Case Information
UNITED STATES DISTRICT COURT NORTHERN DISTRICT OF CALIFORNIA DEREK MOBLEY, Case No. 23-cv-00770-RFL Plaintiff, ORDER GRANTING PRELIMINARY
v. COLLECTIVE CERTIFICATION Re: Dkt. No. 106 WORKDAY, INC., Defendant.
Derek Mobley brings this action for employment discrimination against Workday, Inc., alleging that Workday’s artificial intelligence (“AI”)-based applicant recommendation system discriminated against job applicants on the basis of race, age, and disability. Mobley is joined by four other plaintiffs over the age of forty, who allege that they too have applied for hundreds of jobs via Workday and have been rejected almost every time without an interview, allegedly because of age discrimination in Workday’s AI recommendation system. Mobley now moves for preliminary certification of a collective action on the age discrimination claim, which would allow him to notify similarly situated individuals of the lawsuit and provide them an opportunity to opt-in to having their claims heard on a collective basis. (Dkt. No. 106.) After additional discovery, Workday would then have an opportunity to present evidence that the collective is not, in fact, similarly situated. Workday could then ask the Court to revisit its preliminary decision by de-certifying the collective and requiring each plaintiff to proceed individually. Mobley’s motion for preliminary certification of the collective is GRANTED .
The proposed collective is similarly situated because Mobley has substantially alleged the existence of a unified policy: the use of Workday’s AI recommendation system to score, sort, rank, or screen applicants. The critical issue at the heart of Mobley’s claim is whether that system has a disparate impact on applicants over forty. That issue is susceptible to common proof—it cuts across the proposed collective, regardless of the degree to which particular employers place weight on those recommendations, the extent to which the system makes discriminatory recommendations across different employers, or the natural variations in the qualifications or rejection rate for particular members of the proposed collective. At this stage, that is sufficient. The proposed collective need not be identical in all ways, because its members are alike in the central way that matters: they were allegedly required to compete on unequal footing due to Workday’s discriminatory AI recommendations.
Workday has raised concerns that there may be logistical hurdles to identifying members of the collective. On the current record, those challenges do not appear insurmountable. And, of course, challenges in identifying the collective are not an excuse for denying notice altogether. Nor does the size of the collective provide a basis to decline to provide notice. If the collective is in the “hundreds of millions” of people, as Workday speculates, that is because Workday has been plausibly accused of discriminating against a broad swath of applicants. Allegedly widespread discrimination is not a basis for denying notice. The parties shall meet and confer and engage in further targeted discovery to develop a notice plan. If necessary, publication notice via social media or electronic notice to applicants using Workday’s platform could be considered, but only if targeted notice cannot be accomplished.
I. BACKGROUND
A. Allegations and Supporting Declarations
Mobley seeks preliminary certification of a nationwide collective as to his Age Discrimination in Employment Act (“ADEA”) claim. The proposed collective includes “[a]ll individuals aged 40 and over who, from September 24, 2020, through the present, [] applied for job opportunities using Workday, Inc.’s job application platform and were denied employment recommendations.” (Dkt. No. 106 at 14.) [1]
Workday allegedly provides “human resource management services” including applicant screening services on a subscription basis to businesses spanning numerous different industries. (Dkt. No. 47 (“FAC”), at 2, ¶¶ 89–90.) Workday provides a platform on the customer’s website to collect, process, and screen job applications. ( See id. at 2, ¶¶ 49–55.) Workday’s website states that it can “reduce time to hire by automatically dispositioning or moving candidates forward in the recruiting process.” ( Id. ¶ 94.) Workday allegedly “embeds artificial intelligence . . . into its algorithmic decision-making tools, enabling these applications to make hiring decisions.” ( Id. ¶ 99.) Workday’s applicant screening tools allegedly integrate “pymetrics” that “use neuroscience data and AI,” in combination with existing employee referrals and recommendations. ( Id. ¶¶ 100–01.) According to Mobley, these tools “determine whether an employer should accept or reject an application” and are designed in a manner that reflects employer biases and relies on biased training data. ( at 3, ¶¶ 28, 38–48, 102–03.) Mobley alleges that Workday’s AI recommendation system can score, sort, rank, or screen an applicant and provide that data to the employer. ( Id. at 2–3, ¶¶ 28, 39, 98–101, 110–12, 120.) In many cases, an applicant can advance in the hiring process only if they get past Workday’s screening algorithms. ( Id . ¶ 98.)
Although discovery is ongoing, the parties’ preliminary collective certification briefing discusses two Workday tools. The first tool, Candidate Skills Match (“CSM”), operates within a subscription service called “Workday Recruiting” to “extract[] skills in the employer’s job posting” and the applicant’s materials “and determine the extent to which the applicant’s skills match the role to which they applied. The results of CSM are reported [to the employer] as ‘strong,’ ‘good,’ ‘fair,’ ‘low,’ ‘pending,’ and ‘unable to score.’” (Dkt. No. 109-6 at 4.) The second tool, Workday Assessment Connector (“WAC”) is alleged to use machine learning to “observe that a client-employer disfavors certain candidates who are members of a protected class, [and] decrease the rate at which it recommends those candidates.” (FAC ¶¶ 39, 100–01; Dkt. No. 107 at 29–30.) Workday stated at oral argument—without introducing any evidence on the issue—that WAC acts as a bridge to allow employers to access only “third-party” AI features. (Dkt. No. 120 at 8:6–17; Dkt. No. 107 at 29 n. 6.) Via declaration, Workday also asserts that customers with Workday Recruiting subscriptions may “enable, disable, use, or ignore its many features,” including its AI features. (Dkt. No. 107 at 9 (citing Dkt. No. 108 ¶ 10).)
Since 2017, Mobley has allegedly applied to over 100 positions with companies that use Workday’s screening features for talent acquisition and hiring. (FAC ¶ 49.) Mobley was allegedly denied employment for every one of the 100-plus applications that he submitted to companies using Workday’s platform. ( ¶ 88.) In support of his motion, Mobley submits the declarations of four Opt-In Plaintiffs who likewise state that they are over forty and have submitted “hundreds of employment applications to prospective employers” through the Workday system and have been rejected nearly every time. ( See Dkt. No. 106 at 6, 9–13.) Each Opt-In Plaintiff alleges that they received at least one “automated” rejection email for a job for which they met the qualification requirements. ( Id. )
B. Motion to Dismiss Order
Mobley has already been found to have stated a plausible ADEA claim on a disparate impact theory. As explained in the order denying Workday’s motion to dismiss, Mobley has alleged that “Workday’s customers delegate traditional hiring functions, including rejecting applicants, to the algorithmic decision-making tools provided by Workday.” (Dkt. No. 80 at 9.) Specifically, “Workday’s software is not simply implementing in a rote way the criteria that employers set forth, but is instead participating in the decision-making process by recommending some candidates to move forward and rejecting others.” ( Id. ) “Given Workday’s allegedly crucial role in deciding which applicants can get their ‘foot in the door’ for an interview,” Mobley had plausibly alleged that “Workday’s tools are engaged in conduct that is at the heart of equal access to employment opportunities.” ( Id. ) Mobley has also adequately alleged that these tools created a discriminatory disparity. ( at 15.)
II. LEGAL STANDARD
Mobley seeks preliminary collective certification of his ADEA claim. The ADEA prohibits employers from discriminating against a qualified individual on the basis of a disability “in regard to job application procedures” and “the hiring” of employees. 42 U.S.C. § 12112(a). In proving a claim for disparate impact under the ADEA “a plaintiff must (1) show a significant disparate impact on a protected class or group; (2) identify the specific employment practices or selection criteria at issue; and (3) show a causal relationship between the challenged practices or criteria and the disparate impact.” Bolden-Hardge v. Off. of Cal. State Controller , 63 F.4th 1215, 1227 (9th Cir. 2023).
The ADEA is enforced “in accordance with the powers, remedies, and procedures” of the Fair Labor Standards Act (“FLSA”). 29 U.S.C. § 626(b); Heath v. Google LLC , 345 F. Supp. 3d 1152, 1164 (N.D. Cal. 2018). Section 216(b) of the FLSA provides that one or more employees may bring a collective action against any employer on behalf of “themselves and other employees similarly situated.” 29 U.S.C. § 216(b). Under this provision, “workers may litigate jointly if they (1) claim a violation of the FLSA, (2) are ‘similarly situated,’ and (3) affirmatively opt into the joint litigation.” Campbell v. City of Los Angeles , 903 F.3d 1090, 1100 (9th Cir. 2018). The FLSA does not define the term “similarly situated.” See 29 U.S.C. § 216(b); Campbell , 903 F.3d at 1100 (“Given the[] gaps [in the FLSA], much of collective action practice is a product of interstitial judicial lawmaking or ad hoc district court discretion.”). However, “what matters is not just any similarity between party plaintiffs, but a legal or factual similarity material to the resolution of the party plaintiffs’ claims, in the sense of having the potential to advance these claims, collectively, to some resolution.” Campbell , 903 F.3d at 1115 (emphasis in original). Frequently, plaintiffs satisfy the “similarly situated” analysis with “substantial allegations, supported by declarations or discovery, that the putative class members were together the victims of a single decision, policy, or plan.” See Villa v. United Site Servs. of Cal., Inc. , No. 12-cv-318-LHK, 2012 WL 5503550, at *13 (N.D. Cal. Nov. 13, 2012) (collecting cases). If such similarities exist, the collective may be granted preliminary certification even if there are differences that will ultimately require a degree of individual adjudication. Id. at *14 (“certification is proper based on the allegations that Defendant operated under [a] common policy, even though to prove liability, individual plaintiffs may still have to prove that they never actually took their breaks, and thus were underpaid”).
In Campbell , the Ninth Circuit described with approval the common practice in this district of using a two-step process to determine whether a collective is “similarly situated.” See Campbell , 903 F.3d at 1110 (finding that “[t]here is good reason for th[e] consensus” on the two- step approach). First, “plaintiffs will typically move for preliminary certification.” Id. at 1109. At this stage, the standard of review is “loosely akin to a plausibility standard, commensurate with the stage of the proceedings,” and “the district court’s analysis is typically focused on a review of the pleadings but may sometimes be supplemented by declarations or limited other evidence.” Id . [2] If preliminary certification is granted, the “sole consequence . . . is the sending of court-approved written notice to workers who may wish to join the litigation as individuals.” Id. at 1101 (quoting Genesis Healthcare Corp. v. Symczyk , 569 U.S. 66, 75 (2013)).
After a collective has secured preliminary certification, the second stage of the analysis “will come at or after the close of relevant discovery.” Id. at 1109. “The employer can move for ‘decertification’ of the collective action for failure to satisfy the ‘similarly situated’ requirement in light of the evidence produced to that point.” (citing 1 McLaughlin on Class Actions § 2:16 (21st ed. 2024); 7B Fed. Prac. & Proc. Civ. § 1807 (3d ed. 2025)). “The district court will then take a more exacting look at the plaintiffs’ allegations and the record.” Id. “[T]he two-step process, culminating in a decertification motion on or after the close of relevant discovery, has the advantage of ensuring early notice of plausible collective actions, then eliminating those whose promise is not borne out by the record.” Id. at 1110.
III. DISCUSSION
A. Burden of Proof
As an initial matter, Workday argues that Mobley should be held to a significantly higher burden of proof at the first stage of the “similarly situated” analysis because the parties have already engaged in some discovery. (Dkt. No. 107 at 11, 31–32.) In essence, Workday proposes a sliding scale, in which the more discovery has occurred, the more evidence plaintiff should have to provide. At oral argument, consistent with that approach, Workday repeatedly faulted Mobley for relying on his well-pled allegations and for failing to come forward with sufficient evidence. The Court declines to apply a sliding-scale standard, which is contrary to the procedure used by virtually all district courts in this circuit. Such an approach would be impossible to apply with consistency, would encourage gamesmanship, and is contrary to the policy considerations laid out in Campbell .
Unlike the straightforward standard described in Campbell , a sliding scale standard would require a court to weigh competing evidence on an incomplete record, attempting to fairly account for what percentage of discovery has yet to be exchanged. Such a system would lead to inconsistent rulings. How much weight should a court assign conflicting evidence proffered by a defendant when the court estimates that only 20% of relevant discovery has been exchanged? What burden of proof should a plaintiff be held to where 90% of discovery is complete? What should a court do where, as here, the parties disagree on how much critical discovery remains to be exchanged? Even more concerning, the system would encourage gamesmanship. Parties could withhold critical discovery until after preliminary collective certification in an effort to win an early—and potentially dispositive—evidentiary battle. It would also allow defendants to engineer a higher burden of proof for a plaintiff simply by negotiating for a lengthier discovery period prior to preliminary certification, just as Workday did here in its proposed schedule at the initial case management conference. ( See Dkt. No. 92.)
Campbell endorses a two-step process where step one is “loosely akin to a plausibility standard” and step two “resemble[s] a motion for partial summary judgment.” Campbell , 903 F.3d at 1109. Its reasons for doing so are clear. The system “ensur[es] early notice of plausible collective actions” to the interested parties, and “then eliminating those whose promise is not borne out by the record.” Id. at 1110. Workday has provided no compelling reason for departing from the Ninth Circuit’s directive here, especially where Workday’s own evidentiary submission is limited to excerpts of Plaintiffs’ deposition transcripts, four interrogatory responses, and a single declaration. (Dkt. Nos. 108, 109.) Mobley’s motion will therefore be assessed under the “substantial allegations” of similarity standard set forth above.
B. Mobley’s Proposed Collective
Mobley argues that the proposed collective is “similarly situated” because all proposed members, like Mobley and the Opt-In Plaintiffs, were “together the victims of a single . . . policy”: “Workday’s artificial intelligence and machine learning screening products [which] treat applicants over age 40 discriminatorily and have a disparate impact against them.” (Dkt. No. 106 at 15, 18.) He points to the allegations in the complaint, which have already been found plausible, and the declarations of Opt-In Plaintiffs who similarly experienced a high rejection rate of their applications processed through Workday. ( at 18.)
Mobley has substantially alleged an “identifiable factual or legal nexus [that] binds together the various claims of the [collective] members in a way that hearing the claims together promotes judicial efficiency and comports with the broad remedial policies underlying the FLSA,” making the proposed collective similarly situated. See Rabin v. PricewaterhouseCoopers LLP , No. 16-cv-2276-JST, 2018 WL 3585143, at *4 (N.D. Cal. July 26, 2018) (citing Russell v. Wells Fargo & Co. , No. 07-cv-3993-CW, 2008 WL 4104212, at *3 (N.D. Cal. Sept. 3, 2008)). Specifically, he has shown that the elements of his prima facie disparate impact claim are susceptible to common proof. Because Mobley has identified a unified policy applicable to all the putative collective members, common evidence can be used to “(1) show a significant disparate impact on [individuals over 40]; (2) identify the specific employment practices or selection criteria at issue; and (3) show a causal relationship between the challenged practices or criteria and the disparate impact.” See Bolden-Hardge , 63 F.4th at 1227 (listing the elements of a prima facie disparate impact case). Even under the more exacting “commonality” standard, the Ninth Circuit has found that the identification of a uniform policy that allegedly generated a disparate impact creates a common question as to a prima facie case of disparate impact age discrimination. See Stockwell v. City & Cnty. of San Francisco , 749 F.3d 1107, 1114–15 (9th Cir. 2014); see also Wal-Mart Stores, Inc. v. Dukes , 564 U.S. 338, 350–56 (2011) (“If the employer ‘used a biased testing procedure to evaluate [] applicants for employment. . . , a class action on behalf of every applicant . . . who might have been prejudiced by the test clearly would satisfy the commonality and typicality requirements of Rule 23(a).’”) (quoting Gen. Tel. Co. Sw. v. Falcon , 457 U.S. 147, 159 n. 15 (1982)).
The mechanisms behind Workday’s AI recommendation system, the effects on applicants who are over 40, and whether those “effects amount to a disparate impact on account of age, [] will be so for all [collective] members or for none; their claims rise and fall together.” Stockwell , 749 F.3d at 1115. Therefore, as discussed in further detail below, preliminary collective certification is proper.
C. Challenges to Any Proposed Collective
Workday advances several arguments for why it is impossible for any collective to be similarly situated under Mobley’s theory of the case. First, Workday claims that it “does not offer ‘employment recommendations,’ so no one belongs in this collective, including [Mobley].” (Dkt. No. 107 at 31 (emphasis in original).) Second, Workday argues that the policy Mobley has identified is not uniform, defeating preliminary collective certification. Finally, Workday suggests that the natural variation in the proposed collective members’ qualifications for the jobs to which they applied, number of jobs applied to, and rejection rate mean that no collective could ever be similarly situated. Each of these arguments is addressed in turn.
1. Whether Workday “Recommends” Applicants Citing its interrogatory responses, Workday argues that “Workday does not recommend, screen out, or otherwise assess or predict applicants’ likelihood of success in a role.” (Dkt. No. 107 at 31.) Therefore, Workday claims that no one has ever been “denied employment recommendations” by Workday and the proposed collective has no members. But this position is contrary to the well-pled allegations in the complaint (incorporating Workday’s statements on its own website), as well as Workday’s interrogatory responses. For example:
• Workday’s website states that “[o]ur skills intelligence foundation helps you build diverse teams by expanding candidate pools with equitable, AI- and ML-driven job recommendations .” (FAC ¶ 111 (emphasis added).) [3] • In discovery, Workday stated that: “Workday customers who purchase Workday Recruiting have access to Candidate Skills Match (‘CSM’), which they can choose to turn on or off. If a customer chooses to turn on CSM, CSM utilizes artificial intelligence to parse an employer’s job posting and an applicant’s application and/or resume; extract skills in the employer’s job posting, on the one hand, and skills from the application and/or resume on the other hand; and determine the extent to which the applicant’s skills match the role to which they applied. The results of CSM are reported as ‘strong,’ ‘good,’ ‘fair,’ ‘low,’ ‘pending,’ and ‘unable to score .’” (Dkt. No. 109-6 at 4 (emphasis added).)
• Plaintiff alleges that Workday’s “algorithms are only trained on incumbent employees at a company, allowing the pymetrics Workday Assessment Connector to build a homogenous workforce not representative of the applicant pool.” (FAC ¶ 102.) Workday fails to explain why the conduct described above does not constitute “recommend[ing], screen[ing] out, or otherwise assess[ing] or predict[ing] applicants’ likelihood of success in a role.” Workday appears to take the position that because its AI recommendation system supposedly cannot auto-reject applicants without some degree of participation by the employer, Workday does not “recommend.” But Mobley’s disparate impact claim is based on the theory that Workday’s AI “participat[es] in the decision-making process,” a concept that is broader than auto-rejections. (Dkt. No. 80 at 9.) Therefore, even if Workday is taken at its word that its AI recommendation system cannot auto-reject an applicant, Workday is incorrect that the proposed collective is memberless. An individual who received a “low” CSM score, for example, could qualify as having been denied an employment recommendation. Similarly, Mobley has substantially alleged that the WAC operates to recommend certain applicants over others in a way that causes a disparate impact.
Questions remain regarding where to draw the line between someone who was and someone who was not “recommended.” Does anything less than a “strong” CSM score qualify as a denial of a recommendation? How should the results of any other features be measured? However, the fact that these questions remain outstanding does not mean that no collective may be certified. See , e.g. , Heath v. Google Inc. , 215 F. Supp. 3d 844, 859 (N.D. Cal. 2016) (certifying the collective and “instruct[ing] the parties to meet and confer on how to use applicants’ graduation date information as a proxy for age . . . to identify the candidates to whom the third-party administrator will send notice”). Where the court has found that a collective should receive preliminary certification, it cannot withhold notice simply because it will be difficult to identify individual putative members. See Campbell , 903 F.3d at 1110 (stating that “in a valid collection action, ‘forbid[ding] the sending of notice altogether’ would be an abuse of discretion”) (quoting Woods v. N.Y. Life Ins. Co. , 686 F.2d 578, 580 (7th Cir. 1982)); see also Carrillo v. Schneider Logistics, Inc ., No. 11-cv-08557, 2012 WL 556309, at *12–13 (C.D. Cal. Jan. 31, 2012) (authorizing alternative notice where members of the collective could not be easily identified), aff’d on other grounds , 501 F. App’x 713 (9th Cir. 2012). Accordingly, the parties are ordered to meet and confer to determine how to define a “recommendation,” along with the other issues further detailed below.
2. Whether Mobley Has Identified a Unified Policy Workday next argues that Mobley has failed to show that any uniform policy applies to all applications submitted through Workday, and that “[w]ithout that uniformity, there can be no common decision, policy, or plan justifying collective adjudication.” (Dkt. No. 107 at 29.) Specifically, Workday points out that employers can choose whether or not to use Workday’s AI features. ( at 30.) Furthermore, with respect to the “Workday Assessment Connector” feature which Mobley alleges is used to “build a homogenous workforce” at a particular company, Defendant argues that whether or not this feature causes a disparate impact would necessarily vary based on the existing makeup of the employer’s incumbent workforce. ( at 29–30.)
As to the use of AI features by Workday’s clients, the proposed collective already encompasses only individuals whose applications were subject to Workday’s AI because the employer turned on AI features. By limiting the collective to those who were “denied employment recommendations” by Workday, the collective necessarily includes only individuals who were scored, sorted, ranked, or screened by Workday’s AI and who either had this information conveyed to the employer or who had their application withheld from the employer by Workday. That is because an individual who was scored, sorted, ranked, or screened based only on criteria that were specifically selected by the employer alone has not been “denied [an] employment recommendation” by Workday , even if the employer-directed screening was effectuated through Workday. In that hypothetical scenario, Workday would be no different from an excel spreadsheet operated by the employer that sorted employees by age, so that the employer could refuse to interview those above a certain cutoff. (Dkt. No. 80 at 11.) As previously held in the motion to dismiss order, Workday would not be liable under an agency theory in that scenario. ( Id. ) Accordingly, the individual would not be part of the proposed collective. The unified policy that Mobley has identified is Workday’s AI recommendation system, and the proposed collective reflects that limitation. Workday will have the opportunity at a later stage of the case to present evidence that its AI system does not operate in a unified way.
Workday’s argument that certain of its AI features could vary in their impact across different employers does not defeat the existence of common issues either. In nearly every large disparate impact case, certain units of the whole—for example, regions of a national company or divisions of an organization—will demonstrate the effects of a unified discriminatory policy to a greater or lesser extent than others, or may demonstrate no discriminatory effect at all. But where a unified policy exists and the net disparate impact of that unified policy can be proven through statistical evidence, such unit-level differences do not defeat the prima facie discrimination case. The case remains subject to collective proof. For example, in Ellis v. Costco Wholesale Corp. , 285 F.R.D. 492 (N.D. Cal. 2012), Costco opposed certification in a case related to gender discrimination in promotion. at 522–23. Costco’s expert argued that the court should look at “promotion numbers by region and evaluate each region separately,” noting that some regions showed no evidence of discrimination. Id. The court found the region- by-region data unpersuasive for several reasons, including that Costco’s top-down “promotion practices support a nationwide statistical analysis.” Id. Here, as well, Mobley has identified a unified policy that is the source of the alleged discriminatory disparate impact. At this stage, the fact that the policy may not yield discriminatory results in every possible scenario does not mean that the proposed collective is not similarly situated.
Furthermore, if Workday intends to argue that any observable disparate impact is the result of employer-driven bias and not attributable to Workday, that issue does not defeat preliminary certification either. How Workday’s AI recommendation system works, and the extent to which any disparate impact is generally driven by employer preferences rather than Workday’s own algorithms, are issues susceptible to common proof. To the extent Workday contends that its advertised AI recommendations are mere rote implementation of employer preferences, and thus that no uniform policy can be traced to its AI, that is an issue for the second stage of the “similarly situated” analysis, and not this preliminary stage. See Rabin , 2018 WL 3585143, at *4 (“[c]onsiderations involving the merits of claims are more appropriately addressed at the second stage of the analysis when fewer facts are in dispute”) (citation omitted); see also Stockwell , 749 F.3d at 1114 n. 3 (in the more demanding Rule 23(b) context, affirmative defenses are “not pertinent to the commonality question, as long as there is a common question as to the [] prima facie case of disparate impact age discrimination”).
3. Whether the Individual Characteristics of Members of the Proposed Collective Make Certification Impossible Workday argues that because Mobley and the Opt-In Plaintiffs (i) did not meet the minimum qualifications for every job they were rejected from, (ii) sometimes omitted indicia of their age from their applications, and (iii) received some interviews or job offers as the result of their applications via Workday, they are not similarly situated to the hypothetical plaintiff who could state a disparate impact claim against Workday. (Dkt. No. 107 at 13–15, 21–24.) Workday sees Mobley and Opt-in Plaintiffs’ slightly varied application experiences as “emblematic” of the fact that the individual characteristics of potential collective members will overwhelm, making “collective adjudication [] impossible .” ( at 19, 24 (emphasis in original).) At oral argument, Workday raised further concerns about whether Mobley would be able to carry his burden at the merits stage of statistically proving a disparate impact claim where the collective comprised both qualified and unqualified candidates for jobs.
Mobley is not required to prove that each member of the proposed collective is identically situated, or even that common questions predominate. Villa , 2012 WL 5503550, at *14 (“conditional certification of a collective action does not require a showing that common claims predominate”). Instead, his burden is to identify legal or factual similarities that are material to the resolution of the case. Id. As discussed above, Mobley has met that burden. Mobley has substantially alleged that the factors of his prima facie disparate impact claim are susceptible to common proof.
It is possible that, at the merits stage, Mobley will face challenges in proving disparate impact for the reasons that Workday identifies. However, that challenge is one that is common to the collective, and does not counsel against certification. See Tyson Foods, Inc. v. Bouaphakeo , 577 U.S. 442, 457 (2016) (“[r]espondents’ ‘failure of proof on this common question’ likely would have ended ‘the litigation and thus would not have caused individual questions to overwhelm questions common to the class.’”) (quoting Amgen Inc. v. Conn. Ret. Plans and Tr. Funds , 568 U.S. 455, 468 (2013)) (cleaned up). The variables may also be relevant to Workday’s defenses in individual cases, and will likely impact individual plaintiffs’ claim for damages. However, the “various defenses available to the defendant which appear to be individual to each plaintiff” is properly assessed at the second phase of the collective certification inquiry, and is not relevant to the preliminary certification analysis. See Rabin , 2018 WL 3585143, at *4. In sum, Workday has not identified a basis to deny preliminary certification.
D. Limits on the Collective
Workday argues that the only individuals who could conceivably be entitled to join a collective on Mobley’s theory of the case are those who applied to a high number of jobs, met the minimum qualification for every job for which they applied, and had a “zero percent success rate.” (Dkt. No. 106 at 12, 14, 20.) Workday claims that no other applicants are “entitled to the same ‘inference of discrimination’ the Court gave [Mobley] at the motion to dismiss.” ( at 20–21 . ) In support, Workday argues that the FLSA was amended to “limit[] private [ADEA] plaintiffs to employees who asserted claims in their own right and free[] employers of the burden of representative actions.” (Dkt. No. 160 at 21 (quoting Hoffmann-La Roche , 493 U.S. at 173) (emphasis added by Workday).)
This case, however, is no longer at the motion to dismiss stage and has moved into discovery. As such, whether each member of the collective could allege disparate impact without the benefit of discovery is irrelevant. Instead, the question is whether the proposed collective members are similarly situated in the sense that they were actually “subject to the particular employment practice with the alleged disparate impact.” See Pottenger v. Potlatch Corp. , 329 F.3d 740, 749–750 (9th Cir. 2003); see also Campbell , 903 F.3d at 1100 (individuals may only “join a collective action if they [] claim a violation of the FLSA , [] are ‘similarly situated’ to the original plaintiff, [] and affirmatively opt in”) (emphasis added). Because the proposed collective encompasses only individuals who submitted applications through Workday and were “denied employment recommendations,” this requirement is satisfied.
It is unnecessary to examine each individual’s qualifications, application volume, or rejection rate to determine if they are members of the collective. Whether Workday’s AI recommendation system has a disparate impact on applicants over forty is a question that is addressed across the collective, not on a member-by-member basis. Workday cites Heath for the proposition the collective cannot include “individuals who were not qualified for the jobs to which they applied.” (Dkt. No 107 at 20–23 (citing Heath , 215 F. Supp. 3d at 857).) But Heath was a disparate treatment case. Heath , 215 F. Supp. 3d at 848. Therefore, unlike in Heath , “qualification” is not an element of Mobley’s prima facie disparate impact claim. Compare id. at 857 (citing Reeves v. Sanderson Plumbing Prods., Inc. , 530 U.S. 133, 142 (2000)) with Bolden- Hardge , 63 F.4th at 1227.
Moreover, applicants do not need to show that they were qualified, got rejected, or would likely have succeeded in being hired for the job in order to have standing to bring a disparate impact claim for injunctive relief. [4] See Carney v. Adams , 592 U.S. 53, 66 (2020) (reaffirming that an “aggrieved party need not allege that he would have obtained the benefit but for the unlawful barrier in order to establish standing”) (quotations omitted). Here, in addition to damages, Plaintiffs seek injunctive relief to stop Workday from continuing to engage in the allegedly discriminatory practice. (FAC at 35.) To allege an injury-in-fact sufficient to satisfy Article III standing to seek injunctive relief, applicants need only show that they were denied the right to compete on equal footing against other candidates due to the allegedly discriminatory nature of Workday’s AI recommendation system, and that they intend to apply for jobs in the future through Workday’s platform.
“In the context of an employment discrimination claim, a plaintiff may claim an injury in fact from the purported denial of the ability to compete on an equal footing against other candidates for a job.” Shea v. Kerry, 796 F.3d 42, 50 (D.C. Cir. 2015) (citing Ne. Fla. Chapter of the Associated Gen. Contractors of Am. v. City of Jacksonville, 508 U.S. 656, 666 (1993)). “And because the injury lies in the denial of an equal opportunity to compete, not the denial of the job itself, the Court need not inquire into the plaintiff’s qualifications (or lack thereof) when assessing standing.” Id. (citing Regents of Univ. of Cal. v. Bakke, 438 U.S. 265, 280–81 & n. 14 (1978)) (emphasis in original); Cerrato v. S.F. Cmty. Coll. Dist. , 26 F.3d 968, 976 (9th Cir. 1994) (“[t]o establish standing, a person need only allege that a discriminatory policy exists that prevents him from competing on an ‘equal basis’”). [5] Individuals over forty who were allegedly subjected to the discriminatory policy— i.e. , they had an application scored, sorted, ranked, or screened by Workday’s AI—therefore have standing regardless of their qualifications, application volume, or rejection rate and may be included in the collective.
To be sure, an applicant’s qualifications could affect their standing to seek damages. See Braunstein v. Ariz. Dep’t Transp. , 683 F.3d 1177, 1186–87 (9th Cir. 2012) (plaintiff lacked Article III standing to pursue a damages claim in an equal protection case because he had not submitted bids or shown that he would have been “in a position to compete equally” absent the alleged discriminatory policy). However, there is no basis for precluding members who have standing to seek injunctive relief from joining the collective simply because they may not be able to state a claim for money damages. At the trial stage, the collective action could proceed first on a representative or group basis as to liability and injunctive relief, before moving into a damages phase on an individual basis. See 7B Fed. Prac. & Proc. Civ. § 1807 (3d ed. 2025). At this stage, it is premature to limit the collective in the way that Workday suggests. [6]
In summary, preliminary certification is granted to the collective comprising “[a]ll individuals aged 40 and over who, from September 24, 2020, through the present, [] applied for job opportunities using Workday, Inc.’s job application platform and were denied employment recommendations.” In this context, being “denied” an “employment recommendation” means that (i) the individual’s application was scored, sorted, ranked, or screened by Workday’s AI; (ii) the result of the AI scoring, sorting, ranking, or screening was not a recommendation to hire; [7] and (iii) that result was communicated to the prospective employer, or the result was an automatic rejection by Workday.
E. Feasibility of Providing Notice
Workday raises a host of concerns regarding how the parties can identify and notify collective members. In addition to its line-drawing concerns already discussed above, Workday argues that it is contractually barred from accessing data about applications, and even if it could access that data, there is no way to determine the age of applicants. Finally, Workday makes the policy argument that the proposed collective is simply too big, and therefore any notice would amount to an improper solicitation of claims.
On the existing record, it appears that Workday will likely be able to access the necessary data to provide targeted notice, and that it would not be unduly burdensome for Workday to gather and provide said data. First, Workday acknowledged at oral argument that it is able to see whether its customers have enabled any AI features. Second, Workday does not dispute that applicant data is hosted by Workday, and that while Workday’s customers sign contracts that prohibit Workday from accessing applicants’ confidential personal data, the contract contains an exception applicable when Workday is subject to a court order. Third, Workday confirmed that, assuming customer consent, it can identify the score that its CSM feature gave an applicant (and can presumably do so for other features as well). Finally, although Workday’s counsel was unable to confirm what age-correlated data Workday customers collect, it is a fair inference that datapoints such as graduation year or employment experience will frequently be available to use as a proxy for age. See Heath , 215 F. Supp. 3d at 859 (ordering the parties to “meet and confer on how to use applicants’ graduation date information as a proxy for age, where available in Google’s gHire database, to identify the candidates to whom the third-party administrator will send notice.”).
The estimated size of the collective does not provide a basis to withhold notice either. Workday states that “1.1 billion applications were rejected using Workday” during the time period at issue and “any notice would still invite potentially hundreds of millions of potential plaintiffs to file their claims in this case.” (Dkt. No. 107 at 24–25 (emphasis in original).) First, this rough estimate ignores the qualifiers in the definition of the collective that will limit its scope. Second, it is not the size of the proposed collective (as Workday argues), but rather the content of the proposed notice, that the Supreme Court instructs courts to police in order to “respect judicial neutrality.” Hoffmann-La Roche Inc , 493 U.S. at 174. In fact, Hoffmann-La Roche reinforces the importance of providing “accurate and timely notice concerning the pendency of the collective action, so that [individuals] can make informed decisions about whether to participate.” at 170. There is no basis for withholding notice simply because Workday is alleged to have discriminated against a large number of applicants.
IV. CONCLUSION
Because the proposed collective is similarly situated, Mobley’s Motion for Conditional Certification of Collective Action (Dkt. No. 106) is GRANTED , and notice should issue to the collective members. Although disputes remain—regarding, for example, the parameters of the “court order exception” in Workday’s customer contracts, the availability of age-related proxies, and how to define a “recommendation”—it is premature to find that any of these issues make targeted notice unfeasible, and none of the identified issues support the withholding of notice altogether. See Rabin v. PricewaterhouseCoopers LLP , No. 16-cv-02276-JST, 2019 WL 9078785, at *4 (N.D. Cal., Apr. 5, 2019) (certifying a collective action but explaining that “[h]aving so ruled, the Court cannot yet order notice to potential members of the collective action because—as the parties themselves agree—further meet-and-confer efforts are required before they and the Court can determine who will receive such notice”).
The parties are directed to meet and confer and engage in further targeted discovery to resolve the remaining disputes. The parties are encouraged to work cooperatively and transparently in identifying the most sensible manner for providing comprehensive notice. In the event that a list of collective members is developed based on incomplete information in discovery, and later discovery reveals that a larger group should have been noticed, the Court may consider whether to authorize further round(s) of notice and which party should be responsible for the costs of that notice. The parties should also consider whether alternative notice is necessary if more targeted notice proves too difficult. Carrillo v. Schneider Logistics , Inc., No. 11-cv-08557, 2012 WL 556309, at *13 (C.D. Cal. Jan. 31, 2012), aff’d , 501 F. App’x 713 (9th Cir. 2012) (collecting cases allowing for alternate notice distribution methods where FLSA workers could not be easily identified).
In addition, Workday identifies a variety of legitimate concerns about the content of the proposed notice, about which the parties are directed to meet and confer. The parties shall file by May 28, 2025 , a case management statement containing a proposed schedule for the targeted discovery, submission of the notice plan and revised notice, and briefing and a hearing on aspects of the notice plan on which the parties cannot reach agreement, if any. If the parties are unable to reach agreement on the proposed schedule, the joint filing shall include each side’s proposal as to that issue and the reasoning behind its requested approach. A case management conference to discuss this schedule is set for June 4, 2025, at 10:00 a.m. via videoconference. IT IS SO ORDERED.
Dated: May 16, 2025
RITA F. LIN United States District Judge
[1] Citations to page numbers refer to the ECF pagination.
[2] “The fact that a defendant submits competing declarations will not as a general rule preclude [preliminary] certification.” Harris v. Vector Mktg. Corp. , 716 F. Supp. 2d 835, 838 (N.D. Cal. 2010); see also Sanchez v. Sephora USA, Inc. , No. 11-cv-03396-SBA, 2012 WL 2945753, at *4 (N.D. Cal. July 18, 2012) (collecting cases that disregarded disputed evidence submitted by defendants when certifying a collective at the notice stage).
[3] Quoting https://www.workday.com/en-us/products/talent-management/talent-acquisition.html.
[4] Workday conceded at oral argument that it was not challenging the proposed collective members’ standing at this stage, but this order addresses the issue to avoid any doubt as to the propriety of defining the collective in the proposed fashion.
[5] Workday contended at oral argument that this line of cases is distinguishable because they involved employment tests that were alleged to discriminate under Title VII rather than the ADEA. But Workday offers no principled reason why the Article III standing analysis should treat injury-in-fact due to age discrimination differently from injury-in-fact due to race or sex discrimination.
[6] It is theoretically possible that the preliminary collective will include some individuals who, notwithstanding that they applied to jobs using Workday in the last four years, do not intend to apply to any jobs using Workday’s software in the future. This is likely to be a small percentage of the total collective, given Workday’s widespread role in the job market. ( See Dkt. No. 108 ¶¶ 6-7.) Furthermore, many such individuals will still have standing under a damages theory. Finally, while there may be a small subset of individuals in the preliminary collective who are not in either camp, this fact does not render the case unmanageable at this stage. See Hoffmann- La Roche Inc ., 493 U.S. at 174 (the court’s “intervention in the notice process [is] for case management purposes”).
[7] The parties should meet and confer to determine what rank, score, or grouping constitutes a “recommendation.”