Plaintiffs Roñica Tabor and Dacia Gray (collectively “Plaintiffs”) worked as inside sales representatives at Hilti, Inc., and Hilti of North America, Inc. (collectively “Hilti”). After being denied promotions to Account Manager (outside sales) positions, they each filed individual claims for gender discrimination under Title VII and moved to certify a class of all female inside sales representatives at Hilti who were denied similar promotions.
The district court refused to certify the class and granted summary judgment for Hilti on all claims. Exercising jurisdiction under 28 U.S.C. § 1291, we affirm in part and reverse in part. We affirm the district court’s grant of summary judgment on Ms. Tabor’s individual claim for retaliation and Ms. Gray’s individual claim for failure to promote. We also affirm the district court’s refusal to certify the class. We reverse with respect to Ms. Tabor’s individual claims for failure to promote and disparate impact, and we remand Ms. Gray’s individual disparate impact claim because the district court did not address the claim in its analysis.
I. BACKGROUND
A. Factual History1
Hilti is a tool manufacturer. The company employs inside sales representatives, *1212who are responsible for providing customer assistance and sales support by phone. A common career track for inside sales employees in the Customer Service Department is promotion to Account Manager. Account Managers are responsible for outside sales or field sales, including site visits to customers within an assigned territory. This promotion sometimes involves transfer to a different city and requires some skills not required for inside sales, such as the ability to lift 60 pounds, engage with customers face-to-face, and offer hands-on demonstrations of the tools.
Hilti established a performance management and reporting process it called the “Global Develop and Coach Process” (“GDCP”).2 GDCP included multiple components that tracked different aspects of an employee’s readiness to promote. An important component was a priority rating, or “P” rating, indicating a direct supervisor’s or manager’s subjective assessment of an employee’s promotion-readiness based upon his or her skills in areas such as “Functional Expertise,” “Understanding the Business,” “Getting Things Done,” “Working with Others,” and “Living Our Values.” Aplt. Appx. at 877-78. A PI rating indicated the employee was ready for promotion within zero to 12 months, while a P5 rating indicated the employee was currently ineligible for promotion. Another important component was an “M” rating, which indicated the employee’s reported mobility, i.e., willingness to relocate. Still another important GDCP component was employee’s career goal, e.g., to become an Account Manager or a Team Leader in the Customer Service Department.
Hilti considered GDCP its official method for identifying employees who would be promoted internally. However, Hilti did not maintain careful records. Hilti’s Applicant Flow Log data (“AFL data”) indicated that 282 individuals were promoted between 2005 and 2008, but fewer than 24% had been assigned a P rating at the time of promotion; fewer than 37% of promoted employees were assigned M ratings; fewer than 8% .of individuals who were promoted to outside sales positions had actually identified outside sales as a future career goal; and more than 64% of employees were missing both P rating and M rating at the time of promotion.
Hilti managers also did not always follow the GDCP ratings in making promotion decisions. For example, of the promoted employees who had been assigned a P rating at the time of promotion, only 28% had a PI. Furthermore, 33 promoted employees were assigned a P rating of P5 at the time of promotion. A P5 rating indicated the employee was currently ineligible to promote because he or she did not meet the minimal qualifications, e.g., tenure in current position. Plaintiffs allege that a number of male inside sales representatives were placed in Account Manager positions through “tap on the shoulder” promotions, that is, extending promotion offers to male employees without posting an open position or allowing other interested employees to apply. Aplt. Br. at 8. Plaintiffs also allege that males who were ineligible for promotion under the GDCP system were allowed or even invited to apply for Account Manager positions, even as Plaintiffs and other female employees were told they could not apply for promotion- until they earned a PI rating.
One prerequisite for earning a PI rating for the Account Manager position was completion of field training. This training *1213involved filling in for an Account Manager who was on vacation by assuming his or her responsibilities for one to two weeks. Ms. Gray participated in field training in Dallas, Texas, and she requested additional field training on more than one occasion. Ms. Tabor also requested field training. Hilti did not allow either plaintiff to participate in field training during the first half of 2008. The company explains that it was short-handed during this time and that no customer service representative was permitted to leave for field training. Plaintiffs claim that at least two males were allowed to attend field training during this time — Berkeley Smith and an unnamed male. Hilti says Mr. Smith was only allowed a short trip to Arkansas to decide whether to accept a promotion offer to relocate there.
1. Roñica Tabor’s Experiences at Hilti
Roñica Tabor began work at Hilti in January 2006, selling and demonstrating tools to customers face-to-face at a Hilti center in Dallas, Texas. She transferred to the Customer Service Department in Tulsa, Oklahoma, in October 2006. She expressed interest in becoming an Account Manager, and her immediate supervisor assigned her a PI rating.
Ms. Tabor applied for an Account Manager position located in Oklahoma City. This position focused primarily on the company’s Interior Finish product line. Ms. Tabor first interviewed with Regional Manager David Perkins and was selected for a second interview. This second interview was on November 14, 2007, with Division Manager Glenn Teel and Mr. Perkins. During the interview, Mr. Teel and Mr. Perkins mentioned a second position available in Arkansas, and Ms. Tabor expressed interest in that position as well.
During the interview, Mr. Teel made a number of statements related to Ms. Tabor’s gender. He told her that tools “are like guns for men” and using them is “almost like second nature,” Aplt. Appx. at 2816, and that it would take more work for her, as a woman, to learn the tools well enough to demonstrate them for customers or she would be “chewed up and spit out,” Aplt. Br. at 10. Mr. Teel also suggested that as a woman, Ms. Tabor might have some “advantages” in getting men to talk to her even if they were reluctant to talk to a salesman. Aplt. Appx. at 2816. Mr. Teel expressed concern about whether Ms. Tabor should travel as much as the job required because she was a wife and mother. He stated that he would personally not want his wife to hold a job that required travel, and he advised Ms. Tabor to ask her husband about whether she should pursue this type of work.
Ms. Tabor was not offered either of the two Account Manager positions. Berkeley Smith, a male, was offered the Arkansas position. A male employee, Clifford Kid-well, was eventually hired as an Account Manager in Oklahoma City. Around this same time frame, an external female applicant named Paulette Musso was hired as an Account Manager in Tulsa, Oklahoma.
The district court found a dispute of material fact as to whether the Oklahoma City position for which Ms. Tabor interviewed was offered to Mr. Kidwell or to Ms. Musso. Mr. Kidwell was offered an Account Manager position in Oklahoma City, and Ms. Musso was hired as an Account Manager in Tulsa. In its brief, Hilti simply asserted that Ms. Musso was hired for the Oklahoma City position and offered no explanation for the conflicting facts in the record or the district court’s finding that this fact was disputed. At oral argument, Hilti explained that after they interviewed Ms. Tabor, Mr. Teel and Mr. Perkins decided to move the Oklahoma City position to Tulsa and offered that position *1214to Ms. Musso. Around the same time, in what it describes as an unrelated decision, Hilti created a brand new Oklahoma City Interior Finish position, which it offered to Mr. Kidwell.
The posting for the Oklahoma City position stated that the job would focus on the Interior Finish product line, that a bachelor’s degree was strongly preferred, and that the position required ability to work with Spanish-speaking customers. Ms. Tabor had a bachelor’s degree and was fluent in Spanish. The parties agree she exhibited the strongest knowledge of the Interior Finish products. She had experience selling Interior Finish products over the phone and face-to-face from her experience at the Hilti center in Dallas, Texas. Mr. Smith had a bachelor’s degree, but Mr. Kidwell did not. Ms. Musso’s educational qualifications are unknown.3
After the interview, Mr. Teel and Mr. Perkins documented their evaluation of Ms. Tabor. They gave her high ratings in several areas, including personal skills and qualities, working with others, and developing herself and others. They assigned her low ratings in other areas, including time and territory management, knowledge of the business, and construction site etiquette. Mr. Teel and Mr. Perkins assigned Ms. Tabor a rating of P2, indicating she would not be ready for promotion for 12-24 months.
After the interview, Ms. Tabor complained to her supervisor about Mr. Teel’s comments and made a complaint to Human Resources (“HR”). After several days, an HR representative followed up and advised her “to just brush it under the rug ... start fresh ... and to just not speak of it again.” Aplt. Appx. at 2817. The HR representative assured her that her complaint would not affect her future at the company. Ms. Tabor alleges, however, that she was subsequently told her P2 rating would remain in place with respect to any future Account Manager applications. She asked to do additional field training to improve the P rating but was not allowed to do so. Ms. Tabor concedes that as a formal matter her rating in the GDCP tracking system remained PI, but she contends the P2 interview rating affected her eligibility to interview for other Account Manager positions.
Ms. Tabor claims these events caused depression, anxiety, loss of sleep, loss of appetite, and upset stomach; she sought medical attention and was prescribed anti-depression medication. She resigned from Hilti on April 5, 2008.
2. Dacia Gray’s Experiences at Hilti
Dacia Gray became a Hilti Customer Service Representative in January 2005. Like Ms. Tabor, Ms. Gray hoped to become an Account Manager. When Ms. Gray shared this goal with her supervisor, Larry Brown, he told her: “Women do not make it out in the field.... They don’t succeed. They don’t do well. Men don’t respond to women.” Aplt. Appx. at 2525.
Unlike Ms. Tabor, Ms. Gray never applied for an Account Manager position and was never assigned a PI rating. She attempted to complete additional field training to achieve a PI rating but was not permitted to do so.
In a December 2007 performance review, Ms. Gray’s performance was rated as *1215meeting expectations for her current job. Notwithstanding this satisfactory evaluation, several supervisors and managers at Hilti testified to performance deficiencies on Ms. Gray’s part. They described disciplinary problems, such as frequent tardiness, inattentiveness, sleeping on the job, lack of commitment, and poor attitude. One supervisor sent Ms. Gray a letter warning that she had exhausted her sick leave in violation of Hilti policy and was in danger of termination. In addition, Mr. Brown claims to have warned Ms. Gray twice in 2008 that she was in danger of termination due to tardiness and inattentiveness.
On June 30, 2008, Ms. Gray resigned in a letter to Mr. Brown, accusing him of discriminating against her and other female employees on the basis of sex and disability.
B. Procedural History
Ms. Tabor and Ms. Gray brought individual claims against Hilti alleging intentional and disparate impact sex discrimination in violation of Title VII, as well as class claims for disparate impact and pattern and practice discrimination. They filed a motion to certify a class composed of “all women employed by Hilti in the United States denied promotion to Account Manager ...” Aplt. Appx. at 1672.
The district court denied class certification and granted summary judgment for Hilti on all claims. Ms. Tabor appeals individual claims for failure to promote, retaliation, and disparate impact. Ms. Gray appeals her individual claims for failure to promote and disparate impact. Both Plaintiffs appeal denial of class certification.
II. DISCUSSION
“We review de novo the district court’s decision to grant summary judgment.” Turner v. Public Serv. Co., 563 F.3d 1136, 1142 (10th Cir.2009). We view facts in the light most favorable to Ms. Tabor, and Ms. Gray and “draw all reasonable inferences” in their favor. Id. Summary judgment is appropriate only if Hilti shows “ ‘there is no genuine dispute as to any material fact and [it] is entitled to judgment as a matter of law.’ ” Hernandez v. Valley View Hosp. Ass’n, 684 F.3d 950, 957 (10th Cir.2012) (quoting Fed.R.Civ.P. 56(a)). “A fact is ‘material’ if, under the governing law, it could have an effect on the outcome of the lawsuit. A dispute over a material fact is ‘genuine’ if a rational jury could find in favor of the nonmov-ing party on the evidence presented.” E.E.O.C. v. Horizon/CMS Healthcare Corp., 220 F.3d 1184, 1190 (10th Cir.2000) (citation omitted).
We first address each of Ms. Tabor’s individual claims. Next, we consider Ms. Gray’s individual claims. Finally, we consider the Plaintiffs’ class action claims.
A. Ms. Tabor’s Individual Claims
Ms. Tabor appeals three individual claims for gender discrimination under Title VII. The first two claims, failure to promote and retaliation, charge Hilti with intentional discrimination. The third claim charges Hilti with disparate impact discrimination.
1. Failure to Promote
The district court agreed that Mr. Teel’s interview comments were “inappropriate” but nevertheless determined that Hilti’s low ratings of Ms. Tabor’s qualifications were unrelated to her gender. Aplt. Appx. at 2826. The court then concluded that Hilti’s reasons for not promoting Ms. Tabor were non-discriminatory and non-pre-textual and therefore dismissed her claim. We reverse.
*1216a. Legal Background
“Title VII of the Civil Rights Act of 1964 prohibits, among other things, unlawful employment discrimination on the basis of an individual’s sex.” Horizon/CMS, 220 F.3d at 1190; see 42 U.S.C. § 2000e-2. The parties disagree about what legal standard applies to this claim.
When a plaintiff offers direct evidence of discrimination in a Title VII claim, her claim may move forward without being subjected to the burden-shifting framework set forth in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973). See Ramsey v. City & Cnty. of Denver, 907 F.2d 1004, 1007-08 (10th Cir.1990) (“ ‘[T]he McDonnell Douglas test is inapplicable where the plaintiff presents direct evidence of discrimination.’ ” (quoting Trans World Airlines, Inc. v. Thurston, 469 U.S. 111, 121, 105 S.Ct. 613, 83 L.Ed.2d 523 (1985))). The classic example of direct evidence of discrimination comes from Trans World Airlines, where the Supreme Court held that an explicit, mandatory age requirement was direct evidence of age discrimination. 469 U.S. at 121, 105 S.Ct. 613.
Comments in the workplace that reflect personal bias do not qualify as direct evidence of discrimination unless the plaintiff shows the speaker had decision-making authority and acted on his or her discriminatory beliefs. Ramsey, 907 F.2d at 1008. We also have explained that discriminatory statements do not qualify as direct evidence if the context or timing of the statements is not closely linked to the adverse decision. Riggs v. AirTran Airways, Inc., 497 F.3d 1108, 1118 (10th Cir.2007). Furthermore, if the content and context of a statement allow it to be plausibly interpreted in two different ways — one discriminatory and the other benign — the statement does not qualify as direct evidence. Id.
When evidence of discrimination is circumstantial, rather than direct, a plaintiffs claim is subject to the McDonnell Douglas burden-shifting framework. See Ramsey, 907 F.2d at 1008. Under McDonnell Douglas, a plaintiff carries the initial burden of establishing a prima facie case of discrimination. Id. at 1007. The burden at this stage is “ ‘not onerous.’ ” Orr v. City of Albuquerque, 417 F.3d 1144, 1149 (10th Cir.2005) (quoting Texas Dep’t of Cmty. Affairs v. Burdine, 450 U.S. 248, 253, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981)).
To state a prima facie case of discrimination under McDonnell Douglas, a plaintiff must demonstrate by a preponderance of the evidence that (1) she belongs to a protected class; (2) she applied for an available position for which she was qualified; (3) she “was rejected under circumstances which give rise to an inference of unlawful discrimination.” Burdine, 450 U.S. at 253, 101 S.Ct. 1089; see Orr, 417 F.3d at 1149.4
If a plaintiff states a prima facie case, the burden shifts to the employer to proffer “a legitimate non-discriminatory *1217purpose for the adverse employment action.” Orr, 417 F.3d at 1149. If the employer makes this offering, the plaintiff will avoid summary judgment only if she shows her sex “was a determinative factor in the ... employment decision, or show[s] the [employer’s] explanation for its action was merely pretext.” Horizon/CMS Healthcare, 220 F.3d at 1191 (quotations omitted).
Ms. Tabor argues that Mr. Teel’s comments are direct evidence of discrimination and that McDonnell Douglas therefore does not apply. Hilti disputes that Mr. Teel’s comments are discriminatory at all and insists that McDonnell Douglas does apply. We conclude that Ms. Tabor’s claim survives summary judgment under either standard.
b. Application of the Direct Evidence Standard
Hilti does not dispute that Mr. Teel made the alleged comments during the interview, but it insists they are “innocuous and non-discriminatory.” Aplee. Br. at 12. Hilti points to our decisions in Ramsey and Heim v. State of Utah, 8 F.3d 1541 (10th Cir.1993). In these cases, we emphasized that “stray remarks in the workplace” based on sex stereotypes do not constitute direct evidence of discrimination. Heim, 8 F.3d at 1547 (quotations omitted). “The plaintiff must show that the employer actually relied on ... gender in making its decision.” Id. (quotations omitted).
But none of Hilti’s cited cases involve statements by a decisionmaker during an interview expressing discriminatory beliefs about whether members of the plaintiffs protected class are capable of doing the job at issue. We have previously emphasized the importance of context and temporal proximity in determining whether comments reflecting personal bias qualify as direct evidence of discrimination. Riggs, 497 F.3d at 1118. Here, Mr. Teel explicitly stated a view that women have inferior knowledge of tools and inferi- or ability to sell tools. These statements spoke directly to central requirements of the job for which Ms. Tabor was interviewing, and he made them during a discussion about her fitness for the position. The content of his statements, the interview context, and the temporal proximity to the adverse employment decision directly link the discriminatory statements to his decision not to promote Ms. Tabor.5
Given these circumstances, Mr. Teel’s remarks may be considered direct evidence of discrimination.
c. Application of the McDonnell Douglas Standard
Ms. Tabor’s claim also survives summary judgment under McDonnell Douglas. The parties agree that the first two McDonnell Douglas steps are satisfied: 1) Ms. Tabor established a prima facie case, and 2) Hilti responded by asserting a legitimate nondiseriminatory reason for not promoting her — namely, Mr. Teel and Mr. Perkin’s assessment that her knowledge of tools and time management skills were insufficient.
*1218We therefore arrive at the third McDonnell Douglas step. To satisfy this step and overcome summary judgment, Ms. Tabor must show that Hilti’s asserted reason “was not the true reason for the employment decision.” Burdine, 450 U.S. at 256, 101 S.Ct. 1089. A plaintiff can meet this burden to show pretext in either of two ways: (1) by showing that the proffered reason is factually false or (2) by showing that discrimination was a primary factor in the employer’s decision, which is often accomplished by revealing “weaknesses, implausibilities, inconsistencies, in-coherencies, or contradictions in the employer’s proffered reason[],” such that a reasonable fact finder could deem the employer’s reason “unworthy of credence.” Garrett v. Hewlett-Packard Co., 305 F.3d 1210, 1217 (10th Cir.2002) (quotations omitted).
The district court found that Mr. Teel’s comments manifested a discriminatory belief that “women ... have inferior natural skill with tools.” Aplt. Appx. at 2824-25. Nevertheless, the court found that the interviewers’ negative ratings of Ms. Tabor’s qualifications raised “specific and practical” concerns that were “unrelated to generalized concerns over gender.” Id. at 2825. We agree with the district court that Mr. Teel’s comments revealed discriminatory views, but we disagree with its conclusion that Ms. Tabor failed to establish an inference that the negative ratings were related to the discriminatory views.
Ms. Tabor has raised a genuine dispute of material fact as to whether Hil-ti’s proffered reasons for rejecting her are “unworthy of credence.” Garrett, 305 F.3d at 1217 (quotations omitted). A reasonable jury could infer that the negative ratings the interviewers assigned Ms. Tabor did not represent objective individualized assessments of her qualifications but rather a reflection of the discriminatory views Mr. Teel expressed during the interview. See id. (“A plaintiff can withstand summary judgment by presenting evidence sufficient to raise a genuine dispute of material fact regarding whether the defendant’s articulated reason for the adverse employment action is pretextual.”); see also Simms v. Dep’t of Mental Health and Substance Abuse Servs., 165 F.3d 1321, 1328 (10th Cir.1999) (evidence of pretext may include use of subjective criteria). For example, the low rating given to Ms. Tabor on knowledge of the construction business relates closely to Mr. Teel’s expressed belief that women generally have inferior knowledge of tools. Similarly, the low rating in time management relates to Mr. Teel’s expressed concern that she would not be able to handle the workload because she was a wife and mother. As Ms. Tabor argues, a reasonable jury could infer that Mr. Teel’s discriminatory views are “mirrored in his ratings” of Ms. Tabor. Aplt. Br. at 53.
Hilti claims that its alleged choice to hire a single mother (Ms. Musso) for the Oklahoma City position proves that it did not discriminate against Ms. Tabor. One problem with this argument is that Ms. Tabor has raised a genuine dispute regarding whether Ms. Musso was actually the person hired for the Oklahoma City Interi- or Finish position: It is undisputed that Ms. Tabor applied for an Interior Finish Account Manager position in Oklahoma City, and that she did not apply for a position in Tulsa. It is also undisputed that a male candidate, Mr. Kidwell, was hired for an Interior Finish Account Manager position in Oklahoma City and that Ms. Musso was hired for an Account Manager position in Tulsa.
Hilti explains that it relocated the Oklahoma City position for which Ms. Tabor applied to Tulsa and offered it to Ms. Musso. Then, in an unrelated decision, it created a new, separate Oklahoma City *1219position that it offered to Mr. Kidwell.6 Under this logic, Hilti insists, Ms. Musso (and not Mr. Kidwell) was selected over Ms. Tabor and thus any inference of gender discrimination is impossible. Perhaps this explanation is true, but that is for a factfinder to determine. A reasonable jury could reject this explanation and infer that Hilti’s proffered reasons for rejecting Ms. Tabor are unworthy of credence and therefore pretextual.
In short, we find Ms. Tabor has raised a genuine and material dispute as to whether Hilti failed to promote her for discriminatory reasons in violation of Title VII. We therefore reverse the district court’s dismissal of this claim.
2. Retaliation
The district court rejected Ms. Tabor’s retaliation claim. It found she had met the first part of her prima facie burden to show that she engaged in protected opposition to discrimination but failed to show that Hilti took adverse action against her because of this opposition. We agree.
“Title VII forbids retaliation against an employee because she has ‘opposed’ any practice made unlawful by Title VII, or because she has ‘participated ... in an investigation, proceeding, or hearing’ ” regarding a claim of discrimination. Stover v. Martinez, 382 F.3d 1064, 1070 (10th Cir.2004) (quoting 42 U.S.C. § 2000e-3(a)). Because Ms. Tabor does not allege direct evidence of retaliation, the McDonnell Douglas framework applies. See id.
To state a prima facie case of retaliation, Ms. Tabor must show: “(1) that [s]he engaged in protected opposition to discrimination, (2) that a reasonable employee would have found the challenged action materially adverse, and (3) that a causal connection existed between the protected activity and the materially adverse action.” Argo v. Blue Cross and Blue Shield of Kan., Inc., 452 F.3d 1193, 1202 (10th Cir.2006) (footnote omitted) (citing Burlington N. & Santa Fe Ry. Co. v. White, 548 U.S. 53, 66-68, 126 S.Ct. 2405, 165 L.Ed.2d 345 (2006)).
The district court and the parties agree that Ms. Tabor engaged in protected opposition to discrimination when she complained to her supervisor and to HR about Mr. Teel’s discriminatory comments during the interview. The parties disagree whether Hilti took adverse employment action against Ms. Tabor. The district court found that any adverse action could not have been causally connected to Ms. Tabor’s complaint because the company did not take any action after her complaint to HR.
Ms. Tabor argues that Hilti took two adverse actions against her: first, it kept her in a P2 status, and second, it refused to allow her to complete additional field training, a necessary step to changing her status to a PI. These actions, she argues, blocked her from applying for promotions for at least 12-24 months. She correctly notes that the Tenth Circuit construes the term “adverse action” liberally. *1220See Gunnell v. Utah Valley State Coll, 152 F.3d 1258, 1264 (10th Cir.1998) (“[I]n our retaliation cases, we have liberally interpreted the phrase ‘adverse employment action’... Rather than defining a set rule ... this court takes a case-by-case approach to determining whether a given employment action is adverse.”).
But as the district court noted, Ms. Tabor’s argument faces a timing problem. The lowered P rating happened immediately after the interview, before her complaint to HR. She has alleged only that HR “kept” her at this lowered P rating after her complaint, which shows at most that Mr. Teel and Mr. Perkins discriminated against her before the HR complaint and that HR failed to correct their actions — not that the company engaged in an act of retaliation. Leaving the P rating where it was before the HR complaint is not such an act.
Both parties agree that Ms. Tabor requested field training to improve her P rating and that Hilti refused. She contends this refusal was retaliatory. But Hilti asserts a legitimate, non-retaliatory business reason: that no Customer Service employees were allowed to attend field training because the department was short-handed. Ms. Tabor fails to show that this reason was pretextual. The only admissible evidence she provides is that one individual, Mr. Smith, completed a one-day field ride on November 21, 2007. She offers no evidence to dispute Hilti’s assertion that Mr. Smith’s field ride was not field training but a one-day ride-along as part of his application for the Arkansas Account Manager position.
Because Ms. Tabor fails to establish any adverse employment action causally related to her HR complaint, her claim for retaliation fails. We therefore affirm the district court’s dismissal of this claim.
3. Disparate Impact
The district court rejected Ms. Tabor’s individual claim for disparate impact discrimination, finding that she failed to state a prima facie case because her statistical evidence did not compare promotion rates between qualified male employees with qualified female employees. The district court relied heavily on our decision in Carpenter v. Boeing Co., 456 F.3d 1183 (10th Cir.2006). We disagree with the district court’s application of Carpenter to Ms. Tabor’s case and reverse the summary judgment order on this claim.
Title VII forbids not only intentional discrimination based on disparate treatment but also “practices that are fair in form, but discriminatory in operation,” most often referred to as “disparate impact” discrimination. Lewis v. City of Chicago, - U.S. -, 130 S.Ct. 2191, 2197, 176 L.Ed.2d 967 (2010); see 42 U.S.C. § 2000e-2(k). The disparate impact “doctrine seeks the removal of employment obstacles, not required by business necessity, which create built-in headwinds and freeze out protected groups from job opportunities and advancement.” E.E.O.C. v. Joe’s Stone Crab, Inc., 220 F.3d 1263, 1274 (11th Cir.2000) (quotations omitted).
To survive summary judgment on an individual claim for disparate impact requires three steps. First, Ms. Tabor must establish a prima facie case that (a) an employment practice (b) causes a disparate impact on a protected group. See Carpenter, 456 F.3d at 1193. Second, if Ms. Tabor presents a prima facie case, the burden will shift to Hilti “to demonstrate that the challenged practice is job related for the position in question and consistent with business necessity.” 42 U.S.C. § 2000e-2(k)(l)(A)(i); Maldonado v. City of Altus, 433 F.3d 1294, 1304 (10th Cir. 2006), overruled on other grounds by Burlington N. & Santa Fe Ry. Co. v. White, 548 U.S. 53, 126 S.Ct. 2405, 165 L.Ed.2d *1221345 (2006). Third, assuming Hilti shows business necessity, Ms. Tabor may still prevail by “showing that the employer refuses to adopt an available alternative employment practice that has less disparate impact and serves the employer’s legitimate needs.” Ricci v. DeStefano, 557 U.S. 557, 578, 129 S.Ct. 2658, 174 L.Ed.2d 490 (2009); see 42 U.S.C. § 2000e-2(k)(l)(A)(ii).7
We turn to whether Ms. Tabor has established a prima facie case of disparate impact.
a. Challenged Employment Practice
“The first step in raising a disparate-impact claim is to identify the specific employment practice allegedly causing the discriminatory impact.” Carpenter, 456 F.3d at 1193. Ms. Tabor identifies Hilti’s GDCP system for this purpose. All parties agree that the GDCP system is facially neutral, but Ms. Tabor alleges that managers and supervisors at Hilti exercise discretion under the GDCP system in a discriminatory fashion. She alleges that Hilti supervisors choose to assign (or sometimes not assign) the subjective GDCP ratings, in particular P ratings, differently for male and female employees. She also alleges that Hilti managers used their discretion to waive GDCP minimum requirements to promote male employees with low or no P ratings, while requiring female employees to obtain a PI rating before applying for promotion.
Hilti argues that the GDCP system cannot be the basis of a disparate impact claim because the Supreme Court’s recent decision in Wal-Mart Stores, Inc. v. Dukes, — U.S. -, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011), rejects use of a discretionary practice as the basis of such a claim. But Hilti misunderstands the Supreme Court’s opinion in Wal-Mart and the Court’s precedent on this issue. The Court has said that discretionary practices may form the basis for an individual disparate impact claim. See Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 990, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988) (“[Disparate impact analysis is in principle no less applicable to subjective employment criteria than to objective or standardized tests.”). Wal-Mart did not disturb this precedent.8 In Wal-Mart, the *1222Court held that the company’s practice of leaving hiring and promotion decisions to the discretion of local supervisors without a more specific policy could not form the basis for class certification. See 131 S.Ct. at 2554-56. The issue in Wal-Mart was factual commonality across all plaintiffs in the class. The Court found it unlikely that thousands of Wal-Mart managers across different regions of the country “would exercise their discretion in a common way without some common direction.” Id. at 2555.
The Court explicitly acknowledged its prior holding in Watson and emphasized that different considerations are at issue in a class certification analysis compared with an individual disparate impact claim, namely the former’s focus on uniformity. Id. at 2554.
Ms. Tabor has identified a specific employment practice that meets the requirements for a disparate impact claim. We turn to whether she has shown that this practice caused a disparate impact.
b. Causation of a Disparate Impact on Women
The second part of the prima facie case requires Ms. Tabor to show the challenged employment practice caused a disparate impact on women. Ms. Tabor presented statistical evidence for this purpose. “Statistical evidence is an acceptable, and common, means of proving disparate impact.” Carpenter, 456 F.3d at 1196. In determining whether Ms. Tabor’s statistical evidence is sufficiently reliable to make a prima facie case, we are concerned with three issues: (1) the size of the disparity between male and female promotions; (2) the statistical significance of the disparity, measured by standard error rate or standard deviation; and (3) whether the statistical evidence effectively isolates the challenged employment practice. See id. at 1195, 1196, 1202; see also Watson, 487 U.S. at 994-96, 108 S.Ct. 2777.
i. Size of Disparity
A plaintiffs statistical evidence must show a significant disparity in the rate of employees in the protected group receiving an employment benefit or opportunity (e.g., promotion) compared with “the rate for the group with the highest rate.” Carpenter, 456 F.3d at 1202 (quoting 29 C.F.R. § 1607.4(D)); see also Shidaker v. Tisch, 833 F.2d 627, 631 (7th Cir.1986) (“In the use of statistical evidence to demonstrate the disparate impact of an allegedly discriminatory practice, the statistical disparity demonstrated must be significant or substantial.” (quotations omitted)). The disparity rate here measures the difference between male and female promotion rates.
The Equal Employment Opportunity Commission (“EEOC”) guidelines provide that a disparity of 20% or more in selection rate will be considered evidence of adverse impact in a disparate impact claim. 29 C.F.R. § 1607.4(D) (“[A] selection rate for any ... group that is less than four-fifths (4/5) (or eighty percent) of the rate for the group with the highest rate will generally be regarded ... as evidence of adverse impact.”). Although not controlling on courts, this guideline is persuasive. See Carpenter, 456 F.3d at 1202; Maldonado, 433 F.3d at 1305; Smith v. Xerox Corp., 196 F.3d 358, 365 (2d Cir.1999) overruled on other grounds by Meacham v. Knolls Atomic Power Lab., 461 F.3d 134 (2d Cir.2006).
*1223Ms. Tabor’s statistical evidence included a regression analysis performed by Mark R. Killingsworth, a professor of economics at Rutgers University. Dr. Killingsworth analyzed Hilti’s promotion data, controlling for employee age, area at Hilti, and tenure. He concluded that, between 2005 and 2008, the promotion rate for male inside sales representatives in Hilti’s Customer Service Department was 60% greater than the promotion rate for female inside sales representatives. After Hilti argued that this analysis did not cover the correct time frame, Dr. Killings-worth repeated the analysis using only data from Hilti’s proposed time frame of October 18, 2007, to December 31, 2008. These data showed that during Hilti’s preferred time frame “the rate of promotion for women (5.5 percent) was only about half the rate of promotion for men (10.8 percent).” Aplt. Appx., at 1598. In both instances, the disparity rate far exceeded the EEOC guideline of 20% and constituted a significant disparity.
ii. The Statistical Significance
It is not enough for a plaintiff to present data showing a disparity between groups. To be reliable, the result also must be statistically significant. Statistical significance measures the likelihood that the disparity between groups is random, i.e., solely the result of chance. It is expressed in terms of standard errors or standard deviations.9
“The Supreme Court has recognized that a disparity of more than two or three standard deviations in a large sample makes ‘suspect’ the contention that the differential occurs randomly.” Carpenter, 456 F.3d at 1195 (quoting Hazelwood Sch. Dist. v. United States, 433 U.S. 299, 308 n. 14, 97 S.Ct. 2736, 53 L.Ed.2d 768 (1977)). Ms. Tabor’s evidence was statistically significant at 2.777 standard errors. Dr. Kill-ingsworth concluded that “the probability is less than 0.006 that a disparity at least as large as this could occur solely as the result of chance factors, if promotions were unrelated to sex.” Aplt. Appx. at 1083.
iii. Isolating the Challenged Employment Practice
For Ms. Tabor’s statistical evidence to be reliable, it also must “isolat[e] and identify!] the specific employment practices that are allegedly responsible for any observed statistical disparities.” Watson, 487 U.S. at 994, 108 S.Ct. 2777. A plaintiff isolates the specific employment practice by controlling for key factors outside the challenged practice that could potentially cause or contribute to the disparity. See Carpenter, 456 F.3d at 1196. The data need not include “all measurable variables” but must be sufficient to prove discrimination by “a preponderance of the evidence.” Bazemore v. Friday, 478 U.S. 385, 400, 106 S.Ct. 3000, 92 L.Ed.2d 315 (1986).
The requirement to isolate the challenged employment practice is important because it goes directly to causation. The Supreme Court has emphasized that a plaintiff cannot establish her claim “simply by showing that, at the bottom line, there is [an] imbalance in the work force.” Wards Cove Packing Co. v. Atonio, 490 *1224U.S. 642, 657, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989) (quotations omitted), superseded by statute on other grounds by 42 U.S.C. § 2000e-2(k). The imbalance must actually be a result of the challenged employment practice. An employer will not, for example, be liable for a gender imbalance in its work force that “is due to a dearth of qualified [female] applicants (for reasons that are not [the employer’s] fault).” Id. at 651,109 S.Ct. 2115.
When the challenged employment practice involves employer discretion, the plaintiffs statistical analysis must “control for the constraints placed upon the deci-sionmaker’s discretion.” Carpenter, 456 F.3d at 1196. This is necessary “[especially in cases where an employer combines subjective criteria with the use of more rigid standardized rules or tests.” Watson, 487 U.S. at 994, 108 S.Ct. 2777; see also Anderson v. Westinghouse Savannah River Co., 406 F.3d 248, 266-67 (4th Cir.2005).
The issue therefore is whether there are any nondiscretionary factors missing from Ms. Tabor’s analysis that could explain the significant disparities in promotion rates between men and women. Ms. Tabor presented statistical evidence comparing the promotion rates of all female inside sales representatives with all male inside sales representatives in Hilti’s Customer Sendee Department. The district court concluded that she should have restricted this analysis to a narrower pool of “qualified” employees. Aplt. Appx. at 2818. We disagree.
At Hilti, discretion in assigning promotions was not constrained by mandatory objective criteria or by “rigid standardized rules or tests.” Watson, 487 U.S. at 994, 108 S.Ct. 2777. The GDCP system allowed supervisors and managers such broad discretion that there was no such thing as a true “qualified” subgroup for promotion. Hilti management exercised discretion in choosing whether to assign a P rating, in determining what the P rating would be, in allowing or not allowing employees to apply for promotions based upon or in spite of their P ratings, and in selecting employees for promotion either because of or irrespective of P ratings. Even under its own subjective definition of what made an employee “qualified,” Hilti promoted dozens of employees who were unqualified — 33 promoted employees had a P5 rating, indicating they were unqualified or ineligible to promote.
As previously discussed, the law requires plaintiffs to control for constraints placed on an employer’s discretionary choices. But in Ms. Tabor’s case, the record suggests the GDCP system is not subject to any such constraints. Nor has Hilti pointed to any mandatory, objective criteria that actually served to limit supervisor discretion. Ms. Tabor’s statistical evidence captured a broad pool of employees because Hilti selected employees for Account Manager promotions from the same broad pool. See Shidaker, 833 F.2d at 631 (“Where a company is shown to promote from within, the relevant labor pool of qualified applicants for upper level positions may be the group of employees in the company from which promotees will be drawn.”) (citing Hazelwood, 433 U.S. at 308 n. 13, 97 S.Ct. 2736).
In dismissing Ms. Tabor’s disparate impact claim, the district court relied heavily on our statement in Carpenter that “it is not enough for Plaintiffs to show simply that ... men get a higher percentage of ... assignments.... They must compare qualified men to qualified women.” 456 F.3d at 1194. But as we explain below, this conclusion failed to account for critical differences between Carpenter and the present case.
In Carpenter, a certified class of female employees at Boeing alleged that Boeing’s policy of giving supervisors discretion in *1225assigning overtime caused a disparate impact on female employees. 456 F.3d at 1188. The plaintiffs’ statistical evidence showed that male employees received more overtime assignments than female employees, but we rejected that evidence because it did not isolate the challenged employment practice of assigning overtime based on supervisor discretion. Other non-discretionary factors likely contributed to the disparity.
For instance, Boeing supervisors’ discretion was constrained by mandatory, objective criteria outlined in a collective bargaining agreement (“CBA”). Id. at 1194. The CBA required that any overtime offer be made first to the employee regularly assigned to the particular machine, job, crew, or position involved in the overtime work. Id. The plaintiffs’ statistical analysis failed to control for these CBA limitations. Id. at 1196,1198-99. It did not, for example, consider whether available overtime hours had been concentrated among particular crews or positions that had few or no female employees. Id. at 1202-03.10 We explained that because the statistical analysis failed to control for objective, mandatory constraints on supervisors’ discretion, it was not possible to reasonably infer that the discretion caused the disparate impact. Id. at 1202 (“Clearly, something in the overtime process consistently results in males obtaining more overtime ... than females” but because of its flaws, the data “tell[ ] us nothing about what that ‘something’ is.” (emphasis added)).
But Carpenters reasoning does not apply here. In Carpenter, only a narrow subgroup of employees was eligible, i.e., qualified, for overtime. That subgroup was defined by objective, mandatory criteria well before supervisor discretion came into play, and there was no evidence Boeing supervisors departed from CBA rules when assigning discretionary overtime.
As we have already explained, at Hilti, the exercise of supervisor or manager discretion and the process of defining “qualified” employees were one and the same. All inside sales representatives in the Customer Service Department were subject to the challenged discretionary employment practice, the GDCP system, and we do not see any objective factors that genuinely limited the pool of employees from which Hilti selected its promotees — nor has Hilti pointed to any. In theory, the subjective GDCP factors should have limited the pool of employees eligible for promotion. But Hilti plainly did not treat GDCP ratings as mandatory, since it promoted employees who had substandard P ratings or no P ratings.11
*1226In short, the district court erred in requiring Ms. Tabor to identify a subgroup of “qualified” employees at Hilti when no such subgroup existed. Ms. Tabor’s statistical evidence isolates the challenged employment practice enough to raise “an inference of causation.” Watson, 487 U.S. at 994-95, 108 S.Ct. 2777; see also Aiken v. City of Memphis, 37 F.3d 1155, 1163 (6th Cir.1994) (holding that statistical evidence comparing promotion rates among all entry-level officers and firefighters was sufficiently reliable to show disparate impact on the basis of race).12
We conclude that Ms. Tabor has stated a prima facie disparate impact claim and we remand to the district court for further proceedings consistent with this opinion.
B. Ms. Gray’s Individual Claims
Ms. Gray brings two individual claims for gender discrimination under Title VII. The first claim charges Hilti with intentional discrimination for failure to promote based on theories of deterrence and failure to train. The second claim charges Hilti with disparate impact discrimination.
1. Failure to Promote/Deterrence
Ms. Gray did not actually apply for an Account Manager position but asserts that she desired this promotion and she was deterred from pursuing it because of Hil-ti’s intentionally discriminatory actions. The district court held that Ms. Gray failed to state a prima facie case for intentional discrimination because she failed to demonstrate she was qualified for promotion. We affirm.
The Supreme Court has allowed discrimination claims based on a theory of deterrence, reasoning that “[a] consistently enforced discriminatory policy can surely deter job applications from those who are aware of it and are unwilling to subject themselves to the humiliation of explicit and certain rejection.” Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 326, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). To establish a discrimination claim based on deterrence, a plaintiff must show: (1) there were promotional opportunities that were filled by males; (2) she was qualified and available for the job; (3) despite her qualifications she was not promoted; and *1227(4) the employer intentionally discriminated against her. See Sprague v. Thorn Ams., Inc., 129 F.3d 1355, 1362 (10th Cir.1997).
Ms. Gray fails to establish the second factor because she has not shown she was qualified for a promotion. She fails to rebut evidence of disciplinary and performance issues that would likely have prevented her from being selected for any promotion at Hilti. Multiple managers testified in their depositions that Ms. Gray had been warned about many issues, including excessive tardiness and absenteeism, poor attitude, and sleeping at her desk. Ms. Gray argues that she has raised a genuine issue of fact regarding Hilti’s reasons for not promoting her and not encouraging her to pursue promotion, pointing to Mr. Brown’s comments that “women do not make it out in the field” and the fact that Mr. Brown refused to allow her to participate in field training. Aplt. Appx. at 2807.
Although Mr. Brown’s comments may reflect a discriminatory view, Ms. Gray cannot establish a cause of action for intentional discrimination unless she proves that she was qualified for promotion. Unlike in Ms. Tabor’s case, the negative evaluations of Ms. Gray’s performance do not come only or primarily from individuals associated with discriminatory remarks. Two managers other than Mr. Brown testified to concerns regarding Ms. Gray’s work performance and discipline.
We therefore affirm the district court’s dismissal of Ms. Gray’s intentional discrimination claim.
2. Disparate Impact
The district court did not address Ms. Gray’s individual disparate impact claim in its summary judgment order. Where an issue has not been ruled on by the court below, we generally favor remand for the district court to examine the issue. See In re R. Eric Peterson Constr. Co., 951 F.2d 1175, 1182 (10th Cir.1991). We therefore remand to the district court to address Hilti’s motion for summary judgment on this claim.
We note, however, that Ms. Gray’s individual disparate impact claim may face one challenge that Ms. Tabor’s does not. Even when an employer is shown to have violated Title VII’s disparate impact provision, “[e]ach person seeking individual relief [must] show that ... she ... suffered an adverse employment decision and therefore was a potential victim of the proved discrimination.” Chin, 685 F.3d at 151; see also Coe, 646 F.2d at 451 (“[I]n individual actions rather than class actions ... [i]t is not sufficient for an individual plaintiff to show that the employer followed a discriminatory policy without also showing that plaintiff [herself] was injured.”).
As we discuss above, Hilti has offered undisputed evidence that multiple managers warned Ms. Gray about performance and disciplinary problems. If the district court determines there is no genuine dispute that Ms. Gray was unqualified for promotion based upon criteria not connected to the challenged employment practice, then summary judgment in Hilti’s favor is appropriate.
C. Class Certification
1. Background
Plaintiffs appeal the district court’s denial of their motion for class certification. The decision whether to grant or deny class certification “involves intensely practical considerations,” Reed v. Bowen, 849 F.2d 1307, 1309 (10th Cir.1988), and therefore “belongs within the discretion of the trial court,” Monreal v. Potter, 367 F.3d 1224, 1235 (10th Cir.2004) (quotations omitted). In ruling on a class certification question, the court is not limited to the pleadings but may “probe be*1228hind the pleadings” and examine the facts and evidence in the case. Gen. Tel. Co. v. Falcon, 457 U.S. 147, 160, 102 S.Ct. 2364, 72 L.Ed.2d 740 (1982).
The requirements for class certification are outlined in Rule 23 of the Federal Rules of Civil Procedure. To certify a class, Plaintiffs must first meet all of four requirements outlined in Rule 23(a), namely (1) numerosity, (2) commonality, (3) typicality, and (4) adequacy of representation. Second, Plaintiffs must show that at least one of three conditions defined in Rule 23(b) is satisfied. The only Rule 23(b) condition at issue on appeal is 23(b)(3), which allows class certification when the district court finds that common questions of law or fact predominate over individualized questions.
Plaintiffs moved to certify a proposed class of “approximately 294 women in inside sales who were denied opportunities to promote to an outside sales position at Hilti.” Aplt. Appx. at 1672 (quotations omitted). The district court refused to certify Plaintiffs’ proposed class, finding they had failed to meet the numerosity requirement under Rule 23(a)(1) and failed to satisfy any conditions under Rule 23(b). Because failure to satisfy Rule 23(a)(1) is dispositive, the court did not consider whether Plaintiffs met the other three requirements of Rule 23(a).
After the district court issued its decision and while this appeal was pending, the Supreme Court decided Wal-Mart Stores, Inc. v. Dukes, — U.S. -, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011), which substantially clarified the Rule 23(a)(2) commonality requirement. Applying the Court’s most recent guidance, we affirm the denial of Plaintiffs’ motion for class certification on the ground that Plaintiffs have not shown “there are questions of law or fact common to the class.” Fed. R.Civ.P. 23(a)(2); Wal-Mart, 131 S.Ct. at 2549. Although failure to satisfy any requirement under Rule 23(a) is dispositive, we also briefly discuss the district court’s Rule 23(b)(3) analysis as it relates to the commonality issue in Rule 23(a)(2).
2. Rule 23(a)(2)
“A party seeking class action certification must demonstrate, under a strict burden of proof, that all of the requirements of 23(a) are clearly met.” Rex v. Owens ex rel. State of Okla., 585 F.2d 432, 435 (10th Cir.1978). With respect to Rule 23(a)(2)’s commonality requirement, a plaintiff must show that class members “have suffered the same injury,” Gen. Tel. Co., 457 U.S. at 157, 102 S.Ct. 2364, and that “the incidents of discrimination complained of ... present a common issue that could be resolved efficiently in a single proceeding,” McReynolds v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 672 F.3d 482, 488 (7th Cir.2012), cert. denied, Merrill Lynch, Pierce, Fenner & Smith, Inc. v. McReynolds, — U.S. -, 133 S.Ct. 338, 184 L.Ed.2d 157 (2012).
In Wal-Mart, a nationwide class of current and former female employees sued Wal-Mart under Title VII. 131 S.Ct. at 2548. They alleged that Wal-Mart’s policy of leaving personnel decisions to the discretion of local managers had a disparate impact on women and resulted in disparate treatment toward them. Id. The Court explained that “in resolving an individual’s Title VII claim, the crux of the inquiry is the reason for the particular employment decision.” Id. at 2552. “What matters to class certification ... is not the raising of common ‘questions’ — even in droves — but, rather the capacity of a classwide proceeding to generate common answers apt to drive the resolution of litigation.” Id. (quotations omitted).
The Court found that the Wal-Mart policy of allowing broad discretion to local managers “is just the opposite of a uniform *1229employment practice that would provide the commonality needed for a class action.” Id. at 2554. It held that neither the disparate impact nor the disparate treatment sex discrimination claim met the commonality requirement. In arriving at this conclusion, the Court emphasized the extraordinary size and scope of the proposed class, which included 1.5 million female employees at 3,400 Wal-Mart stores. It was unlikely that thousands of Wal-Mart managers across different regions of the country would exercise their discretion in a common way without some common direction, and the plaintiffs had “not identified a common mode of exercising discretion that pervades the entire company.” Id. at 2554-55.
After Wal-Mart, federal courts reviewing class certification questions have generally denied certification when allegedly discriminatory policies are highly discretionary and the plaintiffs do not point to “a common mode of exercising discretion that pervades the entire company.” In re Wells Fargo Residential Mortg. Lending Discrimination Litg., No. 08-MD-01930 MMC, 2011 WL 3903117, at *4-5 (N.D.Cal. Sept. 6, 2011); see also, e.g., Rodriguez v. Nat’l City Bank, 277 F.R.D. 148, 154-55 (E.D.Pa.2011); Daskalea v. Wash. Humane Soc’y, 275 F.R.D. 346, 360 (D.D.C. 2011). Other courts have allowed certification of smaller plaintiff classes in cases challenging policies that grant only limited discretion to supervisors. See, e.g., McReynolds, 672 F.3d at 488-89, 492 (allowing class certification in Title VII claim where plaintiffs pointed to a uniform company policy that based account distributions on employees’ past success and gave limited discretion to managers); Ross v. RBS Citizens, N.A., 667 F.3d 900, 909-10 (7th Cir.2012) (allowing class certification in FLSA claim where plaintiffs pointed to a uniform, unofficial company-wide policy compelling employees to work without overtime).
In the current case, Plaintiffs challenge a highly discretionary policy for granting promotions. They have not shown that Hilti maintained “a common mode of exercising discretion that pervade[d] the entire company.” 131 S.Ct. at 2554-55. To the contrary, the record suggests that Hilti failed to maintain the GDCP system in any uniform manner. Even if Plaintiffs’ statistical evidence demonstrates (at least facially) that this haphazard policy caused an overall disparate impact on women, Plaintiffs have not shown that the facts and circumstances involved in Hilti’s promotion choices are common across the class of female employees. The very circumstances at issue in the individual claims of the two current plaintiffs illustrate this point: Ms. Tabor received a PI rating and applied for at least two promotions, whereas Ms. Gray did not receive a PI rating, did not apply for promotion, and apparently received disciplinary warnings. In each case, Hilti offers a very different defense to allegations of discrimination.
Given the broad discretion involved in Hilti’s alleged discriminatory employment practice and the highly individualized facts and circumstances raised in each employment decision, we cannot say that the proposed class “presents] common issue[s] that could be resolved efficiently in a single proceeding.” McReynolds, 672 F.3d at 488; see Fed.R.Civ.P. 23(a)(2).
3. Rule 23(b)(3)
Although our determination that Plaintiffs’ proposed class fails to meet Rule 23(a)(2) is dispositive, we briefly address the district court’s 23(b)(3) analysis because both requirements hinge on some of the same fact-intensive questions.
In addition to meeting all Rule 23(a) requirements, to certify a class, “Plaintiffs must satisfy at least one subsection of *1230Rule 23(b).” Monreal, 367 F.3d at 1235. Rule 23(b) has three subsections, but only subsection (b)(3) is raised in this appeal. Rule 23(b)(3) allows certification of a class when the court finds that “questions of law or fact common to class members ’predominate over any questions affecting only individual members, and that a class action is therefore superior to other available methods for fairly and efficiently adjudicating the controversy.” Fed.R.Civ.P. 23(b)(3) (emphasis added).
In determining whether common issues of law and fact predominate over individual ones, the following considerations are relevant:
(A) the interest of members of the class in individually controlling the prosecution or defense of separate actions;
(B) the extent and nature of any litigation concerning the controversy already commenced by or against members of the class;
(C) the desirability or undesirability of concentrating the litigation of the claims in the particular forum; and
(D) the difficulties likely to be encountered in the management of a class action.
Fed.R.Civ.P. 23(b)(3).
The district court noted that “[t]he defendants allege that both named plaintiffs were denied promotion for specific, objective, and individualized reasons.” Aplt. Appx. at 1681. The court determined that Hilti’s factual defenses “raised the inference that ... individualized concerns predominate over the common questions.” Id. It therefore concluded that Hilti’s promotion decisions involve “highly individualized” facts and defenses that cannot be effectively resolved in a class suit. Id. We agree with this analysis.
We therefore affirm the district court’s refusal to certify the class.
III. CONCLUSION
We affirm the grant of summary judgment as to Ms. Tabor’s retaliation claim and Ms. Gray’s failure to promote claim. We further affirm the district court’s refusal to certify Plaintiffs’ proposed class. We remand Ms. Gray’s disparate impact claim because the district court failed to analyze the claim in its order. Finally, we reverse the grant of summary judgment as to Ms. Tabor’s failure to promote and disparate impact claims and remand for further proceedings consistent with this opinion.