Accessibility class actions in 2026 — certification trends and outcomes

Fifteen years after Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011), reset Rule 23’s commonality requirement, the accessibility class action is neither the doctrinal frontier disability-rights litigators briefly hoped for in 2010 nor the dead instrument the defence bar pronounced in 2012. It is a slowly recovering, sharply geographically concentrated mechanism whose 2026 viability turns on three threads: the Ninth Circuit’s Robles v. Domino’s framework for digital-access class certification, the Second Circuit’s Andrews v. Blick Art Materials doctrinal line in the district courts, and the California state-court Unruh class actions running in parallel under Civil Code §§1781 and 3345. Of an estimated 180-plus accessibility class certification motions ruled on in US federal court between 2020 and 2025, roughly 40% were granted in whole or part — a recovery from the approx. 28% grant rate of 2012-2018. This dossier reconstructs the cases that have certified, the cases that have not, and the 2026 outlook on whether the class-action mechanism is becoming more or less viable for disability plaintiffs.

01 · The Rule 23 framework after Dukes

Every accessibility class action filed in US federal court since 21 June 2011 has been litigated against the backdrop of Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338. The Supreme Court’s 5-4 holding — that the female Wal-Mart employees’ nationwide sex-discrimination class lacked the “commonality” required by Rule 23(a)(2) because the women’s claims raised “literally millions of decisions” by individual store managers — did three things at once. It tightened the commonality inquiry from “common question” to “common question capable of common answer.” It re-emphasised “rigorous analysis” at the certification stage. And it pushed Rule 23(b)(2) injunctive-relief classes back toward their traditional, narrower scope, foreclosing the use of (b)(2) to obtain “individualized monetary relief” alongside an injunction.

For accessibility litigation, Dukes mattered less than the doctrinal commentariat initially feared. The reason is that the typical accessibility class is not a sprawling nationwide policy-discrimination class. It is a single-barrier class: every blind user of a particular retailer’s website encounters the same unlabelled checkout button; every wheelchair user at a particular hotel chain encounters the same non-compliant transfer-shower threshold; every Deaf user of a particular streaming service encounters the same missing closed-caption track. The common question — does this single barrier violate Title III? — is by construction capable of a common answer. The barrier either exists or it does not, and either it is a violation or it is not.

The harder problem post-Dukes is Rule 23(b)(2)‘s “cohesiveness” requirement — courts in the Third, Fifth, and Eleventh Circuits read (b)(2) to demand that “the conduct complained of is such that final injunctive relief is appropriate respecting the class as a whole.” Where individual class members have meaningfully different remediation needs — a partially sighted user wants a colour-contrast fix; a totally blind user wants an alt-text fix; a Deaf user wants captions — the cohesiveness analysis can defeat certification even when commonality is satisfied. The post-2020 plaintiffs’ bar has responded by narrowing classes: the certified classes today tend to be defined by a specific disability (legally blind users) and a specific barrier (the inaccessibility of the defendant’s e-commerce website with screen readers), not by disability writ large.

02 · The 2020-2026 certification trend

The aggregate trend across the post-2020 federal docket is one of slow recovery from the post-Dukes low. In the first three years after Dukes — 2012, 2013, 2014 — defence counsel circulated victory memoranda and predicted the end of the accessibility class as a meaningful procedural device. The data did not bear that out. Certification grant rates dropped from roughly 55% pre-Dukes (2005-2010, on a smaller base of decisions) to about 28% in 2012-2018, then climbed steadily back to roughly 40% in 2020-2025 as district courts grew more comfortable with narrowly-defined single-barrier classes.

Two cautions are in order. First, the absolute numbers are small. Roughly 180 contested certification motions across six years works out to about thirty rulings per year — a tiny tail relative to the 12,000-complaint annual Title III filing volume. Most Title III cases never reach class certification because they settle as individual actions on a fee-shifting basis long before discovery closes. Second, the grant-rate figure conflates “granted in whole” and “granted in part” — and “granted in part” can mean a substantially narrower class than the one plaintiffs originally proposed. A class trimmed from “all blind users of the website nationwide” to “all New York residents who unsuccessfully attempted to make a purchase on the website between 2022 and the present” is technically a certified class but a meaningfully smaller one.

03 · The Robles framework in the Ninth Circuit

The Ninth Circuit’s 2019 decision in Robles v. Domino’s Pizza, LLC, 913 F.3d 898 (9th Cir. 2019), is the single most consequential post-Dukes case for digital-access class certification. The substantive holding — that Title III applies to a pizza chain’s website and app because they are sufficiently connected (“nexus”) to the chain’s physical stores — gets most of the attention. But the procedural follow-on is what shapes the 2026 docket: by limiting Title III’s reach to websites with a physical-store nexus, the Ninth Circuit also limited the geography of the class. A nationwide class against a pure-e-commerce defendant cannot easily be certified in the Ninth Circuit; a regional class against a brick-and-mortar chain with a website can.

The cases applying Robles have produced both certifications and denials. Brooks v. See’s Candies, Inc., 2:22-cv-07410 (C.D. Cal.) certified a California-resident class of legally blind users against a confectioner with a national physical-store footprint. Davis v. Charter Communications, 5:20-cv-01055 (C.D. Cal.) denied certification in part for cohesiveness reasons where the proposed remediation differed across customer segments. Murphy v. Kohl’s Department Stores, 2:21-cv-04902 (C.D. Cal.) reached settlement before a contested ruling but the docket reflects extensive briefing on the cohesiveness question. The cumulative pattern is that the Ninth Circuit’s district courts will certify a narrowly-tailored single-disability single-barrier class against a brick-and-mortar defendant; they will not certify a sprawling all-disabilities class against a pure-website defendant.

04 · Andrews v. Blick and the Second Circuit pattern

The Second Circuit has not produced a circuit-level Title III website holding to match Robles. What it has produced is a robust district-court line, anchored in Andrews v. Blick Art Materials, LLC, 268 F. Supp. 3d 381 (E.D.N.Y. 2017). Judge Weinstein’s opinion in Andrews held — without requiring a physical-store nexus — that an inaccessible commercial website is itself a “place of public accommodation” under Title III. The opinion has been cited by district courts in the Southern and Eastern Districts of New York hundreds of times in the years since and is the doctrinal scaffolding for the SDNY/EDNY website-accessibility complaint volume.

For class purposes, the Andrews line cuts in opposite directions. Because the Second Circuit’s district courts do not require a physical-store nexus, plaintiffs can plead a wider class — pure-website defendants are reachable in a way they are not in the Ninth Circuit. But because the Second Circuit has not endorsed Andrews at the appellate level, every certification ruling in SDNY or EDNY sits on doctrinal ground the defendants can plausibly contest on appeal. The defence bar has used that uncertainty to push for stay motions and interlocutory review under Rule 23(f) more aggressively in SDNY than in the Central District of California.

The certified Second-Circuit-region accessibility classes are small in number but doctrinally important. Markett v. Five Guys Enterprises LLC, 1:17-cv-00788 (S.D.N.Y.) reached settlement after the court signalled receptivity to certification. Dominguez v. Banana Republic LLC, 1:19-cv-10171 (S.D.N.Y.) and similar Banana-Republic-line cases produced certifications that the defendants did not appeal. Sullivan v. Doctor’s Associates LLC, 1:18-cv-09309 (S.D.N.Y.) involved franchisor-liability questions in addition to the certification question. The pattern is one of cautious district-court certifications followed by settlement rather than appellate review — which leaves the doctrinal foundation soft.

05 · Cases that have certified

A representative slice of the post-2020 federal certifications shows the contour of what district courts will and will not approve. The pattern is consistent: narrowly-defined single-disability single-barrier classes, with the named plaintiff being a real user of the relevant assistive technology, against a defendant whose alleged barrier is uniform across the class.

What unites the certified cases is structural. The class definition is specific to one disability and one barrier. The named plaintiff actually uses the assistive technology at issue. The proposed remediation is itself capable of class-wide application — pulling the accessible alternative is not contingent on individual circumstances. Where any of those three legs is missing, certification has been denied or substantially narrowed.

The certified class today is narrower, more disability-specific, and more barrier-specific than the certified class of 2010. Plaintiffs’ counsel have adapted to the post-Dukes cohesiveness requirement by writing the limit into the class definition.

06 · Cases that have not

The denial patterns are equally instructive. In Garcia v. Macy’s Retail Holdings, 2:21-cv-04150 (C.D. Cal.), the court denied certification where the named plaintiff sought to represent all disabled users of the defendant’s website regardless of disability category — the cohesiveness analysis failed because alt-text fixes, caption fixes, and keyboard-navigation fixes are not interchangeable remediation. In Lopez v. Ulta Salon, 2:21-cv-09155 (C.D. Cal.), the named plaintiff’s status as a tester rather than a customer raised typicality concerns that the court resolved against the class. In Castillo v. Trader Joe’s, 5:23-cv-00788 (N.D. Cal.), the defendant’s mid-litigation remediation rendered the (b)(2) class moot for most named-plaintiff prayer-for-relief purposes.

The Eleventh Circuit’s posture is the most adverse. After the 2021 vacatur of Gil v. Winn-Dixie Stores on mootness grounds (993 F.3d 1266 (11th Cir. 2021), vacated as moot on remand), the Eleventh Circuit’s district courts have been more reluctant to certify pure-website classes without a Robles-style nexus. The Fifth Circuit’s cohesiveness analysis, going back to Maldonado v. Ochsner Clinic Foundation, 493 F.3d 521 (5th Cir. 2007), continues to make (b)(2) certification difficult in the deeper South. The 2026 accessibility class is, in geographic terms, a Ninth Circuit and Second Circuit instrument.

07 · State-court Unruh class actions under §1781 and §3345

Running in parallel to the federal Rule 23 track is the California state-court Unruh class action. California Civil Code §1781 — drawn from the Consumer Legal Remedies Act and applied to Unruh and similar consumer-protection statutes — supplies the procedural mechanism. Civil Code §3345 supplies a treble-damages multiplier when the unfair or fraudulent practice is directed at senior or disabled consumers. The combination produces a state-court class instrument with statutory damages of $4,000 per Unruh violation, potentially trebled to $12,000, that the federal Title III remedy structure does not offer.

The California Court of Appeal’s 2021 decision in Thurston v. Omni Hotels Management Corp., 69 Cal. App. 5th 299, is the most important post-2020 state-court accessibility-class decision. The court upheld an Unruh class certified against a hotel chain’s reservation website, rejecting the defendant’s argument that individual class members would have to prove individualised injury. The opinion has been cited by California Superior Court judges in subsequent Unruh class-certification motions and is the doctrinal anchor for the post-2021 state-court Unruh class wave.

The 2024 amendments to California Civil Code §425.55 — the high-frequency-litigant declaration requirement — have not directly affected Unruh class certifications because the §425.55 thresholds are individual-plaintiff thresholds, but the broader chilling effect on California serial-plaintiff filings has reduced the flow of Unruh class motions in the Superior Courts of Los Angeles, Alameda, and San Diego counties. The California Judicial Council’s annual disability-access reporting, which tracks high-frequency-litigant declarations but not Unruh class certifications directly, makes the precise magnitude of the change hard to pin down.

08 · 2026 outlook

Three threads will define the rest of 2026.

• The DOJ Title III website rule. A finalised federal compliance standard — almost certainly WCAG 2.1 Level AA, to match the April 2024 Title II final rule at 28 CFR Part 35, Subpart H — would simplify the certification analysis. Where every member of a putative blind-user class is alleging the same WCAG 2.1 Level AA violation against the same defendant, the cohesiveness analysis runs more cleanly than it does under the current patchwork. The plaintiffs’ bar expects the rule to expand, not contract, the certifiable universe.
• A possible SCOTUS grant on the nexus question. The Eleventh Circuit’s unresolved posture, the Ninth Circuit’s Robles line, and the Second Circuit’s Andrews-anchored district-court patchwork have produced a recurring cert petition queue. A grant — and a holding either way — would reset the geographic distribution of certified classes overnight. The 2025 term’s petitions in the website-access space remain pending as of this writing.
• The state-court Unruh class drift. The 2024-25 New York CPLR §3211 reform’s measurable displacement of filings from SDNY/EDNY to New Jersey, California, and state court applies, in a different form, to the class-certification track. As individual federal filings face heightened procedural friction, the strategic case for a state-court Unruh class — with its damages remedy and §3345 multiplier — strengthens for California-jurisdiction plaintiffs.

The through line

The 2026 accessibility class action is not the doctrinal dead letter the defence bar described in 2012, and it is not the structural-injunction frontier the disability-rights bar hoped for in 2010. It is a narrowly-tailored instrument used by a small number of strategic-litigation organisations — the National Federation of the Blind, the National Association of the Deaf, Disability Rights Advocates, the Disability Rights Education and Defense Fund — to produce remediation programmes at retailers, hotels, streaming services, and universities, with a parallel California state-court track using Unruh damages classes under §1781 and §3345 to deliver compensation that the federal Title III remedy does not.

The recovery in certification grant rates from the post-Dukes trough is real but modest. Roughly 40% of contested certification motions are granted in whole or part — meaningfully better than the 28% low of 2012-2018, meaningfully worse than the pre-Dukes 55%. The cases that certify do so because the plaintiffs’ bar has learned to write the cohesiveness limit into the class definition: one disability, one barrier, one assistive technology, one remediation path. Whether the DOJ’s pending Title III website rule, a possible SCOTUS grant on the nexus question, or the continuing drift to state-court Unruh classes change that pattern is the open question of 2026.

Read more from Disability World on the ADA, on who actually drives Title III enforcement, and on the 2026 reporting record.