appropriate. See Testerman v. EDS Technical Products Corp., 98 F.3d 297, 303 (7th Cir.1996) (“ ‘If the only reason an employer offers for firing an employee is a lie, the inference that the real reason was a 'forbidden one ... may rationally be drawn.’ ”), quoting Shager v. Upjohn Co., 913 F.2d 398, 401 (7th Cir.1990). Even if this evidence of pretext were not enough for Florez, the inference of retaliation is strengthened further by testimony from Leal that Berry bribed her to say she did not see the October 2007 confrontation between Florez and Strauss by offering her a new office.

Evidence that Berry asked Leal to lie about the incident goes directly to the honesty of Berry’s belief that Florez behaved unprofessionally. Berry was one of the decision-makers in DeVry’s decision to terminate Florez, and she used this very incident as a basis to recommend his termination. If Florez did not behave the way Strauss and Berry have claimed, then a reasonable trier of fact could find that DeVry’s “volatile behavior” explanation is unworthy of credence. See Coleman, 667 F.3d at 852. The evidence of pretext is more than enough to require denial of

summary judgment on Florez’s retaliation claim. The district court held in the alternative that even if Florez could prove that Berry and Strauss fabricated these reports, a reasonable trier of fact would not be able to find causation because Florez failed to produce evidence linking DeVry’s decision to terminate him to his April 2007 complaint about Giambone. Castro, 941 F.Supp.2d at 984. Again, we respectfully disagree. To establish this link, Florez supplemented his evidence of pretext with an email from Hurt (the HR manager) to Maher (the HR director).

Hurt sent the email on February 4, 2008, the week after the confrontation between Florez and Berry. We quote the email in full because it is essential to our analysis: Mike Florez is an enrollment advisor. He’s been here since October 2005. We are having a number of problems with his performance. 1. His performance is not consistent (starts goals vs. actual for the last three classes were 9/11, 9/5, 3/5) 2. His behavior is volatile (I have several documented incidences of ‘blow ups’ where he has been, less than, cooperative with his supervisors) His last ‘blow up’ was with Kathy Berry last week.

To refresh your memory, he is the employee who was going to Daniel Hamburger and Dave Pauldine, because he did not feel he got resolution on the run-in he had with Julie Strauss a few months ago. He is also one of the people who complained about a previous supervisor (with Liz Castro). He is constantly insubordinate and challenging every decision his supervisor makes as racially motivated. Supervisors/managers feel that Mike’s negativity adversely impacts the team and would like to separate him at this time. His coming review will like[ly] be ‘Meets Standards’. I agree with the supervisors.

Let’s discuss when you have a moment. (Emphases added.) The February 4 email is striking for two reasons. First, it specifically referred to the activity protected by Title VII as a basis for recommending Florez’s termination. Second, the email made clear that Hurt considered Florez’s performance adequate (or at least probably adequate), which is in tension with DeVry’s long defense of his firing based on poor performance. As it happens, Maher did not respond to Hurt’s February 4 email.

Hurt waited until February 15; 2008 and then sent a follow-up email to Maher saying that she had reviewed Florez’s semi-annual performance evaluation. Although his performance “met standards,” she wanted to move forward with his termination. She explained in pertinent part: The totality of his review is that his performance is inconsistent, his inability to effectively follow the direction of his supervisors is a continuous problem and his refusal to accept management decisions will continue to impede his performance. While separating him is a risk, I feel comfortable that it is the right decision.

Maher still did.not respond. Four days later Hurt emailed Maher a third time: “Let’s discuss the email below when you have a moment.” Florez was terminated the following week, on February 21, 2008. The district court concluded that the February 4 email would not permit a reasonable trier of fact to find a causal link between the protected activity and Flo-rez’s termination for two reasons. Neither reason justifies the conclusion, at least for

summary judgment. First, the district court explained, Berry made the critical recommendation to terminate Florez and because there is no evidence that Berry knew about the protected activity, a reasonable trier of fact could not infer that Berry was motivated by retaliation. Contrary to the district court’s assertion, however, there is evidence in the record that Berry did in fact know about the protected activity before Florez’s termination. Berry testified in her deposition that she learned of plaintiffs’ complaint about Giambone “a couple months” after it happened.

Although Berry could not remember a specific date, she said it “could have been as much as four or five months” after she took over the Chicago office in July 2007. Berry Dep. at 28. Florez was terminated seven months after Berry took over the Chicago office, so under any version of the timing, it is reasonable to infer that she knew of plaintiffs’ protected activity before she recommended Florez’s termination. Even if this were not enough (and it is), there is evidence not only that Berry knew of the protected activity but also that De-Vry lied to conceal this fact from the EEOC.

In response to Florez’s EEOC charge, Hurt drafted a position statement saying that Florez’s termination had been initiated by Berry and that Berry “had no knowledge of’ Florez’s April 2007 complaint about Giambone. A reasonable trier of fact could find that this statement to the EEOC was deliberately false and could infer further that DeVry was lying to cover up a retaliatory motive. See, e.g., McInnis v. Alamo Community College District, 207 F.3d 276, 283 (5th Cir.2000) (reversing

summary judgment for employer where employee produced evidence that employer made false statements to EEOC in connection with termination). Neither DeVry in its response brief nor the district court in its opinion addressed this critical fact. The district court’s second reason was that the February 4 email did not suggest a retaliatory motive because Hurt testified that she included the reference to the April 2007 complaint merely because it represented a risk of litigation if DeVry went ahead with terminating Florez.

DeVry argued in the district court, as it does on appeal, that when read in context with Hurt’s later email from February 15, which mentions that terminating Florez was a “risk,” the February 4 email does not suggest pretext. At most, DeVry contends, it was just “practical advice” from the HR Department about discharging Florez. That’s one way to read the emails, but it’s also reasonable to read them as signaling retaliatory intent. In accepting De-Vry’s argument, the district court drew an inference against the non-moving party, Florez, and in favor of the moving party, which is of course not appropriate at the

summary judgment stage. See, e.g., Hutchens v. Chicago Board of Education, 781 F.3d 366, 373-74 (7th Cir.2015) (reversing

summary judgment for employer where reasonable trier of fact could either accept employer’s stated reason as true or find that it was pretext for discrimination); accord, e.g., Lewis v. City of Chicago, 496 F.3d 645, 651 (7th Cir.2007) (“The district court erred by improperly weighing the parties’ evidence during

summary judgment.”). Here, each side offers a permissible

interpretation of the February 4 email. Nothing in the record makes Flo-rez’s

interpretation unreasonable. In fact, the February 4 email did not mention “the risk of litigation.” It did, however, list reasons to terminate Florez. The statement that Florez was “one of the people who complained about a previous supervisor” appears in the same paragraph as the statement that he had a “blow-up” with Berry — the disputed factual basis for terminating him. A reasonable jury could interpret the February 4 email as listing the reasons to terminate Florez, with the protected conduct among them.

In light of both DeVry’s concession on appeal that Florez’s performance did not justify his termination and the evidence casting doubt on the honesty of the “volatile behavior” explanation, a reasonable trier of fact could find DeVry’s stated reasons unworthy of belief and conclude that unlawful retaliation is the more likely explanation for its decision. See Forrester v. Rauland-Borg Corp., 453 F.3d 416, 418 (7th Cir.2006) (“if the stated reason, even if actually present to the mind of the employer, wasn’t what induced him to take the challenged employment action, it was a pretext”).

In sum, Florez has produced evidence from which a reasonable trier of fact could find that DeVry fired him because he had complained to HR about Giambone. VI. Brooks’s Evidence of Pretext Like Florez, Brooks attempts to prove retaliation by showing that DeVry’s stated reasons for firing her fifteen months after the HR complaint were pretexts. According to DeVry, Brooks was terminated for two reasons: (1) multiple acts of dishonesty and (2) inconsistent performance. DeVry contends that Brooks lied about her activities during several work days and failed to meet important performance benchmarks during 2008.

DeVry cites two events to support its dishonesty explanation. First, in October 2007 Brooks told Strauss that she needed to leave the office at 7:00 p.m. to pick up an application fee from a student. Strauss later conducted a routine quality assurance call with the student. She was told that Brooks had actually picked up the deposit earlier in the afternoon. When Strauss confronted Brooks about the discrepancy, she denied that she had been dishonest with Strauss and explained that there had been an honest miscommunication involving the student’s grandmother.

Strauss told Brooks that “honesty is the best policy” but did not reprimand her further. The second event occurred the following summer. On June 17, 2008, Brooks told Strauss that she would be stopping on her way to work to visit a prospective student. When Brooks still had not shown up at work two hours later, Strauss called the student to see if Brooks had stopped by. He said no. That same day Brooks had also told Strauss that she spent the morning calling several prospective students while working from home. Berry conducted quality assurance calls with these prospective students.

They told her they had not spoken with Brooks that day. Based on the two incidents, Strauss believed that Brooks had been dishonest with her. She wrote a report saying that “La-Tonya [Brooks] has repeatedly lied and is not on task. LaTonya’s production is slipping tremendously....

It is clear LaTonya is not focused on her job and her responsibilities.” Strauss also connected these incidents of dishonesty to Brooks’s performance, noting that her enrollment numbers for the July 2008 admissions class were “decreasing ... by the hand fulls.” Strauss addressed other concerns with Brooks in two other write-ups, both dated June 24, 2008. She described several acts that were, in Strauss’s view, examples of insubordination, disrespect, and a failure to cooperate with management. The next day, Strauss sent her write-ups to Hierl and Berry.

Hierl then sent an email to Hurt with Strauss’s write-ups attached. Hierl told Hurt that she wanted either to terminate Brooks immediately or to put her on month-to-month probation resulting in her termination at the end of the July 2008 énrollment period. Thirty minutes later, Hurt emailed back: “Can you send me a copy of her last review and the improvement plan mentioned?” There is no evidence that Hierl responded to. this email.

On July 1, 2008 Hurt emailed Maher (the HR director) about Brooks’s “inconsistent” performance, summarizing her “starts” for the admissions classes in January, March, May, and July of 2008. De-Vry’s “start” targets are like sales quotas: “starts” refer to the number of students recruited by an admissions officer who enroll in a given class. After listing several enrollment periods for which Brooks failed to achieve her start targets, Hurt’s email concluded: “We really need a more solid performer, especially for a Level II advis- or. Would like to separate from the organization.

We could live with a month-to-month probation, but we would be .probably prolonging the inevitable.” When the July.2008 enrollment period ended, Brooks had missed her starts target for the class. This was the same enrollment period Strauss had emphasized in her write-up about the June 17, 2008 incidents regarding Brooks’s dishonesty. DeVry fired Brooks a few days later, on July 8, 2008. Brooks argues that both of DeVry’s justifications were pretexts.

She contests the dishonesty explanation in two ways: (1) she denies that she was ever dishonest with Strauss, and (2) she argues that because the July 1, 2008 email recommending her termination mentioned her performance but not her dishonesty, DeVry’s reasons for her termination are “shifting” and “inconsistent.” The performance explanation was a pretext, she says, because Hurt provided “false” numbers in the July 1 email to Maher when describing her starts for two enrollment periods. We begin with the dishonesty rationale.

A reasonable trier of fact could not find on this record that this explanation was pre-textual. Brooks disputes that she was actually dishonest with Strauss, suggesting that any discrepancies discovered by Strauss should be chalked up to innocent miscommunications. Unlike Florez in his disputes with Strauss and Berry, though, Brooks does not dispute that Strauss honestly believed Brooks had been dishonest with her. Brooks also does not dispute that Strauss reported these incidents to Berry and Hierl, who in turn relayed the reports to Hurt and Maher.

Brooks’s dispute about whether she was actually dishonest with Strauss therefore does not bear on the decisive question: Did Strauss, Berry, Hierl, Hurt, and Maher honestly believe Brooks had been dishonest when they made the decision to fire her? Cf. O’Leary, 657 F.3d at 635. Brooks attempts to fill this gap in her theory by arguing that DeVry has given “shifting” and “inconsistent” explanations for her termination because Hurt’s July 1 email did not mention dishonesty.

As a general rule, a reasonable trier of fact can infer pretext from an employer’s shifting or inconsistent explanations for the challenged employment decision. See, e.g., Hitchcock v. Angel Corps, Inc., 718 F.3d 733, 738 (7th Cir.2013). The record here would not permit such an inference, though. Hurt’s decision to emphasize the performance rationale in her email to Maher did not create an inconsistency. She did not disavow or contradict the dishonesty justification; she simply failed to mention it. In fact, Strauss’s initial write-ups about Brooks raised both performance issues and concerns about dishonesty.

Strauss’s write-up about the June 17, 2008 incidents involving both a student and Brooks’s claim that she had been working from home that morning explicitly linked the dishonesty rationale to the declining number of students she had enrolled for the July 2008 admissions class and noted her “slipping” production. It is undisputed that Berry, Hierl, Hurt, and Maher relied on this write-up to justify terminating Brooks. Thus, the

summary judgment record establishes beyond reasonable dispute that DeVry relied on both explanations to terminate Brooks. Where an employer relies on multiple reasons for the termination, its failure to address all of the reasons in each communication about the employee is not enough to show contradictions or shifts in rationales that suggest pretext. See Johnson v. Nordstrom, Inc., 260 F.3d 727, 733-34 (7th Cir.2001) (employer’s supplementation of reasons for adverse employment decision not evidence of pretext so long as reasons do not conflict and employer does not retract a reason); O’Connor v.

DePaul University, 123 F.3d 665, 671 (7th Cir.1997) (employer’s “flux in terminology” not evidence of pretext where three explanations “focused on different aspects of [employee’s] behavior” but same underlying conduct). Brooks next attacks DeVry’s performance rationale. Recall that Hurt’s July 1 email to Maher summarized her starts targets for the admissions classes in January, March, May, and July of 2008.

According to Hurt’s email, Brooks missed her target for three of these classes: January (she enrolled one student against a goal of three), May (zero students against a goal of two), and July (sixteen students against a goal of twenty). Brooks concedes that she missed her starts target for the July class. But she contends that she made her targets for the January and May classes and that two of the numbers in Hurt’s email are therefore false. She argues that this evidence is sufficient to raise a genuine issue of material fact on whether DeVry’s performance explanation was pretextual. See Reeves v.

Sanderson Plumbing Products, Inc., 530 U.S. 133, 147, 120 S.Ct. 2097, 147 L.Ed.2d 105 (2000) (“In appropriate circumstances, the trier of fact can reasonably infer from the falsity of the explanation that the employer is dissembling to cover up a discriminatory purpose.”). Brooks’s argument runs into an eviden-tiary problem. The only evidence she offers to support this theory are two documents produced by DeVry in discovery. One document suggests that Brooks enrolled six students for the January class; the other suggests that she enrolled four students for the May class.

If these documents accurately reflect final starts numbers, then Brooks achieved her targets for the January and May classes, contrary to Hurt’s claim in the July 1 email. The district court excluded this evidence, however, because the documents were not properly authenticated. Castro, 941 F.Supp.2d at 987 & n. 10. Plaintiffs argue that the district court should have admitted the documents as business records under Federal Rule of Evidence 803(6) for two reasons: (1) the documents were properly authenticated by Castro’s sworn declaration, and

(2) DeVry’s production of the documents in the litigation served as implicit authentication under Thanongsinh v. Board of Education, 462 F.3d 762, 777-78 (7th Cir.2006). We review only for abuse of discretion a district court’s ruling on the admissibility of evidence on

summary judgment. Bradley v. Work, 154 F.3d 704, 708-09 (7th Cir.1998); accord, e.g., Makowski v. SmithAmundsen LLC, 662 F.3d 818, 822 (7th Cir.2011). The district court did not abuse its discretion in excluding this evidence. We agree with the district court that Castro’s declaration did not lay a sufficient foundation to authenticate the documents. Rule 803(6) requires authentication by a “custodian or another qualified witness.” Castro’s declaration did not establish that she was a custodian or otherwise qualified witness. It merely stated: “Attached as Ex.

F is a document generated by DeVry in the normal course of business reflecting registered students by advisor for January of 2008. Ex. F is a document produced by DeVry in this litigation on a disc.” Castro’s declaration includes an identical statement about the second document, which was offered to show Brooks’s starts for the May 2008 class. Brooks contends that Castro could properly authenticate these documents because she had been a “supervisor” who was qualified to opine about DeVry’s performance evaluations. But that’s not enough.

To lay a proper foundation for admitting this evidence under Rule 803(6), Castro needed to be familiar with DeVry’s record-keeping practices. See, e.g., Joseph P. Caulfield & Associates, Inc. v. Litho Productions, Inc., 155 F.3d 883, 888 (7th Cir.1998) (no abuse of discretion where district court excluded evidence because foundational witness did not testify as to the business’s “regular record-keeping practices”). Her declaration was silent on this issue. The fact of production in the litigation says nothing about Castro’s familiarity with DeVry’s record-keeping practices.

Plaintiffs’ reliance on Thanongsinh is also unpersuasive. In that case, we identified a narrow exception to the rule that admission under Rule 803(6) at

summary judgment requires an affidavit from a custodian or other person familiar with the type of record. The exception applies “when the party challenging the document’s admissibility relied on that same document ‘for its accuracy’ in earlier proceedings, or otherwise ‘conceded the accuracy-of the documents that the [opposing party] sought to introduce.’ ” Thanongsinh, 462 F.3d at 778 (alteration in original), quoting Woods v. City of Chicago, 234 F.3d 979, 988 (7th Cir.2000). The exception does not apply here. DeVry has not relied on these documents for their accuracy.

Nor has it conceded the accuracy of the starts numbers reflected in the documents. In Thanongsinh, by contrast, the party opposing admission had “admitted in discovery that the [exhibit] is what the plaintiff purports it to be.” Id. No such admission exists here. The mere act of producing a document in response to a discovery request based on the content of the document does not amount to an admission of the document’s authenticity. A party’s duty to produce documents under Federal Rule of Civil Procedure 34(

a) applies to responsive documents in its “possession, custody, or control.” They must be produced regardless of their authenticity, accuracy, or reliability, so the act of production does not say anything about authenticity, accuracy, or reliability. Those are matters for followup requests for admissions or other discovery tools. Without those two unauthenticated documents, no evidence in the

summary judgment record casts doubt on DeVry’s performance rationale. This is especially true in light of Brooks’s concession that she missed her starts target for the July 2008 class — the enrollment period immediately preceding her termination and the only class mentioned in Strauss’s decisive writeup. A reasonable trier of fact could not find pretext as to Brooks. Accordingly, Brooks has failed to produce evidence that would permit a reasonable trier of fact to find that DeVry fired her because she had complained to HR about Giambone. VII.

Castro’s Evidence of Pretext Castro also attempts to demonstrate that DeVry’s stated reason for terminating her was a pretext for retaliation for the HR complaint thirty months earlier. According to DeVry, Castro was terminated because of poor performance over a sustained period. It is undisputed that Castro failed to achieve several monthly performance goals in 2008 and 2009. She was ultimately fired on November 3, 2009.

For the September 2009 enrollment period, she achieved only three starts against an original target of twenty, which had been reduced to ten for Castro’s benefit because she had taken a leave of absence. Castro conceded during her deposition that three starts for the September 2009 class “was not a good number.” Nevertheless, Castro argues that De-Vry’s stated reason was pretextual.

Citing her success in achieving several monthly performance goals in 2008 and 2009, Castro argues that her failure to meet other performance goals during this span was caused by her leaves of absence and not having her targets adequately prorated. She also argues that she was terminated prematurely, before she had a meaningful opportunity to satisfy her starts target for the November 2009 class. Castro has failed to establish a genuine issue of material fact regarding pretext. Her ability to satisfy some of her monthly performance goals does not raise a genuine issue.

DeVry has not claimed that Castro failed to satisfy all of her performance targets. DeVry acknowledges that she achieved some of her goals in 2008 and 2009, but it explains that Castro’s performance was too inconsistent to justify keeping her with the company. Inconsistent performance — though not uniformly bad— is a legitimate, non-retaliatory reason for termination. See Roberts v. Separators, Inc., 172 F.3d 448, 451-52 (7th Cir.1999) (“poor performance” was legitimate, nondiscriminatory reason where plaintiffs “performance declined after an initial period of excellence”).

And because DeVry has not denied that Castro achieved some of her performance goals during the relevant span, a reasonable trier of fact could not find based on this record that DeVry is dissembling. Castro fares no better with her argument that she failed to achieve certain monthly performance goals because her targets were not adequately prorated. It is undisputed that DeVry reduced several of her targets for her benefit. In fact, the goal she missed just before she was fired-had been reduced from twenty to ten, and Castro still missed it.

Castro’s subjective belief that her targets should have been prorated more than they were does not undermine the honesty of DeVry’s stated explanation for terminating her and thus could not support a finding of pretext. See Ineichen v. Ameritech, 410 F.3d 956, 961 (7th Cir.2005) (“it is not ‘the court’s concern that an employer may be ... too hard on its employee’ ”), quoting Ransom v. CSC Consulting, Inc., 217 F.3d 467, 471 (7th Cir.2000); accord, Silverman v. Board of Education of City of Chicago, 637 F.3d 729, 738 (7th Cir.2011) (collecting cases).

Castro also argues that DeVry’s stated reason for firing her was a pretext because she did not have a meaningful opportunity to achieve her starts goal for the November 2009 admissions class. DeVry concedes that Berry recommended Castro’s termination on October 26, 2009, before the sign-up period for the November 2009 class officially ended. DeVry maintains, however, that the decision was reasonable because, in light of Castro’s pace at enrolling students during October, she could not possibly enroll nineteen students in less than a week.

Even assuming, as Castro contends, that DeVry jumped the gun in concluding that she would not be able to hit her target by the end of the sign-up period, this evidence does not establish pretext. It does not undermine the honesty of DeVry’s belief. Even if that belief had been unreasonable or overly harsh, there is no evidence that it was dishonest. See, e.g., Ineichen, 410 F.3d at 961; Hartley v. Wisconsin Bell, Inc., 124 F.3d 887, 890 (7th Cir.1997) (an employer’s “foolish or trivial or even baseless” explanation suffices, so long as it is “honestly believed”).

Castro’s argument overlooks the fact ■ that DeVry had already placed her on probation because she had repeatedly failed to hit performance targets over a sustained period. After Castro enrolled only three students for the September 2009 class against a prorated goal of ten, Berry placed her on probation.

In a memo dated August 27, 2009, Berry told Castro: “During this probation period, failure to meet or make substantial progress towards the target and weekly activity levels outlined below may result in termination at any time’ during this period.” The memo then outlined “weekly activity mínimums” of 6 appointments, 4 interviews, and 2.25 applications. Castro does not dispute that she failed to achieve these “weekly activity mínimums.” Her focus on the total starts goal for the entire enrollment period is therefore misplaced.

Ultimately then, Castro has failed to produce evidence from which a reasonable trier of fact could find that DeVry’s reason for firing her was pretextual. We return finally to the evidence that shortly after the July 2007 meeting Strauss told Florez and Brooks that “the people that have went to HR no longer work here.” A reasonable trier of fact could infer based on this record that Strauss was referring to Castro. Castro had been demoted and transferred from Chicago to the Addison office shortly before the comment was made and within three months of the complaint to HR.

A reasonable trier of fact could also interpret the comment as evidence of retaliatory intent — that it was a signal to Florez and' Brooks that Castro was demoted and transferred because she complained to HR and that they might suffer the same fate if they complained again. This evidence takes Castro only so far. She has not based her retaliation claim on her demotion and transfer. She has complained only about her termination. The critical question, then, is whether Strauss’s comment to Florez and Brooks would permit a reasonable finding that DeVry terminated Castro because of her complaint about Giambone.

Strauss’s comment would not permit such a finding. The comment was made more than two years before Castro’s termination. As we have explained, the record establishes beyond reasonable dispute that Castro had a long history of performance problems leading up to her termination. Absent other evidence of retaliation, this comment does not create a genuine issue for trial. See Geier v. Medtronic, Inc., 99 F.3d 238, 242 (7th Cir.1996) (“To be probative of discrimination, isolated comments must be contemporaneous with the discharge or causally related to the discharge decision making process.”).

Finally, plaintiffs argue that if any one of them has raised a genuine issue as to retaliation, then

summary judgment should be reversed as to all of them. They all engaged in the same protected activity, and all suffered the same penalty of termination imposed by the same key decision-makers. While Florez has raised a genuine issue of material fact about the reasons for his termination, we do not believe that finding carries over to Brooks or Castro. Florez has offered evidence that would allow a reasonable jury to find that Strauss, Berry, and other DeVry managers retaliated against him.

That evidence allows a finding that those managers were capable of retaliatory motive in response to the April 2007 HR complaint. But the three plaintiffs were treated in different ways, at different times spanning nearly two years. The undisputed evidence of performance problems for both Brooks and Castro persuades us that a reasonable jury could not find that the later and separate decisions to fire them were motivated by retaliation for the 2007 HR complaint. Accordingly, we AFFIRM the district court’s judgment in favor of DeVry on all of plaintiffs’ claims except Florez’s claim of retaliation.

On that claim, we REVERSE and REMAND for further proceedings consistent with this opinion. . Plaintiffs offer evidence that Giambone had „ a history of making offensive racial and ethnic remarks in the office. For example, he regularly asked about the race or ethnicity of prospective students and connected race or ethnicity to the ability to pay an application fee.

On one occasion, for instance, he commented that a Mexican-American prospective student would be able to pay the fee because "Mexicans always have $100 in their mattresses because it's communion money.” He also at times remarked that an applicant who could not pay the fee was "probably Puerto Rican or black” and that Puerto Rican and black people were "always broke.” We accept this evidence as true for purposes of

summary judgment. . Cf. Hasan v. Foley & Lardner LLP, 552 F.3d 520 (7th Cir.2008) (reversing

summary judgment for employer; circumstantial evidence suggested discrimination in spite of year-long period between discriminatory comment and termination); Lang v. Illinois Dep’t of Children & Family Services, 361 F.3d 416 (7th Cir.2004) (reversing

summary judgment for employer; convincing mosaic of circumstantial evidence of retaliation where employee began receiving reprimands shortly after filing grievance, culminating in his termination approximately one year later); Veprinsky v. Fluor Daniel, Inc., 87 F.3d 881, 891 n. 6 (7th Cir.1996) (recognizing principle that

summary judgment is inappropriate when record would permit reasonable trier of fact to find that employer "waited in the weeds” for years looking for an opportunity to fire employee). . We recognize that Strauss's comment was backward-looking in one respect: a reasonable trier of fact could infer that Strauss was referring to Castro, who had been- transferred from the Chicago office to Addison shortly before the comment was made. Because this

interpretation of the comment applies only to Castro, we address it below when analyzing Castro's remaining evidence of retaliation. . DeVry does not explain why it might have been justified in disciplining Florez because he "made threats about lawyers and the EEOC." In fact, such a "threat” might well qualify as protected activity. E.g., Davis v.

Time Warner Cable of Southeastern Wisconsin, L.P., 651 F.3d 664, 674 (7th Cir.2011) (informal complaints can constitute protected activity for purposes of Title VII retaliation claim); accord, 42 U.S.C. § 2000e-3(a) (making it unlawful to retaliate against anyone “because he has opposed any practice” prohibited by Title VII). Florez denies having made any statement about lawyers or the EEOC, however, so we do not examine this issue further. .

The statement in the February 4 email that Florez "is constantly ... challenging every decision his supervisor makes as racially mo-' tivated” sounds like a reference to other complaints that might have qualified as protected activity under Title VII. Florez does not make this argument, however. He relies exclusively on the April 16, 2007 complaint about Giambone as the protected activity in this case. Id. at 487-88. There, the employee had failed to offer any evidence linking the employer's "(alleged) desire to lie to the EEOC investigator” with the adverse employment decision. Id. at 488.

Here, by contrast, Florez has offered the February 4 email written by the same person who drafted the false statement to the EEOC. That email, unlike the evidence in Rabinovitz, explicitly linked the adverse employment action to the protected activity. . Rabinovitz v. Pena, 89 F.3d 482 (7th Cir.1996), is not to the contrary. In Rabinovitz, we held that an employer’s false statement to the EEOC that it did not know the employee was Jewish did not create a genuine issue of material fact on the employee’s religious discrimination claim. . DeVry claims that Berry made the ultimate decision to terminate Brooks.

Appellee’s Br. 16-17. For this proposition, it cites Dkt. 49-7 at 120:8-17 (Berry's deposition); Dkt. 49-13 at 135:9-136:4 & 136:24-137:5 (Hurt's deposition); and Dkt. 49-14 at 11:1-15 (Maher's deposition). The deposition excerpts do not support this proposition. In fact, the excerpted transcript of Berry’s deposition that DeVry cites ends with page 113. There is no page 120. Based on the

summary judgment record, a reasonable trier of fact could find that Strauss, Berry, Hierl, and Hurt all participated in the decision to terminate Brooks. . Woods v. City of Chicago, 234 F.3d 979 (7th Cir.2000), provides a useful contrast. There, the plaintiff attached a police case report to his response to the defendants' motion for

summary judgment. The report summarized the contents of an arrest record and a misdemeanor complaint. In opposing

summary judgment, the plaintiff argued that the case report accurately described those two underlying documents. Id. at 988. The district court granted

summary judgment to defendants, relying in part on the underlying arrest record and misdemeanor complaint. On appeal, the plaintiff argued that the district court abused its discretion by considering this evidence because neither the arrest record nor the misdemeanor complaint was admissible under Rule 803(6). We agreed with the plaintiff that neither document had been properly authenticated, but we affirmed the district court’s decision anyway because the plaintiff had conceded the accuracy of the underlying documents by relying on the case report that summarized them.

We explained that requiring authenticating affidavits under the specific circumstances of that case “would be an empty formality” and concluded that the district court therefore did not abuse its discretion when it considered the documents without such affidavits. Id. at 989. . Castro cites evidence that Casey Tobin (her direct supervisor) "whited out” two students’ names from her list of starts for the July 2009 enrollment period. Castro argues that this evidence would permit a reasonable trier of fact to find that DeVry falsified her performance numbers in an effort to terminate her.

Even accepting this characterization of the record as true, it does not establish a genuine issue on pretext because DeVry did not justify her termination based on her performance in July 2009. As Castro conceded in the district court, her mid-year 2009 review said that she "made her starts goal” for that period. De-Vry has never claimed otherwise. . Castro asserts that “no one on Tobin’s team, except Tobin” achieved their September 2009 starts goals.

But these purported comparators all had at least seven starts for the period, and Castro does not attempt to demonstrate that they were situated similarly to her. .Castro could in theory establish pretext by producing evidence of disparate treatment in the application of the company’s proration policies. Suppose, for example, that Castro and a similarly situated employee took leaves of absence of the same length and, upon returning to the office, the comparator’s targets were reduced by 50 percent while Castro’s targets were reduced by only 10 percent.

See, e.g., Coleman, 667 F.3d at 858 (explaining that selective enforcement of company policy can establish pretext); Gordon v. United Airlines, Inc., 246 F.3d 878, 892 (7th Cir.2001) (same). But the record here does not fit this pattern. There is no evidence that employees at DeVry were entitled to proration under the company's policies. And Castro has not produced evidence' that a similarly situated employee was treated better in this respect.

In fact, she testified during her deposition that whether DeVry would prorate an employee’s targets was a matter of "discretion” and that there was "no magic number” for determining by how much an employee’s targets should be reduced. . Castro points out that when she was terminated on November 3, no other advisor had more than one start for the November 2009 class. She makes no attempt, however, to demonstrate that the other advisors were situated similarly to her.

For example, she does not present evidence that any of these employees had a comparable history of past performance problems or were already on probation as she was at the time.