WOOD, Chief Judge.

Richard L. Darst, Attorney, Cohen, Garelick & Glazier, Indianapolis, IN, for Plaintiff–Appellant.

Jan Michelsen, Attorney, Ogletree, Deakins, Nash, Smoak & Stewart, P.C., Indianapolis, IN, for Defendant–Appellee.

Before WOOD, Chief Judge, FLAUM, Circuit Judge, and KENNELLY, District Judge.

Opinion

WOOD, Chief Judge.

Chontel Miller began work as a sample processing technician at Polaris Laboratories in Indianapolis in August 2009. Judging from reports of her daily productivity, her performance left something to be desired. In April 2010, Miller was fired for repeated failures to meet an average daily quota of 260 samples processed per day. Yet that is not the whole story. Miller, who is African–American, asserts that during her employment at Polaris she suffered from racial discrimination that had an effect on her work performance. She sued Polaris, contending that it violated Title VII as well as 42 U.S.C. § 1981 in several ways. The district court granted summary judgment to Polaris. There is no doubt that this is a close case. But construing the record in Miller's favor, as we must do, we conclude that Miller has shown a genuine issue of material fact on both her discrimination and retaliation claims. We therefore return the case to the district court for further proceedings.

I

Our account of the facts comes from the evidence Miller presented at summary judgment, taken in the light most favorable to her. Miller started her job as a sample processor at Polaris on August 17, 2009; she was the only African–American in her area. Her task was to review samples of various fluids that customers submitted and record information about them in a computer database. Miller's direct supervisor was Rhonda Ballard; Ballard's supervisor was Debbie New. Polaris expected Miller to process an average of 260 samples a day, but she had difficulty meeting that number. In February 2010, Ballard wrote a performance appraisal report for Miller analyzing her work from her first day through the end of 2009. Miller's average of 22 samples logged per hour, Ballard wrote, was “much lower than we would like to have her logging,” although Ballard acknowledged that Miller had “been given several challenging customers and has adequately logged those samples.” Her daily average of 123, Ballard wrote, was far below the required number. Ballard concluded the appraisal by giving Miller the goal of 260 samples per day on average by the end of March 2010. Miller did not meet this goal.

Not long after Miller began working at Polaris, Miller learned from her fellow processor Amanda Saperstein that someone in her department—Saperstein was not sure who—had referred to Miller as “the colored girl.” (Miller testified that Saperstein told her that Ballard called her “the colored girl.”) That day, September 18, 2009, Miller had an argument with another processor, Sharon Holmes, according to a report that New filed a few days later. The situation apparently escalated. Miller complained in person to New's supervisor, Joe Culp, that no one had been helping her finish her work that day, that Holmes, Ballard, and another processor named Gina Kemp were racists, and that someone had called her “the colored girl.” Culp referred Miller's complaint to the company's department of human resources, which interviewed Miller, Saperstein, Ballard, and other employees. The investigation did not resolve who made the “colored girl” comment. From that point forward Ballard refused to train and even to talk to Miller. Miller complained several times to New about Ballard.

While Miller worked at Polaris, at least one employee aside from Miller heard other workers use racially derogatory language when referring to Miller. A sample processor named Bobbie Jo Young testified that Kemp was discussing Miller's poor performance at work with Ballard and said, “It wouldn't bother me if that stupid nigger bitch did not come back here.” Young said that when Kemp made this comment, Ballard “laughed and said, [']I'm in that same boat with you.[']” Young also witnessed Kemp “mix[ing] up a tray before she gave it to Chontel,” and Ballard giving “Chontel trays that were more difficult that [sic ] those she gave to other [sic ] white employees.” In particular, Young discussed an incident that occurred after the September 18 incident between Miller and Ballard. In that instance, Young testified, Ballard told Young not to take a particular tray because Ballard intended it for Miller, telling Young, “[D]on't you see what I'm trying to do here?” That tray, Young said, “was the hardest tray” because it had samples from “28 different customers”; a tray containing samples from multiple customers “really slows you down.” (Young also testified that she understood Ballard to be saying she was trying to make Miller's work harder.) In another instance, Young witnessed Kemp “unshuffling” samples on a tray intended for Miller and “taking them out of order.” Once Kemp realized that Miller could see her, Young testified, Kemp went back to the tray and “fixed” it, announcing that she knew “that bitch [Miller] is going to go complain.”

Miller testified that she spotted Ballard changing the composition of the samples Miller was to process in order to make her job more difficult. On one occasion when she came back early from lunch, Miller said, she “caught [Ballard] moving my samples around in my trays.” At other times she “would end up with ... just out-of-the-blue samples that was [sic ] not there” before she went to lunch. On a couple of other occasions, Miller caught Kemp and Ballard putting samples into Miller's tray that were missing labels. This too slowed down Miller's work, because she had to research the samples before she could log them into Polaris's systems. Otherwise, Miller testified, she was given samples from Polaris accounts that took more time to work on because of the paperwork involved, or because the trays included samples from multiple customers. Sometimes, Miller said, she would resort to choosing easier samples on her own while Ballard was at lunch in order to try to boost her numbers.

Miller's employment at Polaris lasted about eight months; at no time did she meet Polaris's productivity goals, although there was some improvement. She completed a general probationary period in February 2010, but in April 2010, she was again placed on probation for poor performance. The April notice, which listed Ballard as Miller's “appraiser,” pegged Miller's average number of samples logged at 189. “Chontel must speed up her logging” by the end of April, Ballard wrote. Miller's monthly averages for each full month on the job were 91, 135, 150, 149, 158, 161, 189, and 184; her overall daily average was 147. Miller exceeded the required 260 number only once: her second-to-last day on the job, when she logged 288 samples. Two days later, on April 30, 2010, before Miller's April probation technically had ended, New along with chief operating officer Mark Minges and human resources director Chad Ziegler terminated Miller's employment. They cited her inadequate production numbers as the reason. In July 2011 Miller filed a complaint in the Southern District of Indiana, alleging race discrimination and retaliation for complaining about discrimination.

The district court initially granted only partial summary judgment for Polaris. It denied summary judgment to Polaris on Miller's discrimination claim because it believed that Miller had presented sufficient evidence under a cat's paw theory of a causal connection between the racial bias of Kemp and Ballard and the decision of Polaris management to fire Miller. It granted summary judgment for Polaris on Miller's retaliation claim, concluding that Miller had no evidence indicating that she was treated poorly because of her complaints of discrimination, nor that the stated reason for her firing—her poor performance—was pretextual. Polaris moved for reconsideration of the court's decision on Miller's discrimination claim, arguing that Miller had no evidence showing that manipulation of her work product by Ballard and Kemp “could systematically reduce” her output over eight months of employment. The district court was persuaded and ruled in Polaris's favor on all counts; this appeal followed.

II

We turn first to Miller's employment discrimination claim, which she brings under both 42 U.S.C. §§ 2000e–2 and 2000e–5 and 42 U.S.C. §§ 1981 and 1981a.

A discrimination claim under Title VII or Section 1981 requires a plaintiff proceeding under the direct method of proof to raise an inference through direct or circumstantial evidence that discrimination motivated an adverse employment action. Langenbach v. Wal–Mart Stores, Inc., 761 F.3d 792, 802 (7th Cir.2014) ; see also Mintz v. Caterpillar Inc., 788 F.3d 673, 679 (7th Cir.2015) (same standards apply in Title VII and Section 1981 claims). Miller has not offered any evidence that the actual decisionmakers at Polaris who decided to terminate her employment—Culp, New, and Minges—harbored discriminatory feelings toward her. She relies instead on the so-called cat's paw theory of liability, the name of which is derived from a fable we have recounted before. See, e.g., Smith v. Bray, 681 F.3d 888, 897 n. 3 (7th Cir.2012). Liability for discriminatory action under this theory is premised on the ability of the culpable actor to use an innocent decisionmaker as an “unknowing tool” of the former's animus. Hutchens v. Chicago Bd. of Educ., 781 F.3d 366, 373 (7th Cir.2015) ; see also Staub v. Proctor Hosp., 562 U.S. 411, 422, 131 S.Ct. 1186, 179 L.Ed.2d 144 (2011) (“[I]f a supervisor performs an act motivated by [discriminatory] animus that is intended by the supervisor to cause an adverse employment action, and if that act is a proximate cause of the ultimate employment action, then the employer is liable.” (citation and footnote omitted)). An employer is liable under Title VII or Section 1981 when “a non-decision-making employee with discriminatory animus provided factual information or input that may have affected the adverse employment action.” Matthews v. Waukesha Cty., 759 F.3d 821, 829 (7th Cir.2014). In cat's paw cases, we regard the biased subordinate's actions as direct evidence of discrimination. Nichols v. Mich. City Plant Planning Dep't, 755 F.3d 594, 600 (7th Cir.2014).

The present case is one in which the reason for firing the employee is straightforward at first glance. Miller was supposed to log an average of 260 samples per day; she consistently came up short; and so she was fired. End of case, Polaris says: its decision to fire Miller “was the direct consequence” of her poor performance. Miller does not fight the numbers; she says instead that she was set up to fail, for racially discriminatory reasons. Ballard and Kemp, she contends, acting upon their racial animus, intended to cause (and did cause) Miller's termination through tampering with her work, among other tactics. The nominal decisionmakers at Polaris accepted the numbers unquestioningly, unaware that Miller's ability to meet her quota had been sabotaged.

Miller cannot prevail on this theory unless she can show a dispute of fact on the question whether Ballard and Kemp's actions were motivated by racial animus. Construing the facts in Miller's favor, we conclude that she has passed that hurdle. There is evidence in the record from both Miller and Miller's coworker Young that Ballard and Kemp displayed discriminatory feelings toward Miller. Kemp's “stupid nigger bitch” comment obviously qualifies, considering Young's testimony that it could only have been intended for Miller, the sole African–American processor in the workplace and the person whose performance Miller and Ballard were discussing at the time. According to Young, Ballard explicitly concurred with this statement, laughing and saying that she was “in the same boat” with Kemp. Polaris's sole response to this allegation at oral argument was to contend that Miller was not aware of this comment at the time. That may be so, but it has nothing to do with whether Ballard and Kemp actually harbored discriminatory animus. There is also the “colored girl” comment, which Miller attributes to Ballard. Although Saperstein, the coworker who told Miller about the comment, testified that she did not know who said it, Miller herself testified that Saperstein reported Ballard as the speaker. This evidence is sufficient to show a dispute of fact as to whether Ballard and Kemp displayed (and thus possessed) racial animus toward Miller.

The more difficult question is whether Ballard and Kemp took actions that proximately caused Miller's termination. Miller insists that her inability to hit the 260 mark during her employment at Polaris was the result of Ballard's sabotage of her work as well as Ballard's refusal to help and train her. Polaris responds that even if Ballard sabotaged Miller's work on certain occasions and failed to train her, that alone cannot possibly explain the consistently low numbers. In other words, Polaris argues, Miller cannot possibly show that the sabotage was “systematic.” But Miller does not have to prove systematic tampering at this point; she need only produce evidence that raises an inference that such tampering occurred to such an extent that it torpedoed her output.

In order to meet that burden, Miller provides evidence from her coworker Young and from her own sworn statements that one way in which Ballard and Kemp sabotaged Miller's work was by deliberately creating trays for her that were harder to complete and took more time. There is the incident in which Young attempted to take a tray that appeared to have more time-consuming samples on it, but Ballard told her it was for Miller, asking Young, “Don't you see what I'm trying to do here?” A trier of fact could draw an inference from this exchange that Ballard, from whom Young had overheard racist comments, was acting on that animus to make Miller's work more difficult. And Young testified about another incident in which Kemp tampered with a tray intended for Miller only to reverse her tampering once she was aware Miller had seen her. Miller testified about other occasions in which she would return from lunch or a break and catch Kemp or Ballard manipulating samples on her trays.

Polaris argues that “mathematical certainty” underlies its assertion that Ballard and Kemp could not have meddled with Miller's output to the extent that it caused her poor results. But that argument is based on the assumption that Ballard and Kemp manipulated Miller's work only three times—times at which they were actually seen by others—and there is no reason to make that assumption. A trier of fact might draw that conclusion, but it equally might find that the observed incidents were part of a larger pattern. At summary judgment, we must draw the latter inference, especially in light of the evidence of Ballard and Kemp's racial epithets (which at this stage we must assume they uttered, although we appreciate the fact that they have denied saying any such things). Polaris is welcome to argue to a jury that the sabotage of Miller's output would have had to be so extensive as to be practically impossible; a jury may weigh that point against Miller's evidence. But we are not the ones charged with resolving this matter; it is for the trier of fact. Polaris otherwise appears to think that the only way for cat's paw liability to take effect in this case would be for Miller to prove that Ballard “hacked” Polaris's own data system to reduce Miller's productivity numbers falsely. No one is arguing that. A jury reasonably could believe based on evidence in this record that Ballard expressed her animus more subtly, by manipulating the inputs in a way that dictated Miller's low output.

Polaris also contends that Miller is misrepresenting how work among its sample processors is assigned. Primarily relying on an affidavit of another of its employees, Terry Shotts, Polaris contends there is no such thing as “harder” work among the sample processors. Shotts was the “opener” in Miller's department during Miller's employment there; she swore in an affidavit that a sample processor's workday began with the processor making “a few trays of their own accounts so they could begin logging.” Then, Shotts said, she would open boxes of samples and arrange them by customers assigned to particular sample processors; after that, all remaining samples were placed in trays and put on a front table (Shotts does not say by whom). After this occurred, “[s]ometimes the Sample Processors just grabbed whatever tray they desired,” Shotts said, “and other times I would pass them out.” Each processor had four of the same accounts, but she would not know what accounts other processors “logged on a given day,” nor would she know the “estimated difficulty of samples.” Shotts left this last statement at that; she did not back it up with any explanation. Her affidavit thus did not exclude the possibility that Ballard tried to saddle Miller with trays requiring more paperwork. Shotts admitted that the difficulty of the paperwork can be ascertained by removing it from a carton of samples. Shotts also conceded that Ballard would sometimes assist her in composing trays of samples.

In all, Shotts's account (and Saperstein's, which Polaris also cites) is just evidence on Polaris's side of the balance. It does not foreclose Miller's and Young's contentions about Ballard's and Kemp's manipulation of Miller's work. It is not as if, for example, Shotts had total control over the samples. In any event, whether Shotts or Miller is correct is a question for a factfinder. We also note that Miller asserts that other actions by Ballard also contributed to her low output, such as Ballard's failing to answer Miller's questions, forcing Miller to waste time by leaving her workspace to pose her questions to New. A factfinder is entitled to hear Shotts's statements on how work is performed at Polaris and compare it to Miller's and Young's accounts of tampering. A trial is also the time for Miller to present statistics such as those in her reply brief, which (the trier of fact may conclude) support the possibility that she was indeed assigned harder work than some other processors.

It may be difficult to marshal evidence that coworkers and lower-level supervisors harbored discriminatory animus against a plaintiff and thwarted her ability to perform her work effectively, and it may be even harder to show that this is what lies behind a decisionmaker's adverse job action. Taking the summary judgment record in the light most favorable to Miller, however, as we must, we conclude that she has made it over the line. Although Polaris has some evidence that Miller's work could not have been manipulated to the extent she claims, it is not ironclad. We cannot simply eyeball the numbers, as Polaris urges, and decide that Miller could not have made up the difference in her productivity absent sabotage from Ballard and Kemp. Miller has demonstrated a dispute of material fact on her discrimination claim.

III

We briefly address Miller's retaliation claim, on which the district court also granted summary judgment in Polaris's favor. In pursuing such a claim under Title VII using the direct method of proof, “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.” Castro v. DeVry Univ., Inc., 786 F.3d 559, 564 (7th Cir.2015). We may either proceed through these factors one by one, or else “ask the fundamental question directly—could a reasonable trier of fact infer retaliation?” Id. A cat's paw theory of liability may apply to a retaliation claim. See Smith, 681 F.3d at 897–98.

Miller can establish that she engaged in a protected activity by complaining to New and then Culp on September 18, 2008, after Ballard (or someone in the department) had called her “the colored girl.” She also has evidence of racially charged language from Kemp and Ballard. Her accusation prompted an investigation of the relevant parties. In the course of that inquiry, both Kemp and Ballard were interviewed. Both Miller and Young testified that Ballard's and Kemp's subsequent manipulation of Miller's work occurred after the September 18 incident. Admittedly, neither Miller nor Young placed precise dates on the treatment they allege. But given the evidence of discriminatory statements by Kemp and Ballard, the timing and content of the poor treatment is sufficient to raise the inference that Ballard and Kemp were responding to Miller's report to New and Culp. It is not beyond any dispute of material fact whether the actions of these workers kept Miller's productivity scores artificially low. In that sense this case differs from Jajeh v. County of Cook, 678 F.3d 560 (7th Cir.2012), in which there was far more mathematical and factual certainty about whether an employee would have received a sufficient performance score absent discriminatory meddling. Finally, it is undisputed that Miller suffered an adverse action—the loss of her job—because of her low numbers.

Given this evidence, Miller has adequately advanced a dispute of material fact on whether Kemp and Ballard retaliated against her for the September 18 incident. This claim, too, should be resolved by a finder of fact.

IV

We Reverse the judgment of the district court and Remand this case for proceedings consistent with this opinion.