Comparator Proof in Relative-Ranking Regimes: Seventh Circuit Holds Employer-Defined Assessment Groups Control Title VII Comparators and Bars Undisclosed Cross-Case Evidence

Introduction

In Kara Mitchell v. Exxon Mobil Corporation, No. 24-2823 (7th Cir. July 14, 2025), a Seventh Circuit panel (Judges Easterbrook, Brennan, and Pryor; opinion by Judge Brennan) affirmed summary judgment for ExxonMobil on a Title VII sex-discrimination claim. The case arose from ExxonMobil’s annual, relative-ranking performance assessment process, which places employees into lettered categories (A–D) within defined “assessment groups” and, in the relevant cycle, required a standard distribution of ratings within each group. Mitchell, a laboratory technician designated as a Management Assigned Category (MAC) employee due to her prior relevant experience, received a D rating following the group process and was separated pursuant to company policy.

Mitchell alleged that her discharge was because of sex, arguing that preliminary assessments by local supervisors “tainted” the outcome, that a male supervisor harbored bias against women, and that several women at the Cicero facility had been terminated around the same time. She also sought to leverage evidence from another case involving the same employer and facility. The district court granted summary judgment. On appeal, the Seventh Circuit addressed two evidentiary rulings and then evaluated the merits under both the McDonnell Douglas framework and the holistic approach of Ortiz.

The opinion’s most salient legal clarifications are twofold:

When an employer uses a formal relative-ranking system with defined assessment cohorts, the “similarly situated” comparator analysis for Title VII must track those cohorts; co-workers outside the relevant assessment group are not appropriate comparators even if they share title, location, or supervisor.
Evidence from other litigation—even if publicly accessible—cannot be used at summary judgment if it was not disclosed under Rule 26; Rule 37(c)(1) bars its use absent substantial justification or harmlessness.

Summary of the Opinion

The court affirmed summary judgment for ExxonMobil. It first upheld two evidentiary rulings: (1) a supervisor’s declaration recounting the assessment group meeting was properly considered; calling it “self-serving” did not render it inadmissible where record facts corroborated his attendance and role; and (2) the district court properly excluded depositions and documents from a different case (Ontiveros v. Exxon Mobil Corp.) because Mitchell failed to disclose them under Rule 26, and their late use would have prejudiced ExxonMobil under Rule 37(c)(1).

Turning to the merits, the court held that Mitchell failed the fourth prong of McDonnell Douglas because she did not show she was treated less favorably than “similarly situated” employees outside her protected class. Her preferred comparators—two male lab technicians in Cicero—were not in her assessment group, and she offered no evidence of how either fared in their own groups. The court also concluded that, even if a prima facie case were assumed, ExxonMobil articulated a legitimate, nondiscriminatory reason (her poor relative performance within her assessment cohort), and Mitchell failed to demonstrate pretext. Under Ortiz’s totality-of-the-evidence approach, her pattern-or-practice theory failed for lack of admissible, probative evidence tying any alleged pattern to her discharge or showing regular, purposeful discrimination.

Analysis

Precedents Cited and Their Role

Igasaki v. Illinois Department of Financial & Professional Regulation, 988 F.3d 948 (7th Cir. 2021): Reiterates summary-judgment standards, the “singular question” of causation under Title VII, and use of both McDonnell Douglas and Ortiz approaches.
McDonnell Douglas Corp. v. Green, 411 U.S. 792 (1973); Texas Department of Community Affairs v. Burdine, 450 U.S. 248 (1981); Swierkiewicz v. Sorema N.A., 534 U.S. 506 (2002); Ames v. Ohio Department of Youth Services, 145 S. Ct. 1540 (2025): Frame the burden-shifting method, emphasize that the prima facie threshold is “not onerous,” and caution against rigid formulae.
Ortiz v. Werner Enterprises, Inc., 834 F.3d 760 (7th Cir. 2016); Brooks v. Avancez, 39 F.4th 424 (7th Cir. 2022); Vega v. Chicago Park District, 954 F.3d 996 (7th Cir. 2020); Reives v. Illinois State Police, 29 F.4th 887 (7th Cir. 2022): Direct courts to consider the totality of the evidence and clarify that McDonnell Douglas remains an organizational tool, not a substantive straitjacket.
Comparator cases: Gamble v. County of Cook, 106 F.4th 622 (7th Cir. 2024); Downing v. Abbott Laboratories, 48 F.4th 793 (7th Cir. 2022); Smith v. City of Janesville, 40 F.4th 816 (7th Cir. 2022): Define “similarly situated” as “directly comparable in all material respects,” with the purpose of eliminating other explanatory variables and isolating discriminatory animus.
Summary judgment and burden-shifting clarity: Stockwell v. City of Harvey, 597 F.3d 895 (7th Cir. 2010); Wince v. CBRE, Inc., 66 F.4th 1033 (7th Cir. 2023); Anderson v. Street, 104 F.4th 646 (7th Cir. 2024); Napier v. Orchard School Foundation, 137 F.4th 884 (7th Cir. 2025): Discuss light burden to articulate a legitimate reason, absence-of-evidence showings, and pretext standards.
Pretext/business judgment: Galvan v. Indiana, 117 F.4th 935 (7th Cir. 2024); Grant v. Trustees of Indiana University, 870 F.3d 562 (7th Cir. 2017): Reinforce that courts are not “super personnel departments” second-guessing business judgments.
Pattern-or-practice: Puffer v. Allstate Insurance Co., 675 F.3d 709 (7th Cir. 2012); Council 31, AFSCME v. Ward, 978 F.2d 373 (7th Cir. 1992); International Brotherhood of Teamsters v. United States, 431 U.S. 324 (1977): Require proof that discrimination is the employer’s “standard operating procedure,” i.e., regular and purposeful.
Evidentiary rulings: Butts v. Aurora Health Care, Inc., 387 F.3d 921 (7th Cir. 2004): Self-serving affidavits without factual support carry no weight. The panel distinguishes this by finding corroboration for the supervisor’s declaration.
Federal Rules of Civil Procedure 26(a) and 37(c)(1): Mandate disclosure of witnesses and information a party may use to support claims and sanction failures by exclusion at summary judgment absent substantial justification or harmlessness.

Legal Reasoning

1) Evidentiary Gatekeeping

The court affirmed two key evidentiary rulings:

Supervisor declaration: Although characterized by Mitchell as “self-serving,” the declaration was supported by record facts (including email correspondence directed to “Assessment Meeting Participants” and the declarant’s supervisory responsibilities to deliver feedback after Hayes retired). Therefore, it was properly considered at summary judgment; it was not an unsupported, conclusory affidavit within the Butts prohibition.
Cross-case evidence exclusion: Mitchell attempted to submit depositions and records from Ontiveros v. Exxon Mobil Corp. to prove a pattern or practice at the Cicero plant. Because she never disclosed those witnesses or documents under Rule 26(a), Rule 37(c)(1) barred their use at summary judgment. The panel emphasized that public availability does not override disclosure obligations, and late use would have prejudiced ExxonMobil by denying it the chance to test the evidence in this case.

2) Comparator Definition Anchored to Employer’s Assessment Cohorts

The court’s central doctrinal move is to align the “similarly situated” inquiry with the employer’s own formal comparison set. ExxonMobil’s system grouped employees by role and experience into assessment cohorts and applied a standard distribution of A–D categories within each cohort, with MAC status identifying comparability among new hires with prior relevant experience.

Mitchell’s preferred comparators—two male lab technicians at her facility—were outside her MAC assessment group. Her actual cohort consisted of three MAC employees in similar operational roles, including one woman and two men, all rated B while Mitchell was rated D. The court held that, in this relative-ranking context, co-workers outside the assessment group are not “similarly situated” for Title VII comparator purposes, even if they share job title, location, or local supervision. The comparator set is defined by the employer’s neutral grouping used to make the decision being challenged.

This focus serves the comparator doctrine’s purpose: eliminating alternative explanatory variables by ensuring apples-to-apples comparisons. Because the adverse action flowed from within-cohort ranking, the relevant comparators are within that cohort.

3) Preliminary vs. Final Rankings; Process Irregularities

Mitchell argued the preliminary ranking stage “tainted” the outcome and that the final meeting rubber-stamped earlier decisions. The court rejected this, citing a dynamic record in which preliminary categories shifted across multiple supervisors before the final meeting. The manual’s guidance on preliminaries was described as “vague,” and even if some deviation occurred, Mitchell failed to show prejudice—no evidence established that the final meeting was pretextual or that preliminary categories invariably controlled outcomes for any comparator.

Critically, Mitchell lacked evidence of the final categories assigned to her preferred comparators and of the initial categories for her actual cohort, undermining claims that only her preliminary ranking changed or that comparable employees’ ranks were fixed from the outset.

4) Pretext and the Legitimacy of Forced Distributions

ExxonMobil articulated a legitimate, nondiscriminatory reason: within a competitive MAC cohort, other employees were perceived as going “above and beyond,” and Mitchell did not stand out. The court credited testimony about the final meeting and noted that ranking distributions are a feature of the process. The mere existence of a standard distribution requiring some D ratings does not demonstrate pretext; absent evidence of discriminatory application, such a system is a business judgment insulated from judicial second-guessing.

Mitchell provided no comparative performance evidence from her cohort (e.g., the others’ self-assessments, peer feedback, or contrary witness testimony) and admitted unfamiliarity with the comparators’ performance. Her attempt to ascribe bias to the plant manager also failed: she testified he had not shown bias or made derogatory comments, and the record suggested he did not participate in the preliminary emails and only conveyed final results after the meeting.

5) Ortiz Holistic Review; Pattern or Practice

Under Ortiz, the court considered all evidence together. Mitchell’s pattern-or-practice theory faltered because the core proof came from excluded material and, even if considered, lacked ties to her assessment cohort, supervisors, or the decision process at issue. The women she identified were in different categories (managerial/professional vs. operational), some opted to retire or were contractors, and there was no showing that the alleged decision-maker had a meaningful role in her own assessment outcome. The record thus did not show that sex discrimination was ExxonMobil’s “standard operating procedure.”

Impact

A. Title VII Litigation in Relative-Ranking Systems

Comparator alignment: Plaintiffs challenging outcomes of a formal relative-ranking process should be prepared to identify comparators from the same assessment cohort that actually drove the decision. Traditional indicia—same supervisor, title, or location—may be insufficient when the employer’s decision framework defines comparability differently.
Evidence gaps are fatal: Plaintiffs need concrete, admissible proof of both initial and final rankings for cohort members to show disparate treatment, irregular application of distributions, or pretext.
Forced distributions as legitimate rationale: The opinion tacitly validates standard distributions as a legitimate nondiscriminatory reason, putting the burden on plaintiffs to show discriminatory application (e.g., uneven criteria, biased inputs, or inconsistent treatment across protected classes).

B. Discovery Practice and Cross-Case Evidence

No “public record” safe harbor: Parties cannot rely on the public availability of depositions or filings from other cases to bypass Rule 26; failure to disclose means exclusion at summary judgment under Rule 37(c)(1) absent a showing of harmlessness or substantial justification.
Strategic lesson: If a plaintiff anticipates using cross-case testimony to prove pattern or animus, those witnesses and materials must be identified in initial disclosures or supplemental disclosures, with time for the opponent to test them.

C. Employer Compliance Considerations

Document cohort definitions and rationale: Clearly define who is in an assessment group and why. The court’s deference to employer-defined cohorts makes accurate documentation pivotal.
Record the “why”: Meeting notes explaining relative rankings, especially specific “standout” behaviors, help establish legitimate reasons and rebut pretext.
Guard against bias in inputs: Because the system aggregates supervisor judgments and peer feedback, ensure training and checks to prevent biased inputs from contaminating outcomes.
Monitor adverse-impact risk: Although not litigated here, forced distributions can raise disparate-impact concerns. Regular statistical monitoring and validation can mitigate that risk.

Complex Concepts Simplified

Title VII discrimination: Federal law prohibits discharging or otherwise disadvantaging an employee “because of” protected characteristics, including sex.
McDonnell Douglas framework: A three-step proof structure. Step 1: Plaintiff makes a prima facie case (protected class, qualified, adverse action, and better treatment of similarly situated non-protected employees). Step 2: Employer articulates a legitimate reason. Step 3: Plaintiff shows that reason is pretext—a dishonest cover for discrimination.
Ortiz holistic approach: Courts must consider all evidence together to decide whether a reasonable jury could find discrimination—no rigid categories of “direct” or “indirect” evidence.
“Similarly situated”: Comparators must be alike in “all material respects” relevant to the decision. Here, because the decision turned on within-cohort relative ranking, the material respect was membership in the same assessment cohort.
Pretext: Not simply a bad or unfair decision. It is a false explanation that hides the real, discriminatory motive.
Pattern or practice: Proof that discrimination is the employer’s routine way of operating, not just isolated instances.
Rule 26 and Rule 37: Parties must disclose witnesses and evidence they plan to rely on; undisclosed evidence is generally excluded at summary judgment unless the failure is harmless or justified.
MAC and Management of Lower Relative Performance: MAC identifies certain new hires with prior relevant experience; employees rated D enter a remediation or separation track. In the cycle at issue, MAC employees rated D could choose pay in lieu or termination without benefits.
Forced or standard distribution: A rating system that requires set proportions in each category within a cohort. Such systems can be legitimate, but they must be applied without discrimination.

Conclusion

Mitchell v. Exxon Mobil Corporation offers two durable guideposts for Title VII disputes involving relative-ranking systems. First, comparator analysis must be tethered to the employer’s formal cohort that actually governed the adverse action; employees outside that cohort—even those with the same title, supervisor, or location—are not “similarly situated” where the decision mechanism is cohort-based. Second, evidentiary discipline matters: litigants cannot import “public record” material from parallel litigation at summary judgment without timely disclosure under Rule 26; Rule 37(c)(1) exclusion will apply.

On the merits, the court reaffirmed familiar principles: McDonnell Douglas remains a flexible organizational tool; Ortiz requires a holistic look; and pretext requires proof of dishonesty, not mere disagreement with business judgments. In a record devoid of admissible comparator outcomes, marked by a competitive cohort including another woman who fared better than the plaintiff, and lacking admissible pattern evidence, the panel concluded no reasonable jury could find sex discrimination.

The opinion’s practical signal is clear. Plaintiffs must gather and disclose cohort-specific data and admissible evidence showing disparate treatment or biased application of rating distributions; employers should define cohorts transparently, document relative-performance reasoning, and ensure that multi-step assessments are consistently applied. Within that framework, courts will defer to legitimate, well-documented relative-ranking decisions absent concrete proof of discriminatory causation.