Dennis Speerly v. General Motors, LLC

RECOMMENDED FOR PUBLICATION

Pursuant to Sixth Circuit I.O.P. 32.1(b)

File Name: 25a0170p.06

UNITED STATES COURT OF APPEALS

FOR THE SIXTH CIRCUIT

┐

DENNIS SPEERLY; JOSEPH SIERCHIO; DARRIN

│

DEGRAND; DANIEL DRAIN; WAVERS SMITH; RICHARD

│

FREEMAN; CHRISTOPHER GILES; LOUIS RAY; RICHARD

│

SULLIVAN; JAMES NORVELL; MICHAEL BANKS; GUY

│

CLARK; MARIA BARALLARDOS; CARY SHERROW; >

JASON KEVIN SINCLAIR; KIMBERLY COULSON; TROY │ No. 23-1940

COULSON; ANDRE MCQUADE; DONALD DYKSHORN; │

TAIT THOMAS; JAMES PAUL BROWNE; WILLIAM │

FREDO; JON ELLARD; RHIANNA MEYERS; RANDALL │

JACOBS; MICHAEL PONDER; PHILIP WEEKS; KATRINA │

FREDO; JIMMY FLOWERS; STEVEN BRACK; KEVIN │

WESLEY; BRIAN LLOYD; GREGORY BUTSCHA; JERRY │

CARROLL; KIMBERLY CARROLL; DOMINIC EATHERTON; │

THOMAS EDMONDSON; RICHARD FILIAGGI; ROBERT │

HIGGINS; DAVID THOMPSON; DONALD SICURA, │

Plaintiffs-Appellees, │

│

│

v. │

│

GENERAL MOTORS, LLC, │

Defendant-Appellant. │

┘

On Petition for Rehearing En Banc.

United States District Court for the Eastern District of Michigan at Detroit.

Nos. 2:19-cv-11044; 2:19-cv-11802; 2:19-cv-11808; 2:19-cv-11875; 2:19-cv-12371—David M.

Lawson, District Judge.

Argued En Banc: March 19, 2025

Decided and Filed: June 27, 2025

Before: SUTTON, Chief Judge; MOORE, COLE, CLAY, GRIFFIN,

KETHLEDGE, STRANCH, THAPAR, BUSH, LARSEN, NALBANDIAN,

READLER, MURPHY, MATHIS, BLOOMEKATZ, and RITZ, Circuit Judges.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 2

_________________

COUNSEL

ARGUED EN BANC: Richard C. Godfrey, QUINN, EMANUEL, URQUHART &

SULLIVAN, LLP, Chicago, Illinois, for Appellant. Douglas J. McNamara, COHEN MILSTEIN

SELLERS & TOLL, PLLC, Washington, D.C., for Appellees. ON SUPPLEMENTAL BRIEF:

Richard C. Godfrey, R. Allan Pixton, QUINN, EMANUEL, URQUHART & SULLIVAN, LLP,

Chicago, Illinois, John F. Bash, QUINN, EMANUEL, URQUHART & SULLIVAN, LLP,

Austin, Texas, Stephanie A. Douglas, BUSH SEYFERTH PLLC, Troy, Michigan, Renee D.

Smith, Cole T. Carter, KIRKLAND & ELLIS LLP, Chicago, Illinois, Jason M. Wilcox,

KIRKLAND & ELLIS LLP, Washington, D.C., for Appellant. Douglas J. McNamara, Karina

Puttieva Madelyn Petersen, COHEN MILSTEIN SELLERS & TOLL, PLLC, Washington, D.C.,

Theodore J. Leopold, COHEN MILSTEIN SELLERS & TOLL PLLC, Palm Beach Gardens,

Florida, for Appellees. ON AMICI BRIEFS: John M. Thomas, DYKEMA GOSSETT PLLC,

Ann Arbor, Michigan, Kyle M. Asher, DYKEMA GOSSETT PLLC, Lansing, Michigan, Aaron

D. Van Oort, John L. Rockenbach, FAEGRE DRINKER BIDDLE & REATH LLP,

Minneapolis, Minnesota, Brian D. Schmalzbach, MCGUIRE WOODS LLP, Richmond,

Virginia, Philip S. Goldberg, SHOOK, HARDY & BACON L.L.P., Washington, D.C., Stephen

A. D’Aunoy, KLEIN THOMAS LEE & FRESARD, St. Louis, Missouri, Brandon L. Boxler,

KLEIN THOMAS LEE & FRESARD, Richmond, Virginia, Wendy Liu, PUBLIC CITIZEN

LITIGATION GROUP, Washington, D.C., William Alvarado Rivera, AARP FOUNDATION,

Washington, D.C., David J. Shea, SHEA LAW, PLLC, Southfield, Michigan, Adam J. Levitt,

John E. Tangren, DICELLO LEVITT LLP, Chicago, Illinois, Jason L. Lichtman, LIEFF

CABRASER HEIMANN & BERNSTEIN, LLP, New York, New York, for Amici Curiae.

SUTTON, C.J., delivered the opinion of the court in which GRIFFIN, KETHLEDGE,

THAPAR, BUSH, LARSEN, NALBANDIAN, READLER, and MURPHY, JJ., concurred.

THAPAR, J. (pp. 38–52), delivered a separate concurring opinion in which KETHLEDGE and

MURPHY, JJ., concurred. NALBANDIAN, J. (pp. 53–65), delivered a separate concurring

opinion in which GRIFFIN, J., concurred. MOORE, J. (pp. 66–112), delivered a separate

dissenting opinion in which COLE, CLAY, STRANCH, MATHIS, BLOOMEKATZ, and RITZ,

JJ., concurred.

_________________

OPINION

_________________

SUTTON, Chief Judge. With the introduction of automatic transmissions in cars, clutch

pedals became obsolete for drivers who do not enjoy using a stick shift. General Motors’ Hydra-

Matic became the first readily accessible automatic transmission on the market in 1939. It had

four gears. From 2015 to 2018, GM offered an eight-gear Hydra-Matic transmission.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 3

That version had two alleged problems. It led some GM cars occasionally to shudder, and it led

some GM cars occasionally to lurch. The two problems had nothing to do with operator error,

even if some of the symptoms called to mind the experience of getting a ride from a novice stick-

shift driver.

A group of car buyers filed a class action against GM to answer for the two defects under

a range of state common law and statutory theories: express warranty, implied warranty,

consumer protection laws, and fraudulent omission. The district court certified 26 state-wide

subclasses with a total of 59 state-law claims on behalf of roughly 800,000 individual car buyers.

Because the subclasses do not meet the rigorous requirements for handling all of these cases in

one district court, whether as one case or as 26 cases, we vacate the class-certification order.

I.

A transmission translates power into movement. The engine ignites gasoline, converting

the chemical energy stored in fuel into the mechanical energy of a rotating crankshaft. The

transmission prompts a range of gears to control how much of the crankshaft’s torque reaches the

wheels, all by selecting an appropriate gear depending on the car’s speed. When used

effectively, the gears use the engine’s torque efficiently and protect the crankshaft from wearing

down.

In 2015, GM released the “Hydra-Matic 8-Speed Transmission” as an upgrade to its prior

models. R.220-3 at 138. Compared to their 6-speed predecessors, the “8L” transmissions

offered “8 gear ratios,” which created two more refinements when it comes to engaging the most

efficient gear “for performance, fuel economy and general drivability.” R.224-17 at 13, 15. It

did so, better yet, while taking up the same space and less weight than the six-speed

transmissions.

As GM tells it, the 8L’s hardware brought together “unique algorithms” that delivered

“world-class shift-response times” with “performance that rival[ed] the” “transmissions found in

many supercars—but with the smoothness and refinement that comes with a conventional

automatic” transmission. R.41-5 at 3. As the class-action representatives tell it, GM knew about

problems with the 8L transmission before the first car rolled off the lot.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 4

The class identifies two problems with the transmission. The first turns on the 8L’s

transmission fluid, a lubricant that protects the gears and valves as the transmission operates. As

customers drive, especially in humid climates, the fluid absorbs moisture “introduced via the

vent system.” R.177-2 at 18–19, 63–64. The moisture changes the fluid’s viscosity, which is

calibrated to prevent the gears from slipping. When the gears lose traction and slip, the

transmission vibrates. Some customers, especially those driving in high gears, feel a

“shake/shudder feeling” akin to “driving over rumble strips or rough pavement.” R.177-3 at 2.

A GM engineer noted as early as 2013 that the “shudder is terrible” on test cars. R.224-9 at 2.

The problem led to a rash of warranty claims. After testing an “Option B” fluid that “did

not address the sensitivity to water,” R.245-5 at 6, GM developed a “Mod1a” moisture-resistant

fluid in December 2018, R.177-3 at 9–10. The company did not conduct a “field action”—

similar to a recall—to notify customers that they could receive the new fluid. R.221-1 at 266.

But it instructed dealers to resolve shudder complaints by flushing customer transmissions with

the new fluid. Most of the customers who received the replacement never complained about the

shudder problem again.

The second problem is “unrelated” and concerns how the transmission shifts gears.

R.220-3 at 91. What makes a transmission automatic is its ability to monitor a vehicle’s speed

and decide on its own which gear to apply. The 8L is no exception. Its “Transmission Control

Module” first analyzes input from sensors across the vehicle. R.220-3 at 140. When the module

software detects the need for a gear change, it instructs the hardware to manipulate the

“pressure” and “direction” of transmission fluid to create hydraulic pressure, forcing the gears to

change. R.220-3 at 140.

The 8L transmissions, as it happens, sometimes add too much pressure and sometimes

fail to “purg[e] trapped air” leaking into the valves. R.177-6 at 2. That design issue has a range

of impacts. Some customers feel nothing at all. Some customers who shift their vehicle gears

from park to drive in the mornings feel a “[h]esitation” for one to five minutes, as if they remain

in neutral, before they can drive their cars. R.177-6 at 2; R.225-3 at 10–11. Other customers

feel the vehicles “jerk forward” when the transmission moves from higher to lower gears, as if

they have been “rear-ended.” R.225-1 at 10, 15. The lunge forces customers to slam the brakes

No. 23-1940 Speerly et al. v. General Motors, LLC Page 5

to stop. Still other customers feel the same “violent[] jerk[]” but only when shifting the

transmission from lower to higher gears. R.225-7 at 10–11.

After customers complained about this problem, GM tried to fix it. The company sent its

dealerships 60 “Technical Service Bulletins”—what amount to warranty-covered hardware and

software fixes—to resolve the issue. R.220-3 at 262; R.177-6 at 23. While some of the

workarounds “[i]mproved drive quality,” R.206-15 at 5, the underlying issues “could not be

resolved without a major redesign of the transmission,” R.177-6 at 21. The company instructed

dealers to tell customers that harsh shifting was “[c]haracteristic” of GM cars, R.206-15 at 5,

while internally noting that “[r]eplacing transmission components or complete assemblies

w[ould] not improve” harsh shift conditions, R.177-6 at 40. A full redesign solved the problem

when GM released a second generation of eight-speed transmissions for the 2022 model-year

(and later) vehicles.

A group of customers who purchased the old vehicles filed this lawsuit on behalf of a

putative class. Seventy-four plaintiffs from 32 states brought 104 claims against GM for breach

of express warranty, breach of implied warranty, violation of state consumer protection statutes,

and fraudulent omission. The district court certified 26 statewide subclasses led by 33 named

plaintiffs who purchased selected GM cars from model year 2015 to model year 2019, all with

the 8L transmissions. The class, as certified, represents roughly 800,000 individuals. GM

appeals.

II.

We start by stopping to consider whether we have jurisdiction over this case. Article III

permits us to decide only “Cases” and “Controversies.” U.S. Const. art. III, § 2. That

“irreducible constitutional minimum” requires that the plaintiff have suffered an “injury in fact,”

“trace[able]” to the defendant’s actions, and “redress[able]” by a favorable decision. Lujan v.

Defs. of Wildlife, 504 U.S. 555, 560–61 (1992) (quotation omitted).

What gives pause is the first prong, injury in fact. It requires that a plaintiff suffer a

“concrete” invasion that is “real, and not abstract.” TransUnion LLC v. Ramirez, 594 U.S. 413,

424 (2021) (quotation omitted). “[T]raditional tangible harms” such as “monetary harms” count.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 6

Id. at 425. But the “mere risk of future harm, standing alone,” does not suffice for money-

damages actions. Id. at 436.

How does this test apply to a consumer who arguably overpays for a product because it

allegedly has a hidden defect? Is that a sufficiently tangible and concrete harm to cross the

Article III threshold? Most of our sister circuits say yes. If someone pays “more for [something]

than they would have, had they known of the risks,” these courts have concluded, they suffer a

concrete injury. In re Aqua Dots Prods. Liab. Litig., 654 F.3d 748, 750–51 (7th Cir. 2011); In re

Evenflo Co., Mktg., Sales Pracs. & Prods. Liab. Litig., 54 F.4th 28, 37–38 (1st Cir. 2022); Cole

v. Gen. Motors Corp., 484 F.3d 717, 722–23 (5th Cir. 2007); Nguyen v. Nissan N. Am., Inc., 932

F.3d 811, 815–16 (9th Cir. 2019); Debernardis v. IQ Formulations, LLC, 942 F.3d 1076, 1086–

87 (11th Cir. 2019). That company includes us, at least if we include an unpublished and non-

binding decision. Loreto v. Procter & Gamble Co., 515 F. App’x 576, 581 (6th Cir. 2013).

Two features of this case complicate the issue. First is the fact that some courts forbid, as

a matter of standing, “purchasers without manifest defects” from “piggyback[ing] on the injury

caused to those with manifest defects.” Johannessohn v. Polaris Indus. Inc., 9 F.4th 981, 988

(8th Cir. 2021). The Seventh Circuit likewise recently cabined overpayment-based injuries to

exclude products containing only a “potential risk of harm.” In re Recalled Abbott Infant

Formula Prods. Liab. Litig., 97 F.4th 525, 530 (7th Cir. 2024).

Second is the class-action posture of this case. The Supreme Court has not decided

whether an unnamed class member’s lack of standing poses an Article III problem. See Gratz v.

Bollinger, 539 U.S. 244, 262–63 (2003) (finding the matter unsettled); Sosna v. Iowa, 419 U.S.

393, 402–03 (1975) (suggesting that only the named plaintiffs need standing). No matter how

the Court resolves the point, all agree that the class representatives at a minimum must show an

injury. Sosna, 419 U.S. at 402–03; U.S. Parole Comm’n v. Geraghty, 445 U.S. 388, 404 (1980).

The parties and we agree that the class representatives in today’s case own cars that experienced

these defects. “[E]very named plaintiff,” the district court found, “testified that their vehicles

‘manifested’ one or both alleged defects, repeatedly and continuously, despite multiple

presentments and failed repairs.” R.284 at 46, 61–74. The named plaintiffs suffered an injury in

No. 23-1940 Speerly et al. v. General Motors, LLC Page 7

fact if their cars shuddered and shifted, even if their alleged overpayment for the cars may or

may not suffice.

The standing of absent class members becomes immaterial once a court, as here, vacates

a certification order. Because “class members are not parties before class certification,” “a court

need not worry about their standing until it certifies the class.” Fox v. Saginaw County, 67 F.4th

284, 296 (6th Cir. 2023). That is why a “class-certification denial will make a[n absent] class

member’s standing problem irrelevant.” Id. at 297. Unless and until the district court certifies

the class on remand, the absent class members’ standing makes no difference. We need to verify

only that the named plaintiffs, who are parties from the start, have standing to pursue their

claims. They do on this record.

III.

We review the court’s decision to certify a class for abuse of discretion. In re Whirlpool

Corp. Front-Loading Washer Prods. Liab. Litig., 722 F.3d 838, 850 (6th Cir. 2013). The court

must exercise that discretion “within the framework of Rule 23.” In re Am. Med. Sys., Inc., 75

F.3d 1069, 1079 (6th Cir. 1996). A district court abuses its discretion if it misstates the law,

relies on clearly erroneous facts, or makes a “clear error of judgment.” In re Ford Motor Co., 86

F.4th 723, 727 (6th Cir. 2023) (per curiam); In re Nissan N. Am., Inc. Litig., 122 F.4th 239, 245

(6th Cir. 2024).

Class certification makes mountains out of molehills—sometimes fairly so, sometimes

not. In this case, the named plaintiffs ask us to turn their 33 individual lawsuits involving dozens

of cars into one lawsuit involving over 800,000 cars. Because this “exponential aggregation” of

parties “magnifies the stakes of litigation,” Ford, 86 F.4th at 726, bypasses our tradition of

litigation by “individual named parties only,” Califano v. Yamasaki, 442 U.S. 682, 700–01

(1979), and must steer clear of violations of the Seventh Amendment, see Ortiz v. Fibreboard

Corp., 527 U.S. 815, 845–46 (1999), Civil Rule 23 imposes stringent requirements before

permitting the aggregation of so many claims by so many people in one court. To these ends, the

class must show that it is “so numerous” that joinder is impracticable. Fed. R. Civ. P. 23(a)(1).

It must find “questions of law or fact common” to the class. Fed. R. Civ. P. 23(a)(2). The

No. 23-1940 Speerly et al. v. General Motors, LLC Page 8

named plaintiffs’ claims or defenses must be “typical” of those of the unnamed class members.

Fed. R. Civ. P. 23(a)(3). And they must show that they will “fairly and adequately protect” class

interests. Fed. R. Civ. P. 23(a)(4). For classes certified under Civil Rule 23(b)(3), like this one,

the plaintiffs must show that common questions “predominate” and that litigation as a class is

“superior to other available methods” of adjudication.

This is not merely a pleading standard. The plaintiffs must affirmatively “prove” that the

class meets the prerequisites for certification. Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338, 350

(2011); cf. id. at 369–72 (Ginsburg, J., dissenting). And they must do so with “significant

proof.” Id. at 353 (majority op.) (quotation omitted). The trial court, for its part, must “probe

behind the pleadings” and conduct a “rigorous” examination to ensure that the class satisfies

Rule 23 before transforming the retail disposition of claims into the wholesale disposition of

them. See id. at 350–51.

“[R]igorous” as in substance and process, we emphasize, not as in effort. Id. at 351. No

one doubts the district court’s effort in this massive case. A glance at this docket confirms that a

trial court facing a multi-state class action like this one deserves considerable respect and

gratitude. In fact, at this stage, we have no reason to dwell on many of the district court’s class-

action determinations: numerosity, typicality, adequacy of representation, and superiority. No

one challenges the district court’s handling of them in this interlocutory appeal.

More complicated are the commonality and predominance inquiries, where we train our

attention. Rule 23 demands that the court conduct a structured two-step approach to assess

commonality and predominance. Step one: check to see that the plaintiffs have identified a

“common question of law or fact.” Fed. R. Civ. P. 23(a)(2). To be common, a question must

(1) yield a common answer with common evidence and (2) meaningfully progress the lawsuit.

The decisionmaker must be able to resolve the question with “a yes-or-no answer for the class in

one stroke.” Doster v. Kendall, 54 F.4th 398, 430–31 (6th Cir. 2022), vacated as moot, 144

S. Ct. 481 (2023) (quotation omitted). The district court must actually decide whether the

questions are common, Szabo v. Bridgeport Machs., Inc., 249 F.3d 672, 676–77 (7th Cir. 2001),

meaning that it cannot “accept [allegations] as true or construe [evidence] in anyone’s favor,”

Brown v. Electrolux Home Prods., Inc., 817 F.3d 1225, 1234 (11th Cir. 2016). If a reasonable

No. 23-1940 Speerly et al. v. General Motors, LLC Page 9

decisionmaker left with the evidence may answer “yes” to a question for some class members

and “no” for others, the class has not shown that it is common. Doster, 54 F.4th at 430–31;

Nissan, 122 F.4th at 252.

Even then, not every question with a common answer meets Rule 23(a). Else, the fact

that every customer sued GM about a GM car would create a common question suitable for class

certification. The plaintiffs must also show that the question “affect[s] at least one” disputed

“element” of the class’s claims. Doster, 54 F.4th at 430. To conduct a rigorous analysis, the

court must “walk through each cause of action, identify the relevant elements, and evaluate

which elements, if any, submit to common answers.” Nissan, 122 F.4th at 246–47.

That leads to step two: ensure that the common questions “predominate over any

questions affecting only individual members.” Fed. R. Civ. P. 23(b)(3). The court must “put the

common issues on one side, the individual issues on the other, then qualitatively evaluate which

side predominates.” Nissan, 122 F.4th at 252 (quotation omitted). If adding or subtracting

plaintiffs from the class substantially varies the “substance or quantity of evidence offered” and

the cost of doing so, see Fed. R. Civ. P. 23(b)(3), the individualized questions likely overwhelm

the common ones. Brown, 817 F.3d at 1235.

In conducting a Rule 23 analysis, the district court will inevitably address issues that

overlap with the merits inquiry. It “cannot be helped” that commonality and predominance’s

element-by-element, claim-by-claim inquiry implicates the merits of each claim. Dukes, 564

U.S. at 351; Amgen Inc. v. Conn. Ret. Plans & Tr. Funds, 568 U.S. 455, 465–66 (2013). How

else could the court assess the elements of Rule 23? While a free-ranging merits prediction—do

the claimants have a weak or strong product liability case?—is “not properly part of the

certification decision,” the district court may—indeed, must—answer merits questions that bear

on Rule 23’s demands. Fed. R. Civ. P. 23 advisory committee’s note to 2003 amendment.

The district court, as a result, must not defer merits questions bearing on commonality

and predominance until summary judgment. That kind of deferral concept is nowhere in Rule

23’s text, which tells the court to refuse certification until it is assured that certification is proper.

Fed. R. Civ. P. 23(a)–(b). The history of the Rules points in the same direction. The deferral

No. 23-1940 Speerly et al. v. General Motors, LLC Page 10

possibility apparently grew out of a previous regime in which a district court could conditionally

certify a class and prune the non-common, non-predominant issues later. But under the Rules

Enabling Act, the Judicial Conference and the Supreme Court considerably narrowed, if not

eliminated, that option in 2003 when it barred conditional certification and mandated that a

district court “refuse certification” until it is “satisfied that the requirements of Rule 23 have

been met.” See Fed. R. Civ. P. 23 advisory committee’s note to 2003 amendment. Postponing

Rule 23 inquiries until everyone has joined the class runs the risk of avoiding an issue for 33

plaintiffs today in order to decide it for 800,000 tomorrow.

GM challenges the district court’s analysis at both steps of the process. That requires us

to address two questions: Did the district court properly identify common questions of fact or

law? And did the court properly find that those questions predominated over individual ones?

IV.

Commonality. A common question must “resolve an issue that is central to the validity of

each one of the claims.” Dukes, 564 U.S. at 350. To ensure the questions, whether factual or

legal, are central, a court must ensure that they “affect at least one element” of all 59 claims.

Doster, 54 F.4th at 430. The court “must walk through each cause of action” and “identify the

relevant elements.” Nissan, 122 F.4th at 246. Because an issue is not “central” unless it affects

at least one contested element in each cause of action, it flows from Wal-Mart that only an

element-oriented analysis permits the court to identify which questions meaningfully move the

lawsuit forward. 564 U.S. at 350. Only after identifying the relevant elements of each cause of

action may it decide whether a “yes-or-no answer” would resolve an element “for the class in

one stroke.” Doster, 54 F.4th at 431 (quotation omitted).

A question is not common if the answer requires “evidence that varies from member to

member.” Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442, 453 (2016) (quotation omitted). In

assessing whether common issues predominate under Rule 23(b)(3), Tyson Foods clarified that a

common question was one “where the same evidence will suffice for each member” to answer it.

Id. (quotation omitted).

No. 23-1940 Speerly et al. v. General Motors, LLC Page 11

The district court reasoned that “three readily discernible common questions that are

crucial to the pleaded causes of action” exist: whether the transmissions have shift and shudder

defects; whether GM knew about them; and whether the defects are material. R.284 at 17.

Every claim, it pointed out, “demand[s] proof of a defect in the vehicles’ transmission design.”

R.284 at 13. Because the court determined that these questions lend themselves to a common

answer and feature “prominently in the disposition of the case,” it found commonality satisfied.

R.284 at 13, 17.

With respect, that does not suffice. A court may not simply ask whether generalized

questions yield a common answer. That would undermine the bedrock principle that courts must

identify common questions with respect to concrete elements of each claim. By hitching all 59

claims to a question about “defect” in the abstract, the court overlooked how significant

differences across each cause of action raise serious commonality concerns.

The plaintiffs, to illustrate the point, claim that the court may answer whether a “defect”

exists in each transmission in one stroke. But that is not necessarily so in the context of the

relevant elements of each claim. Do the plaintiffs mean “defect” in the products-liability sense

because it is “unreasonably dangerous”? Plas-Tex, Inc. v. U.S. Steel Corp., 772 S.W.2d 442, 444

(Tex. 1989). Or do they mean “defect in an implied warranty” sense, as in unfit for the “ordinary

purposes for which” the accused products “are used”? Id. Or do they mean defect in the sense

of a consumer-protection statute, which asks if the feature “diminished” the “value of the

product”? Schiffner v. Motorola, Inc., 697 N.E.2d 868, 874 (Ill. App. Ct. 1998).

That also means we can’t tell which claims the defect question might not be central to.

The district court, for example, reasoned that the plaintiffs must identify a defect that is “covered

by the express warranty to repair or renders the class vehicles unfit for their ordinary or intended

purpose.” R.284 at 20. And it observed that, without “proof of a defect making the vehicles

unsuitable for ordinary use,” “the plaintiffs cannot prevail on any implied warranty claims.”

R.284 at 25. But these two claims ask different things of the class, confirming the imperative of

an element-driven analysis. A court might find that the element of breach in an express-warranty

claim asks a common question: Does a problem exist in each transmission that GM promised to

fix? But the element of breach in an implied-warranty claim asks a different question: Does a

No. 23-1940 Speerly et al. v. General Motors, LLC Page 12

problem in each transmission amount to a defect that makes the car unfit for its ordinary

purpose? The second question might not be common because some class members’ transmission

problems might not seriously disrupt the safety and comfort of each class vehicle, especially for

those customers who merely felt a “light punch” while upshifting while others felt a “lurch[].”

R.225-22 at 18; R.225-26 at 4.

The same goes for knowledge. Is the right question whether GM knew that the

transmissions had problems with harsh shifting in 2015, as relevant to a consumer-protection

claim? That might lend itself to a common answer to the extent test drivers knew about it. Or is

the relevant question that GM knew that it couldn’t fix buyer vehicles and kept selling them

anyway, as relevant to an express-warranty claim? That might not lend itself to a common

answer because GM repaired the shudder defect in 252,059 8L transmissions, most of which GM

presumably covered with its free warranties. All of this suggests that, without an element-by-

element commonality analysis, this court cannot effectively review which questions are truly

central to which claims. We vacate and remand for the district court to conduct an element-by-

element analysis that assesses how each question is common by fitting it into each claim.

The dissent objects to this analysis on several grounds. It worries that this approach

requires a “rote explication of every element for every cause of action” at the commonality stage.

Dissent at 83. Not necessarily and not likely. As the Court said in response to the same critique

in Wal-Mart, the plaintiffs need to identify only one question suitable for common proof with

respect to each cause of action. 546 U.S. at 359. But that question must be central to the claim

at hand; it can’t relate to a non-material point, such as whether the plaintiffs bought a GM car

during the relevant period. The plaintiffs must tie that debated question to “the relevant

elements” of that claim; they cannot simply ask whether a “defect” exists in the abstract. And

the plaintiffs must prove that common evidence would yield a common answer; they cannot

ignore the individualized inquiries that might occur with respect to the allegedly “common”

question. Nissan, 122 F.4th at 246–47; Doster, 54 F.4th at 432; see Dukes, 564 U.S. at 350–51.

Because commonality’s one-question inquiry is less onerous than predominance’s all-questions-

considered inquiry, this approach does not “transform[] Rule 23(a)(2) into a miniature Rule

23(b)(3) analysis.” Dissent at 106.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 13

The dissent insists that an “issue central to a claim does not necessarily mean an element

of that claim,” Dissent at 82, because a “generalized question[], even if not directly tied to an

element of a claim, can suffice to yield a common answer that drives the litigation forward,” id.

at 84. That may be true in some settings. A vital factual question may well turn a case. But an

engineer alone, to put the point in the context of a product liability case, cannot resolve the

commonality inquiry. It takes a lawyer, too. The class must trace the question to a legally

salient element in each cause of action and show that common proof will provide yes-or-no

answers to that factual/legal question. Doster, 54 F.4th at 430–31. Without this analysis, the

district court cannot prove that a common defect drives all 59 claims forward, even if every

plaintiff can prove that they had the same transmission.

The dissent contends that Wal-Mart permits a district court to “craft a common

contention subject to generalized proof.” Dissent at 85. That was possible in Wal-Mart because

the claimants filed a “pattern or practice” claim under Title VII, in which the proof theoretically

could be generalized and theoretically could lead to common yes-or-no answers for the entire

class. Wal-Mart, 564 U.S. at 353 (noting that a “general policy of discrimination conceivably

could justify a class” bringing a Title VII claim if “the discrimination manifested itself in hiring

and promotion practices in the same general fashion”) (quotation omitted). But even that

possibility did not become a reality in Wal-Mart, as the Court rejected certification on

commonality grounds. See id. at 359–60. If Wal-Mart is a refuge for this class-certification

motion, it thus is not a very habitable one. The Court found commonality wanting even in the

setting of a “pattern or practice” claim. Id. at 352. And today’s case does not involve a “pattern

or practice” claim. Quite to the contrary, it involves a swath of distinct state-law claims in which

it is difficult to see any generalized proof that would permit the yes-or-no answers that Wal-Mart

requires. See Doster, 54 F.4th at 430–31; Nissan, 122 F.4th at 246–47, 252; see also Dukes, 564

U.S. at 350–51.

The dissent’s fourth point—that the plaintiffs have identified “the 8L transmissions’ two

universal defects” as “one common contention that is capable of resolution,” Dissent at 75—

illustrates why this analysis matters. Because “defect” bears different legal meanings for

different causes of action, the class needs to explain how and why class-wide proof of each

No. 23-1940 Speerly et al. v. General Motors, LLC Page 14

alleged fault in the transmission resolves a material component of each cause of action.

Otherwise, the word “defect” becomes a chameleon for every cause of action, no matter how

differently state law gauges the point, no matter how differently each alleged fault manifests.

Put another way, we do not object to the district court’s commonality analysis because it

was one page long. The problem is one of process. By choosing to resolve the commonality

question at the level of generality of a “defect,” the court did not measure the question’s impact

on each cause of action. District courts can still efficiently conduct the one-question

commonality analysis by identifying precisely how each common question advances each claim.

To the extent that analysis remains “labored” in this instance, Dissent at 82, that turns on

plaintiffs’ decision to bring 59 claims in one class action, not the imperatives of Rule 23(a).

V.

Predominance. This inquiry requires a district court to make a claim-by-claim

comparison of the common and non-common questions to see which ones, if any, predominate.

Before undertaking that review, two threshold challenges of bringing these 800,000 claims

together deserve attention.

First, there are high costs to a legal system that asks one district court to understand and

apply nearly 60 causes of action across 26 states. In taking on that task, the court must “follow

the decisions of the state’s highest court when that court has addressed the relevant issue.” See

Savedoff v. Access Grp., Inc., 524 F.3d 754, 762 (6th Cir. 2008) (quotation omitted). But not all

high courts have addressed all of the relevant claims as they apply to this action. In that setting,

the court must “anticipate” through an educated guess “how the relevant state’s highest court

would rule.” Kepley v. Lanz, 715 F.3d 969, 972 (6th Cir. 2013) (quotation omitted).

That’s no mean feat. In certifying a class under Rule 23(b)(3), efficiency offers a good

reason, sometimes an overriding reason. But, in making that call, it’s well to remember what

concerns are being overridden. One is the “undesirability of concentrating the litigation of the

claims in the particular forum” and “the likely difficulties in managing a class,” Fed. R. Civ.

P. 23(b)(3)(C)–(D), as opposed to “allowing the claims to be litigated separately in forums to

which they would ordinarily be brought,” Fed. R. Civ. P. 23 advisory committee’s note to 1966

No. 23-1940 Speerly et al. v. General Motors, LLC Page 15

amendment. A bulky multi-state class action forces a federal court to play “central planner” to

26 state economies under “one case, one court, one set of rules,” and “one settlement price for all

involved.” In re Bridgestone/Firestone, Inc., 288 F.3d 1012, 1020 (7th Cir. 2002). And it forces

one federal judge to guess, sometimes in the first instance, the contours of local business norms.

Id.

The central-planner model also asks a lot of trial courts and juries. Would the court use

hundreds of citizens to form 26 juries, one for each state? If it did, how would it demarcate

subtle differences in state laws to comply with the Seventh Amendment’s command that “no fact

tried by a jury, shall be otherwise reexamined”? More complicated still, would the court

empanel one jury to crown the winner, what amounts to empowering six Michigan citizens to

predict 59 rules of the road for 26 state commercial and consumer norms? If the court took this

approach, must it face head-on the “impossible task of instructing a jury on the relevant law” and

provide 59 pages of verdict forms? Am. Med. Sys., 75 F.3d at 1085. Or would the trial court

issue a central-planning “Esperanto” jury instruction that “merg[es] the negligence standards of

the 50 states and the District of Columbia,” letting six people “hold the fate of an industry in the

palm of its hand”? In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293, 1300 (7th Cir. 1995). First

principles of class-action law—Rule 23(b)(3)’s mandate that we consider “the likely difficulties

in managing a class action” in assessing predominance, and the reality that “federal class

action[s] based on state law” “undermine federalism,” Thorogood v. Sears, Roebuck & Co., 547

F.3d 742, 745 (7th Cir. 2008)—deserve serious consideration before certifying such a large class.

Second, this class action presents two theories of defect, each with multiple moving parts,

exponentially increasing these challenges. In asking whether certification will “achieve

economies of time, effort,” “expense,” and “uniformity of decision,” Fed. R. Civ. P. 23 advisory

committee’s note to 1966 amendment, courts should be wary about whether two different

theories will both achieve these economies of scale. If just one theory for four causes of action

in as many as 26 states presents daunting challenges, one should be wary about underwriting

those risks for two theories.

Many of our sister circuits recognize the perils that multi-state, multi-defect, multi-claim

classes pose. They look skeptically on plaintiffs who “aggregate a plethora of discrete claims”

No. 23-1940 Speerly et al. v. General Motors, LLC Page 16

“into one super-claim” because doing so increases “the complexity of performing the required

rigorous analysis” for certification, let alone trial. M.D. ex rel. Stukenberg v. Perry, 675 F.3d

832, 848 (5th Cir. 2012) (quotation omitted); see Marisol A. v. Giuliani, 126 F.3d 372, 378–79

(2d Cir. 1997) (per curiam). Filing “a complex class action complaint alleging differing theories

of recovery,” some of which are common and some not, typically reveals a “failure to

demonstrate that the common issues predominate” across each theory. Gunnells v. Healthplan

Servs., Inc., 348 F.3d 417, 462–64 (4th Cir. 2003). That makes sense. Class actions

“economiz[e] on the expense of litigation” by resolving key issues in one stroke. Thorogood,

547 F.3d at 744. Certifying multiple theories of liability turns a bet that resolving one set of

claims will save time into a multi-leg parlay that resolving every set of claims will save time.

The class responds to this second concern with a citation to a Seventh Circuit case

permitting certification in a case involving two theories of defect. Butler v. Sears, Roebuck &

Co., 727 F.3d 796 (7th Cir. 2013). But Butler underscores our point. The plaintiffs brought

“two class actions” for two defects, prompting the district court to certify one class but not the

other. Id. at 797–98. By contrast, the plaintiffs today ask us to certify both theories. While

some courts have considered multiple-defect classes, see, e.g., Bridgestone, 288 F.3d at 1019

(refusing to certify on other grounds), we approach them with the caution commensurate with

their risk. For like reasons, we cannot agree with the dissent that the union of two distinct

theories, viewed through the prisms of 59 claims, poses only a “superficial difference.” Dissent

at 93.

The dissent claims that we should review only whether the district court’s analysis “was

careful, thorough, and rigorous.” Dissent at 89. But because the plaintiffs “must affirmatively

demonstrate” compliance with Rule 23, Dukes, 564 U.S. at 350, because we don’t defer to

district court readings of state law, Leavitt v. Jane L., 518 U.S. 137, 145 (1996) (per curiam), and

because even one individualized issue risks dramatically increasing the costs of class litigation, a

“rigorous analysis” demands that we confirm that the district court addressed each individualized

concern. Dukes, 564 U.S. at 351.

In addressing these concerns, the district court may not ignore “slight variations across

state law.” Dissent at 89. The inquiry evaluates predominance within, not across, each of the 26

No. 23-1940 Speerly et al. v. General Motors, LLC Page 17

subclasses because they “are each treated as a class.” Fed. R. Civ. P. 23(c)(5). And even a slight

variation in state law—say, a reliance requirement—might markedly increase the “difficulties in

managing [the] class.” Fed. R. Civ. P. 23(b)(3)(C).

That helps to explain why the district court may not certify 26 subclasses on the ground

that some of them do not have defect-manifestation requirements. Dissent at 98. Because a

court must treat each subclass “as a class,” Fed. R. Civ. P. 23(c)(5), “each subclass must

independently meet the requirements of Rule 23,” 7AA Charles Alan Wright & Arthur R. Miller,

Federal Practice and Procedure § 1790 (3d ed. 2005). A subclass with highly individualized

issues may not ride the coattails of another subclass without them.

Moving from the general to the specific, we turn to gauge whether any common

questions for each of these four causes of action predominate over the non-common ones.

A. Breach of Express Warranty

Subclasses for 17 of the states—Alabama, Colorado, Delaware, Georgia, Idaho, Illinois,

Kansas, Kentucky, Maine, Minnesota, New Hampshire, New Jersey, New York, Oklahoma,

South Carolina, Texas, and Washington—bring express warranty claims. Common questions do

not appear to predominate in any of them.

Alabama shows why. It offers a typical approach by handling express warranty claims

under general contract principles. Ex parte Miller, 693 So. 2d 1372, 1376 (Ala. 1997). Under

that state’s law, the claim requires the plaintiff to show a (1) valid contract, (2) performance by

the plaintiff, (3) non-performance by the defendant, and (4) damages. Childs v. Pommer, 348

So. 3d 379, 387 (Ala. 2021). An examination of the first and third elements, as applied to this

class action, reveals the problem.

At one level, the first and third elements lend themselves to some common proof. A

valid warranty consists of an “affirmation of fact” relating to the goods, forming part of the

bargain. Ala. Code § 7-2-313(1)(a). And the record suggests that each member of this subclass

received the same warranty or at least one that was materially the same. With the presentation of

a warranty, GM promised to cover “any vehicle defect” and pay for “repairs, including towing,

No. 23-1940 Speerly et al. v. General Motors, LLC Page 18

parts, and labor” “at no charge.” R.41-2 at 9; R.41-3 at 10; R.245-60 at 10. The third element—

breach—requires a court to answer what looks like one common question: Did GM originally

install defective transmissions in each vehicle?

But the breach inquiry also presents an individualized question that overwhelms the

efficiency gains of the common questions. Because the contract makes “repairs” the “exclusive

remedy” for defect problems, R.245-60 at 17–18; R.41-2 at 17; R.41-3 at 19, GM fails to

perform only if it refuses to repair the buyers’ cars or does so ineffectively. Am. Suzuki Motor

Corp. v. Burns, 81 So. 3d 320, 325 (Ala. 2011).

Whether the class members brought their vehicles in for repair is deeply

individualized. Some people brought their cars in for repair during the warranty period, and

some did not. Some buyers received software repairs, and some received hardware repairs.

Some repairs fixed shudder and shift problems, and some did not. Some technicians could

replicate customer concerns, and some could not. Some customers received “[s]ervice

[c]alibrations” that improved shift quality, and some did not. R.206-15 at 5. All of this means

that the “substance or quantity of evidence offered” varies among unnamed class members.

Brown, 817 F.3d at 1235 (quotation omitted). That is a problem.

Worse, that problem pervades every individual express-warranty claim. By our reading,

every one of the seventeen states provides that a seller breaches a repair warranty only when he

fails to repair something, not when the defect manifests. Cooley v. Big Horn Harvestore Sys.,

Inc., 813 P.2d 736, 744–45 (Colo. 1991); Gutridge v. Iffland, 889 A.2d 283, at *1, *4 n.11 (Del.

2005) (unpublished table decision); Feinour v. Ricker Co., 566 S.E.2d 396, 398 (Ga. Ct. App.

2002); Clark v. Int’l Harvester Co., 581 P.2d 784, 801 (Idaho 1978); Mydlach v.

DaimlerChrysler Corp., 875 N.E.2d 1047, 1059 (Ill. 2007); Voth v. Chrysler Motor Corp., 545

P.2d 371, 378 (Kan. 1976); Moore v. Mack Trucks, Inc., 40 S.W.3d 888, 891 (Ky. Ct. App.

2001); Inniss v. Methot Buick-Opel, Inc., 506 A.2d 212, 216 (Me. 1986); Anderson v. Crestliner,

Inc., 564 N.W.2d 218, 222 (Minn. Ct. App. 1997); Welch v. Fitzgerald-Hicks Dodge, Inc., 430

A.2d 144, 148 (N.H. 1981); Kearney & Trecker Corp. v. Master Engraving Co., 527 A.2d 429,

434–35 (N.J. 1987); Solomon v. Canon USA, Inc., 920 N.Y.S.2d 565, 566 (N.Y. App. Term

2010) (per curiam); Osburn v. Bendix Home Sys., Inc., 613 P.2d 445, 449–50 (Okla. 1980);

No. 23-1940 Speerly et al. v. General Motors, LLC Page 19

Cannon v. Pulliam Motor Co., 94 S.E.2d 397, 400 (S.C. 1956); Northland Indus., Inc. v. Kouba,

620 S.W.3d 411, 417 (Tex. 2020); Schroeder v. Fageol Motors, Inc., 544 P.2d 20, 25 (Wash.

1975).

That is not surprising. Alabama, like the other 16 states with express warranty claims,

follows Article 2 of the Uniform Commercial Code, which permits merchants to limit the

remedies for warranty breaches to the “repair and replacement of non-conforming goods or

parts.” U.C.C. § 2-719(1)(a) (Am. L. Inst. & Unif. L. Comm’n 2022). Because that framework

requires a district court to differentiate those who received a successful repair from those who

did not, a claim-by-claim, element-by-element analysis should have revealed that presentment

poses an individualized task, one unsuitable for class relief.

The class and the dissent seek to fend off this conclusion by noting that the plaintiffs need

not present their vehicles for repair if GM could not fix them. Some states, true enough, do not

require a buyer to seek a futile repair. See, e.g., Nissan, 122 F.4th at 250. But any warranty-

covered Mod1A flushes would have fixed the shudder problems. And some of GM’s software

and hardware calibrations materially improved shift quality. That all suggests that some of the

warranty repairs would not “fail of [their] essential purpose.” Ga. Code Ann. § 11-2-712(2);

Dissent at 94.

Because the district court never conducted that analysis state by state, it left unclear

whether this is an abstract problem or a real one. The court instead dismissed GM’s objections

as “merits issues that are irrelevant at this stage.” R.284 at 52. In fairness to the district court,

the case law in this area is riddled with loose language along these lines—either language that

reserves some merits inquiries for trial or reserves the right to revisit class definitions and

certification orders if problems emerge down the road. In one sense, these cautions are

understandable. No one can predict the future, and some humility is in order about the handling

of a years-away trial. But if the predominance inquiry is to serve its critical function, it cannot

be answered by “maybe,” “perhaps,” and other “what ifs” that leave the hard questions for later

or, worse, that ignore the questions that cut against certification. The arc of a properly handled

class action puts the difficulty of answering individualized legal and factual questions together at

No. 23-1940 Speerly et al. v. General Motors, LLC Page 20

its apex during the class-certification inquiry—and most especially during the predominance

inquiry. It is not a merits question that can be sloughed off until a “later” that may never come.

That makes sense, given what’s at stake during class certification. “While the benefits to

the individual class members are usually miniscule, the possible consequences of a judgment to

the defendant are so horrendous that these actions are almost always settled,” making the class-

certification order too often the main event. See Henry J. Friendly, Federal Jurisdiction: A

General View 119–20 (1973). “[I]ncorrectly certified classes . . . coerce businesses into costly

settlements that they sometimes must reluctantly swallow rather than betting the company on the

uncertainties of trial.” Lab’y Corp. of Am. Holdings v. Davis, 145 S. Ct. 1608, 1612 (2025)

(Kavanaugh, J., dissenting). These “coerced settlements substantially raise the costs of doing

business” for companies, which “in turn pass on those costs to consumers,” investors, and

workers. Id. Properly applied, the predominance inquiry (and sometimes the superiority inquiry

too) avoids that risk by focusing on whether the class action would lead to a fair and efficient

trial, not a settlement. An element-by-element comparison of a cause of action to determine

whether common questions do, or do not, predominate requires a consideration of all of the

elements of each legal claim at the outset. Any other approach does not come to grips with the

central point of the predominance inquiry and subverts the fairness and efficiency considerations

underlying Rule 23.

All in all, we are doubtful that the common questions predominate over the non-common

ones with respect to this cause of action. But we leave it to the district court to apply these

principles in the first instance.

B. Breach of Implied Warranty

Subclasses for 14 states—Arkansas, Colorado, Delaware, Georgia, Illinois, Kansas,

Maine, Michigan, Minnesota, New Hampshire, New Jersey, New York, Oklahoma, and South

Carolina—raise implied warranty claims. Here, too, a predominance problem emerges.

Georgia, as an exemplar of this cause of action, illustrates why. In the Peach State,

breach of the implied warranty of merchantability requires a plaintiff to prove four

elements: (1) the goods were subject to a warranty, (2) the goods were defective, (3) the defect

No. 23-1940 Speerly et al. v. General Motors, LLC Page 21

caused an injury, and (4) that injury caused damages. Mitchell v. BBB Servs. Co., 582 S.E.2d

470, 471–72 (Ga. Ct. App. 2003). Defectiveness bears a technical meaning in the implied-

warranty context. Cf. Nissan, 122 F.4th at 248–49. A car is merchantable if it “pass[es] without

objection in the trade under the contract description,” is of “fair average quality,” and is “fit for

the ordinary purposes for which” it is used. Ga. Code Ann. § 11-2-314(2).

The first and second elements appear to implicate some questions yielding common

answers, though the court should re-assess whether they are central to each claim on remand.

The court must know whether GM’s transmission was defective before it can tell whether that

defect rendered GM cars unmerchantable. And GM can supply a single answer to identify the

“ordinary purpose[]” for which reasonable consumers use cars: driving. Id.

But the second element also has individualized features that rebalance the ledger.

Determining whether the alleged common defects made the cars unmerchantable compels

individualized inquiries twice over. First, different environmental conditions led to different

presentations of the alleged defect—and sometimes meant the defect never appeared. Drivers in

dry climates experienced fewer instances of shudder because less moisture polluted the

automatic transmission fluid. The defendants have attached a compilation of class members who

“never experienced shudder” or “no longer” do. R.245-59 at 2. Under Georgia law, a car may

have “minor problems” that “never render[] the vehicle unusable,” and that’s especially so if the

dealer resolves them “in [a] timely fashion.” Soto v. CarMax Auto Superstores, Inc., 611 S.E.2d

108, 110 (Ga. Ct. App. 2005).

Second, the defects that did manifest did so in different ways. Take the alleged shifting

defect. Some customers felt a “light punch in the arm” while upshifting. R.225-22 at 18. Others

felt a “hesitat[ion]” while accelerating. R.225-16 at 4–5. Still others found their cars “lurch[ing]

forward” when entering second gear. R.225-5 at 8. These individualized inquiries, which

require assessing the presence and extent of the defect, vary from class member to class member.

See Brown, 817 F.3d at 1235. Such variations make it difficult to conclude that the common

questions predominate.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 22

Every other state at issue agrees. Their positions reflect the commercial norm that a

“product is merchantable” if a defective “feature does not make the product as a whole unfit.”

2 Hawkland Uniform Commercial Code Series § 2-314:3 (database updated 2024). And each

state under which the plaintiffs have brought implied-warranty claims has concluded that

unmerchantability does not turn alone on whether there is a defect in a component part. The

class must prove that the defect made the entire product unfit for ordinary use. See, e.g., H.A.S.

of Fort Smith, LLC v. J.V. Mfg., Inc., No. CA03-1445, 2004 WL 2102009, at *3 (Ark. Ct. App.

Sept. 22, 2004); Anderson v. M.W. Kellogg Co., 766 P.2d 637, 643 (Colo. 1988) (en banc),

Freedman v. Chrysler Corp., 564 A.2d 691, 697–99 (Del. Super. Ct. 1989); Malawy v. Richards

Mfg. Co., 501 N.E.2d 376, 383–84 (Ill. App. Ct. 1986); Hodges v. Johnson, 199 P.3d 1251, 1260

(Kan. 2009); Suminski v. Me. Appliance Warehouse, Inc., 602 A.2d 1173, 1175 (Me. 1992);

Rosenbaum v. Toyota Motor Sales, USA, Inc., No. 16-CV-12645, 2016 WL 9775018, at *2 (E.D.

Mich. Oct. 21, 2016), aff’d, 708 F. App’x 242 (6th Cir. 2017) (applying Michigan’s definition of

merchantability to suggest that the Toyota Prius’s poor performance in “electric mode” does not

make it unmerchantable in general); Tellinghuisen v. Chrysler Grp., LLC, No. A13-2194, 2014

WL 4289014, at *3 (Minn. Ct. App. Sept. 2, 2014); Ferrari v. Am. Honda Motor Co., Inc., No.

L-4296-06, 2009 WL 211702, at *3–4 (N.J. Super. Ct. App. Div. Jan. 30, 2009) (per curiam); see

Welch, 430 A.2d at 147–49 (New Hampshire law); Wojcik v. Empire Forklift, Inc., 14 A.D.3d

63, 66 (N.Y. App. Div. 2004); Perry v. Lawson Ford Tractor Co., 613 P.2d 458, 463 (Okla.

1980); Seaside Resorts, Inc. v. Club Car, Inc., 416 S.E.2d 655, 662 (S.C. Ct. App. 1992). Even

if the class may answer with one voice whether the 8L transmissions were defective, the class

must answer with many voices whether their specific cars became unmerchantable given how

their specific transmissions revealed the alleged defect.

(As a sidenote, the GM express warranty contracts say that repair is the only remedy for

implied-warranty claims. If so, the presentment issues identified under the express-warranty

claims may apply here as well.)

The proposed class is not analogous to certifying “a class of people sick with a cold

because some sniffled” and “others coughed.” Dissent at 93. At issue are two distinct

engineering diagnoses involving two different transmission components that GM fixed in

No. 23-1940 Speerly et al. v. General Motors, LLC Page 23

different ways under different repair schedules. These are hardly “superficial differences,”

Dissent at 93, in the context of an implied-warranty claim, to use one example. How the 8L

transmission problems manifest is precisely what matters for merchantability in showing how

bad the driving experience becomes for each consumer. To the extent healthcare offers a useful

analogy to this case, it would be to a class of hospital patients in which some patients had scurvy

and others had gout—different diagnoses and different symptoms.

As with the class’s express-warranty claims, we doubt that the common questions

predominate over the non-common ones with respect to the implied-warranty claims. But we

leave it to the district court to apply these principles in the first instance.

C. Consumer Protection Statutes

Subclasses for 23 states—Alabama, Arizona, Arkansas, Delaware, Florida, Idaho,

Illinois, Kansas, Kentucky, Louisiana, Maine, Michigan, Minnesota, New Hampshire, New

Jersey, New York, North Carolina, Oklahoma, Pennsylvania, Tennessee, Texas, Washington,

and Wisconsin—bring claims under their consumer-protection statutes. These claims face two

general hurdles. The first is a requirement in some states that a product defect must manifest

before a consumer may bring such a lawsuit. The second is a requirement in some states that a

consumer must show he actually relied on a merchant’s misrepresentation or omission. Only the

state regimes that impose neither requirement avoid these predominance pitfalls.

Manifest Defect. States have options when a car company sells a product with a risk of

injury. They may compensate with remedies that sound in tort, allowing recovery only to those

physically injured by a manifest defect. Bridgestone, 288 F.3d at 1017 n.1. But because tort

remedies usually don’t countenance “purely economic losses,” 2000 Watermark Ass’n, Inc. v.

Celotex Corp., 784 F.2d 1183, 1185 (4th Cir. 1986), they may choose remedies that sound in

contract and allow every buyer to recover for the decreased value of a risky product, even if the

risk does not manifest, Bridgestone, 288 F.3d at 1017 n.1. Because the tort-like approach poses

an individualized query, the district court must identify which statutes follow which model.

Texas shows how a manifest-defect requirement overwhelms any common questions. In

the Lone Star State, a plaintiff must show that “(1) he is a consumer, (2) the defendant engaged

No. 23-1940 Speerly et al. v. General Motors, LLC Page 24

in a false, misleading, or deceptive act, and (3) the act constituted a producing cause of economic

damages.” Everett v. TK-Taito, L.L.C., 178 S.W.3d 844, 857 (Tex. App. 2005). To address the

second element, the court must assess whether GM’s 8L transmissions were prone to shudders

and harsh shifts, a matter that might lend itself to common proof.

But the third element, causation and injury, hampers a class-wide yes-or-no answer.

Under Texas law, economic loss “from a defect that has not manifested” is “too remote in time to

constitute an ‘injury.’” Id. at 858. That brings to the surface highly individualized questions,

doubly so in view of the plaintiffs’ two theories of defect: identifying which customers actually

experienced shudders and which customers actually experienced harsh shifts. In addition, the

court must understand which consumers’ shudders and harsh shifts arose from the transmissions

rather than, say, their tires, as worn tires may create a shudder as well. Just as with the implied

warranty claims, those questions appear to overwhelm the common ones.

Texas is not alone. Arkansas, New Hampshire, New York, Oklahoma, and Wisconsin

take the same approach. Wallis v. Ford Motor Co., 208 S.W.3d 153, 161 (Ark. 2005) (“Where

the only alleged injury is the diminution in value of the product, a private cause of action is not

cognizable under the ADTPA.”); Parnell v. FanDuel, Inc., 591 S.W.3d 315, 319–20 (Ark. 2019);

Nichols v. Gen. Motors Corp., No. 99-C-566, 1999 WL 33292839, at *3–5 (N.H. Super. Ct. Dec.

13, 1999) (barring a consumer claim for failure to allege that “the unreasonably [un]safe

condition . . . has manifested itself”); Frank v. DaimlerChrysler Corp., 292 A.D.2d 118, 121–28

(N.Y. App. Div. 2002); Tietsworth v. Harley-Davidson, Inc., 677 N.W.2d 233, 236–37 (Wis.

2004); see Walls v. Am. Tobacco Co., 11 P.3d 626, 629 (Okla. 2000). Although the Alabama

Supreme Court has not directly spoken on this issue, we understand Alabama law to bar

consumer claims generally when a defect doesn’t manifest. Ford Motor Co. v. Rice, 726 So. 2d

626, 628–29 (Ala. 1998). The same is true of Minnesota, which rejects “allegation[s] of

diminished value due to a propensity to fail” as too “conjectural” and “speculative.” Carey v.

Select Comfort Corp., No. 27CV 04-015451, 2006 WL 871619, at *3 (Minn. Dist. Ct. Jan. 30,

2006); see O’Neil v. Simplicity, Inc., 574 F.3d 501, 503 (8th Cir. 2009).

By contrast, 12 states—Arizona, Delaware, Florida, Illinois, Kansas, Kentucky, Maine,

Michigan, New Jersey, Pennsylvania, Tennessee, and Washington—appear not to require a

No. 23-1940 Speerly et al. v. General Motors, LLC Page 25

defect to manifest. Consumer statutes in those states, as we read them, support a price-premium

theory of harm, where someone who purchases a car expecting a defect-free transmission but

receives a defective transmission has suffered a cognizable harm. See Pena v. Opic, No. 1 CA-

CV 09-0401, 2010 WL 1998152, at *5–7 (Ariz. Ct. App. May 18, 2010); Stephenson v. Capano

Dev., Inc., 462 A.2d 1069, 1076 (Del. 1983); Collins v. DaimlerChrysler Corp., 894 So. 2d 988,

990 (Fla. Dist. Ct. App. 2004); Schiffner, 697 N.E.2d at 874 (Illinois law); Ricklefs v. Clemens,

531 P.2d 94, 100 (Kan. 1975); Smith v. Gen. Motors Corp., 979 S.W.2d 127, 131 (Ky. Ct. App.

1998); Corder v. Ford Motor Co., 285 F. App’x 226, 229 (6th Cir. 2008) (suggesting that failing

to disclose that a 2004 truck contained a 2003 engine imposed an ascertainable harm at Kentucky

law); Everest v. Leviton Mfg. Co., No. CV-04-612, 2006 WL 381832, at *2 (Me. Super. Ct. Jan.

13, 2006) (citing Nelson v. Leo’s Auto Sales, Inc., 185 A.2d 121, 123 (Me. 1962)); Mayhall v.

A.H. Pond Co., 341 N.W.2d 268, 271–72 (Mich. 1983) (per curiam) (suggesting that someone

suffers an injury when he “does not receive what he expected”); Thiedemann v. Mercedes-Benz

USA, LLC, 872 A.2d 783, 789 (N.J. 2005) (implying that “expert proof of diminution of value”

may establish an injury under New Jersey’s consumer law); Grant v. Bridgestone Firestone Inc.,

57 Pa. D. & C.4th 72, at *2 (Com. Pl. Pa. 2002); Morris v. Mack’s Used Cars, 824 S.W.2d 538,

539–41 (Tenn. 1992); Tallmadge v. Aurora Chrysler Plymouth, Inc., 605 P.2d 1275, 93–94

(Wash. Ct. App. 1979) (finding an injury when a buyer expected a new car but received a

repaired car).

Some states, like New Jersey, do not treat a defect as manifested if it arises within the

warranty period and the dealer fixes it. Thiedemann, 872 A.2d at 794. That returns the cause of

action and state subclass to a one-by-one determination, in which the individual question

overtakes the common one.

Some states make district courts guess. North Carolina, for one, has avoided deciding

whether its consumer-fraud statute permits plaintiffs to recover for the economic loss of losing

the benefit of a bargain. Coker v. DaimlerChrysler Corp., 617 S.E.2d 306, 314 (N.C. Ct. App.

2005). But the state does not apply that restriction to “claims for fraud brought

contemporaneously with claims for breach of contract,” Bradley Woodcraft, Inc. v. Bodden, 795

S.E.2d 253, 259 (N.C. Ct. App. 2016), leading us to cautiously venture that it would not impose

No. 23-1940 Speerly et al. v. General Motors, LLC Page 26

a manifestation requirement for consumer claims either. The same is true of Idaho and

Louisiana. Taylor v. Taylor, 422 P.3d 1116, 1125 (Idaho 2018); Albert Switzer & Assocs., Inc. v.

Dixie Buick, Inc., 265 So. 2d 313, 314–15 (La. Ct. App. 1972).

The district court offered two reasons to defer this question for trial: (1) This court has

certified a class without a manifested defect before, Whirlpool, 722 F.3d at 856–57, and (2) the

“appropriate opportunity to address claims of absent class members whose vehicles never have

manifested any defect is a Rule 56 motion for summary judgment,” R.284 at 47. That does not

suffice to rigorously compare the common and non-common questions for predominance. In

Whirlpool, this court concluded that “under Ohio law,” “not all class members must demonstrate

manifestation.” 722 F.3d at 857. But the contours of one state’s law—and one not even at issue

in this case—does not bear on whether the 23 states at issue require manifestation. Treating

Ohio as a stand in for everyone would pave over each State’s diverse norms and betray Rule 23’s

requirement that the court conduct a claim-by-claim analysis.

That’s also why the court may not defer the question until summary judgment. Without

showing that each subclass’s consumer statute may proceed on economic injuries alone,

commonality “completely collapses, rendering class certification inappropriate.” Halliburton

Co. v. Erica P. John Fund, Inc., 573 U.S. 258, 283 (2014). Whether each state’s consumer-fraud

statute has a manifestation requirement “must be proved before class certification.” Id.

Halliburton’s teaching, contrary to the dissent’s suggestion, extends beyond the setting of

that case. Halliburton is no more a case just about securities fraud than Wal-Mart is a case just

about employment discrimination. Rule 23 establishes a set of trans-substantive requirements,

and the Court’s interpretations of the Rule apply to all manner of causes of action. Fed. R. Civ.

P. 23 advisory committee’s note to 1966 amendment.

The plaintiffs’ objections come up short. They note that a manifest-defect rule is not

universal because some states do not require a defect to manifest if the claimant alleges a present

economic loss. But that proves our point. If the states have split over whether an alleged defect

must manifest, why disrespect half of them by certifying all 26 state-wide subclasses? The

rigorous-analysis requirement demands that a district court step through each element of each

No. 23-1940 Speerly et al. v. General Motors, LLC Page 27

claim for each state to identify exactly which ones present individualized issues. Certifying now,

winnowing later, betrays the Supreme Court’s imperative that plaintiffs “actually prove” “that

their proposed class[es] satisf[y] each requirement of Rule 23.” Halliburton, 573 U.S. at 275.

As for the states that require manifestation, the plaintiffs argue, the court may dig through

GM’s “warranty service data” to cull class members whose defects did not manifest after they

used their warranties to fix the relevant problem. R.284 at 47. It’s not that easy. What of the

buyers who never brought their cars to a GM dealership? That wouldn’t be on GM’s service

data spreadsheet. What of the buyers whose defects continued to manifest after they turned their

car in? That wouldn’t be on the spreadsheet either. What of the fact that shift defect experiences

ran the gamut, from nothing at all, to a “hesitat[ion]” while accelerating, R.225-16 at 4–5, to a

“light punch in the arm” while upshifting, R.225-22 at 18, to a dangerous “lurch[ing] forward”

when entering second gear, R.225-5 at 8–9? It’s highly unclear—indeed doubtful—that the

spreadsheet answers these questions. If culling in this way through the spreadsheet was that

easy, why didn’t the claimants and the district court agree to limit the class at the certification

stage to only those claimants whose defects manifested? The certify-now-cull-later approach

raises many questions, and none of them, so far as this record shows, favors predominance.

As for Texas, the dissent objects that Everett was a standing case, not a consumer-law

case. Yes and no. Yes, Everett held that an unmanifested defect could not supply statutory

standing—whether “a particular plaintiff has established that he has been injured or wronged

within the parameters of the statutory language.” Nephrology Leaders & Assocs. v. Am. Renal

Assocs. LLC, 573 S.W.3d 912, 916 (Tex. App. 2019). But no, this matter of statutory

interpretation has nothing to do with “the Texas Constitution’s standing requirements.” Id. at

917. The material points remain: Texas law simply does not support a cause of action for an

unmanifested defect if some cars “functioned as represented” for years. Everett, 178 S.W.3d at

858–59.

Reliance. In the second category are states that require a buyer to show that he relied on

a misleading representation or omission. As we have been told before, reliance tends to pose “an

insuperable barrier to class certification.” Dukes, 564 U.S. at 351 n.6; see Halliburton, 573 U.S.

at 265–66. This case does not call for an exception to the rule. GM’s consumers went to

No. 23-1940 Speerly et al. v. General Motors, LLC Page 28

different dealerships, heard different sales pitches, purchased different vehicles, and received

different prices. It thus remains unclear whether they would “react uniformly to a change in

information about . . . the alleged transmission issues.” R.245-37 at 17. That’s why one plaintiff

“read brochures from his dealership” that led him to buy a specific car. R.245-37 at 18. That’s

why another plaintiff chose a different GM model because a salesman marketed it to him. All of

this adds up to a tall task for the plaintiffs, who must show that they were “actually deceived” by

a “statement or omission” that influenced their purchasing decisions. E.g., De Bouse v. Bayer,

922 N.E.2d 309, 316, 319 (Ill. 2009).

Illinois provides a helpful template for what a careful analysis of one state’s consumer

statute should entail. Under the Illinois Consumer Fraud and Deceptive Practices Act, a plaintiff

must show “(1) a deceptive act or practice by the defendant, (2) the defendant’s intent that the

plaintiff rely on the deception, (3) the occurrence of the deception in a course of conduct

involving trade or commerce, and (4) actual damage to the plaintiff that is (5) a result of the

deception.” Id. at 313. The first and third elements may implicate common questions. To know

if GM deceived buyers about the transmissions’ shift and shudder problems, the court must ask

whether the transmissions had faulty fluid and valve-management issues as designed.

But the fourth and fifth elements demand individualized scrutiny. Though the Illinois

statute lacks an express reliance element, Martin v. Heinold Commodities, Inc., 643 N.E.2d 734,

754 (Ill. 1994), a buyer “must have relied on the wrong to some extent in order to establish

proximate cause,” Zekman v. Direct Am. Marketers, Inc., 675 N.E.2d 994, 998 (Ill. App. Ct.

1997), rev’d on other grounds, 695 N.E.2d 853 (Ill. 1998). That principle couples with Illinois’

common-law tradition that a plaintiff “need[s] to prove actual reliance” to establish proximate

cause. Oliveira v. Amoco Oil Co., 776 N.E.2d 151, 161 (Ill. 2002) (quoting M. Polelle & B.

Ottley, Illinois Tort Law § 9.05, at 9–32 (3d ed. 2001)). A deceptive statement, then, does not

cause a buyer’s injury “unless it actually deceives the plaintiff.” Shannon v. Boise Cascade

Corp., 805 N.E.2d 213, 217 (Ill. 2004).

Divining the purchase price of a broken promise is a personal and individualized matter.

To show that GM deceived customers by misstating the efficacy of the 8L transmissions in its

advertisements, the court has to know which customers received those advertisements. If a buyer

No. 23-1940 Speerly et al. v. General Motors, LLC Page 29

did not see GM’s advertisements, or did not attend GM auto shows touting its transmissions,

those representations did not hurt him. To know whether GM’s statements or omissions caused a

consumer to purchase the car, the court must ask each consumer why he purchased it. Siegel v.

Shell Oil Co., 612 F.3d 932, 936 (7th Cir. 2010) (finding proximate cause under the ICFA an

individualized inquiry). Attempting to satisfy this burden, the plaintiffs provide 33 deposition

transcripts from customers who say that they would not have purchased the class vehicles had

they known of the defect beforehand. But that’s merely 33 down with at least 799,967

interviews to go. Not every purchaser says no to a product because of a defect, especially those

who paid under list price or received too-good-to-pass-up deals on other features.

Arizona, Delaware, Kansas, and Wisconsin follow Illinois’s lead. They require buyers to

prove that a merchant’s misrepresentation or omission caused their injury, and causation usually

requires a showing that a buyer relied on the alleged misstatement. Kuehn v. Stanley, 91 P.3d

346, 351–52 (Ariz. Ct. App. 2004); Carroll v. Philip Morris USA, Inc., 163 A.3d 91, 110 (Del.

Super. Ct. 2017) (finding that the “variability in the reasons people engage in a consumer

transaction” an individualized inquiry under the Delaware Consumer Fraud Act); Finstad v.

Washburn Univ. of Topeka, 845 P.2d 685, 691–92 (Kan. 1993); Novell v. Migliaccio, 749

N.W.2d 544, 151–52 (Wis. 2008) (suggesting that “[r]eliance is an aspect of . . . whether a

representation caused” an injury).

Arkansas, Maine, North Carolina, Pennsylvania, and Texas go further and require a

consumer to prove that he relied on a deceptive statement, adhering to “the traditional common

law elements of reliance and causation.” Weinberg v. Sun Co., 777 A.2d 442, 446 (Pa. 2001);

see, e.g., Ark. Code Ann. § 4-88-113(f)(1)(A) (2017) (requiring, after August 2017, a plaintiff to

suffer “an actual financial loss as a result of his or her reliance” on an unlawful trade practice);

Bartner v. Carter, 405 A.2d 194, 201 (Me. 1979) (holding that a consumer cannot show that he

suffered a loss “as a result of false representations by proving merely that the representations had

a ‘capacity or tendency to deceive’”); Bumpers v. Cmty. Bank of N. Va, 747 S.E.2d 220, 226

(N.C. 2013); Toy v. Metro. Life Ins., 928 A.2d 186, 202–03 (Pa. 2007); Henry Schein, Inc. v.

Stromboe, 102 S.W.3d 675, 694 (Tex. 2002).

No. 23-1940 Speerly et al. v. General Motors, LLC Page 30

Alabama, Idaho, Oklahoma, and Louisiana provide little indication either way—and

return us to the multi-state guessing conundrum. The first two states’ consumer-fraud statutes

require that a plaintiff sue only those who “cause[]” him harm. Ala. Code § 8-19-10(a);

Patterson v. Beall, 19 P.3d 839, 846–47 (Okla. 2000). The latter two require the plaintiff to

suffer “as a result of” a deceptive act. La. Stat. Ann. § 51:1409(A); Idaho Code § 48-608(1).

Prospecting these statutes, we understand the laws’ causation requirements to reflect the ordinary

understanding that causation “is not possible without reliance.” Apex Oil Co. v. Jones Stephens

Corp., 881 F.3d 658, 662 (8th Cir. 2018); see Restatement (Second) of Torts § 546 cmt. a, at

102–03 (Am. L. Inst. 1977).

Minnesota and New Jersey take a different approach and sometimes presume reliance if a

consumer can show it through the “materiality and pervasiveness” of the fraud alone, State v.

Minn. Sch. of Bus., Inc., 935 N.W.2d 124, 135 (Minn. 2019), or if “all the representations about

the product are baseless,” Lee v. Carter-Reed Co., L.L.C., 4 A.3d 561, 580 (N.J. 2010). But that

leads to another crossroad. The presumption could apply and commonly prove reliance, or it

could not apply and require courts to trod through thousands of car salesmen pitches to figure out

if the buyer relied on a statement or omission made by GM. When the court faces such a choice,

Halliburton requires it to determine whether the presumption applies. 573 U.S. at 281–83. That

means identifying whether the fraud was so pervasive under Minnesota law, or whether GM’s

representations were so baseless under New Jersey law, that the district court may presume

reliance.

We’re skeptical. If different buyers heard different pitches from different dealers about

different cars, they experienced the misrepresentation differently too. That’s a far cry from what

triggered the presumption in Minnesota, where the State sued two universities for disseminating

identical online advertisements to students. Minn. Sch. of Bus., 935 N.W.2d at 129. Nor does it

fit the prerequisites in New Jersey, where a merchant created an “entire marketing scheme” in

which every single proposition was “fictional.” Lee, 4 A.3d at 527.

Florida, Michigan, New Hampshire, New York, Kentucky, and Tennessee stand on the

other side of the line. Their consumer laws reject the common law and are “intended to make it

easier to sue.” See State ex rel. Humphrey v. Alpine Air Prods., Inc., 500 N.W.2d 788, 790

No. 23-1940 Speerly et al. v. General Motors, LLC Page 31

(Minn. 1993). These states thus do not require reliance as part of their consumer statutes. Davis

v. Powertel, Inc., 776 So. 2d 971, 973 (Fla. Dist. Ct. App. 2000); Dix v. Am. Bankers Life

Assurance Co. of Fla., 415 N.W.2d 206, 209–10 (Mich. 1987) (finding sufficient for consumer

to show that a reasonable person would have relied on allegedly fraudulent representations);

Rikos v. Procter & Gamble Co., 799 F.3d 497, 516 (6th Cir. 2015) (finding New Hampshire’s

consumer law to look to Massachusetts’s consumer-fraud statute for guidance, and finding

Massachusetts’s statute not to require reliance or individualized causation); Stutman v. Chem.

Bank, 731 N.E.2d 608, 612–13 (N.Y. 2000); Corder v. Ford Motor Co., 869 F. Supp. 2d 835,

837–38 (W.D. Ky. 2012) (finding the only probative signal on Kentucky law a state appeals case

suggesting that plaintiffs need prove neither causation nor reliance under the Kentucky

Consumer Protection Act); Messer Griesheim Indus., Inc. v. Cryotech of Kingsport, Inc., 131

S.W.3d 457, 469 (Tenn. Ct. App. 2003).

Washington borrows from federal securities law and presumes reliance when a seller

conceals a defect. Morris v. Int’l Yogurt Co., 729 P.2d 33, 40–41 (Wash. 1986) (en banc). But

the class’s theory of the case sounds in both omission and misrepresentation. Its complaint cites

GM’s statements that eight-speed automatic transmissions offer smooth shifting. Its motion to

certify accuses GM of telling customers that poor shifts were “normal” but also of failing to

disclose defects. To the extent the class suggests that GM lied not once but twice—passively

through omission and actively through misrepresentation—the district court must check if buyers

relied on the affirmative statements. Because we don’t know if GM stood “mute in the face of a

duty to disclose” or “made misstatements of fact and then failed to include other facts,”

Tershakovec v. Ford Motor Co., 79 F.4th 1299, 1309–10 (11th Cir. 2023) (quotation omitted),

we cannot know the answer on this record.

The plaintiffs read the caselaw differently. They claim that most of the state claims do

not require reliance. But a rigorous analysis demands that the district court get it right, not that it

get close. Because we treat each subclass “as a class under” Rule 23(c)(5), we cannot certify any

subclass lacking predominance. The subclasses whose claims require reliance cannot take refuge

in the subclasses whose claims don’t.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 32

The plaintiffs suggest that reliance is a question of state law, one suitable for resolution at

summary judgment. Rikos, 799 F.3d at 512. That’s true enough in one sense. But whether

reliance bars the common questions from “predominat[ing] over any questions affecting only

individual members” under Civil Rule 23(b)(3) is a question of federal law that must be resolved

before certification, not at summary judgment or at trial.

We disagree, at any rate, with the premise that Arizona, Kansas, North Carolina, and

Texas permit a plaintiff to show class-wide reliance through uniform material misrepresentations

or omissions. Even as we review the final decision to certify for abuse of discretion, we “owe no

deference to district court adjudications of state law.” Leavitt, 518 U.S. at 145.

Start with the class’s argument that one may presume reliance under Arizona and Kansas

consumer law. The cited district court opinion determined that an omission may be deceptive; it

did not decide whether it caused injury. See, e.g., In re Ariz. Theranos, Inc., Litig., 308 F. Supp.

3d 1026, 1040 (D. Ariz. 2018) (citing State ex rel. Horne v. AutoZone, Inc., 275 P.3d 1278, 1281

(Ariz. 2012) (en banc)). That explains why the plaintiffs in that case also alleged that they were

“harmed by their reasonable reliance on defendants’ omissions.” Id. at 1042. The Kansas case

does not apply either, where each class member relied on the same communication. See, e.g.,

Delcavo v. Tour Res. Consultants, LLC, No. 21-2137-JWL, 2022 WL 1062269, at *9 (D. Kan.

Apr. 8, 2022). Not so here.

The class’s reading of North Carolina and Texas law, in turn, suggests that reliance can

sometimes be established circumstantially. But they haven’t proven that this is one of those

situations. Take North Carolina law as construed in Rowan County Board of Education v.

United States Gypsum Co., 418 S.E.2d 648 (N.C. 1992). It held that a jury could find that an

architect relied on a supplier’s misinformation, planted in a catalogue he called “an architect’s

Bible” upon which all architects “rely.” Id. at 659–61. True, those circumstances show one

buyer’s reliance. And we acknowledge the reality that, if reliance “often” is “an insuperable

barrier to class certification,” that implies it can be superable in rare cases. Dukes, 564 U.S. at

351 n.6. But it does not prove that in this case hundreds of thousands of plaintiffs all saw and

relied on something GM said. So too with Texas. See Henry Schein, 102 S.W.3d at 694.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 33

***

The dissent identifies some instances in which state courts have affirmed class

certification despite defect-manifestation and reliance hurdles. It then urges the district court on

remand to “consider the ways in which many state courts apply a more flexible approach.”

Dissent at 104.

Much as there may be to learn from state law in some settings, we cannot lightly graft 26

States’ class-action rules onto the federal tree. While Oklahoma courts might think that the “fact

that many members of the class have not yet sustained damage does not prevent class

certification” because the court can “conditional[ly]” certify and hash out the merits later, Hess v.

Volkswagen of Am., 221 P.3d 132, 140 (Okla. Civ. App. 2009) (quotation omitted), we cannot,

Fed. R. Civ. P. 23 advisory committee’s note to 2003 amendment. Illinois courts can experiment

with “rules which can reduce the difficulties of showing individual reliance,” Steinberg v. Chi.

Med. Sch., 371 N.E.2d 634, 645 (Ill. 1977), but we cannot, Amgen, 568 U.S. at 462–63.

Arkansas might find a defect-manifestation rule “inapposite” for class actions because it is good

enough that “at least for some plaintiffs” the “existence of damages . . . will depend on” a

common statement, Phillip Morris Cos. v. Miner, 462 S.W.3d 313, 320 n.4, 321 (Ark. 2015), but

it isn’t enough under Rule 23, Bouaphakeo, 557 U.S. at 453. The list could go on and on, and so

could the lesson. Federal law remains supreme, a point not lost on Congress when it

“facilitate[d] adjudication of certain class actions in federal court” by passing the Class Action

Fairness Act. Dart Cherokee Basin Operating Co. v. Owens, 574 U.S. 81, 89 (2014); 28 U.S.C.

§§ 1332(d); 1453(b).

All of this suggests that common questions do not predominate in the state consumer

protection statutes that require reliance or defect-manifestation. We vacate and remand for the

district court to re-assess, according to these principles, whether common questions predominate

and, if so, as to which subclasses.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 34

D. Fraudulent Concealment

The Illinois, Louisiana, New York, Tennessee, and Washington subclasses bring

fraudulent concealment claims. In none of them, it appears, do common questions of defect

predominate over individualized questions of reliance.

By way of a common example, fraud in Illinois consists of the (1) concealment of a

material fact, (2) intent to induce a false belief, (3) reliance by the listener (who could not have

reasonably discovered the truth) upon the speaker’s silence, (4) prospect that the listener would

have acted differently had they known the concealed information, and (5) causation leading to

injury. Schrager v. N. Cmty. Bank, 767 N.E.2d 376, 384 (Ill. App. Ct. 2002).

As with the consumer claims, reliance typically creates “an insuperable barrier to class

certification.” Dukes, 564 U.S. at 351 n.6. And the plaintiffs do not make any affirmative

argument that common questions predominate in common-law fraudulent concealment claims.

Louisiana, New York, and Tennessee all require reliance in their fraud claims too. Greene v.

Gulf Coast Bank, 580 So. 2d 712, 715–16, 719–20 (La. Ct. App. 1991), rev’d on other grounds,

593 So. 2d 630 (La. 1992); Mandarin Trading Ltd. v. Wildenstein, 944 N.E.2d 1104, 1108 (N.Y.

2011); Chrisman v. Hill Home Dev., Inc., 978 S.W.2d 535, 538 (Tenn. 1998).

That leaves Washington, whose fraudulent-concealment tort offers a reliance-free route

to success. There, a plaintiff may show fraudulent concealment without “affirmatively

plead[ing]” reliance if the defendant “breached an affirmative duty to disclose a material fact.”

Crisman v. Crisman, 931 P.2d 163, 166 (Wash. Ct. App. 1997). That offers the class an escape

hatch. They can try to prove that GM had a duty to reveal its transmission problems to every

consumer, in which case reliance would not pose a barrier to class certification. But that also

means GM can contest this duty if it means that the presumption would not apply and a trial

would yield individualized reliance inquiries. Halliburton, 573 U.S. at 281–83.

It is doubtful that a careful analysis of predominance—identifying each state fraudulent-

inducement regime and clarifying that “class claims, issues, or defenses” raised may not include

common-law fraudulent inducement claims—favors certification. Fed. R. Civ. P. 23(c)(1)(B).

No. 23-1940 Speerly et al. v. General Motors, LLC Page 35

As before, the district court should apply these principles and re-assess predominance for these

state subclasses on remand.

***

One last point applicable to this cause of action and the other theories of liability. The

district court should clarify whether problems with predominance as to the class’s liability

theories disrupt the district court’s conclusion that “formulaic calculation[s]” can adequately

estimate damages for each class member. Fox, 67 F.4th at 301 (quotation omitted). At one

point, the district court wrote that the experts plausibly “proffered reliable statistical means for

estimating” the buyers’ overpayment in the context of every claim because every plaintiff will

seek to recover the difference in price between a car without the 8L transmission problems and a

car with them. R.284 at 48–49. That appears to be overinclusive. A “model purporting to serve

as evidence of damages . . . must measure only those damages attributable” to a specific theory

of harm. Comcast Corp. v. Behrend, 569 U.S. 27, 35 (2013). As one example, an overpayment-

based model for a state consumer law claim might overcompensate in the states that require a

defect to manifest. See, e.g., Wallis, 208 S.W.3d at 161. On remand, the district court should

ensure that each damages model accounts for how each claim works.

E. Arbitration

Arbitration bears on this case as a defense to, not an element of, the class’s claims.

Because Civil Rule 23 asks only whether common “questions of law or fact” predominate, it

applies to affirmative elements and the “defenses of the class” alike. Fed. R. Civ. P. 23(a)(3).

That’s why the class-certification order must define the relevant “claims, issues, or defenses,” the

class notice must list the members’ “defenses,” and a certifying court must consider the costs of

controlling “the prosecution or defense of separate actions.” Fed. R. Civ. P. 23(b)(3)(A),

(c)(1)(B), (c)(2)(B).

An arbitration agreement offers a potential defense. It is one reason why the court might

not issue a final judgment for the class “even if all the allegations in the complaint are true.”

Affirmative Defense, Black’s Law Dictionary (12th ed. 2024); see also Fed. R. Civ. P. 8(c)(1)

(denoting arbitration and award as a defense).

No. 23-1940 Speerly et al. v. General Motors, LLC Page 36

Arbitration raises individualized inquiries of law and fact. GM claims that 20 of the 48

customer purchase agreements submitted during discovery contain arbitration provisions. If the

plaintiffs agree that this number reflects the composition of the entire class, those agreements

would require the district court to bifurcate the class into the 42% who must proceed through

arbitration, and the 58% who may proceed through trial.

Different consumers, moreover, signed different contracts. And they appear to have

signed the agreements with the car dealers, not GM. Because arbitration is “a matter of

contract,” First Options of Chi., Inc. v. Kaplan, 514 U.S. 938, 943 (1995), state law determines

when a car manufacturer may benefit from a car dealer’s arbitration agreement. See, e.g.,

AtriCure, Inc. v. Meng, 12 F.4th 516, 526, 530 (6th Cir. 2021) (equitable estoppel and agency

doctrine); Davis v. Nissan N. Am., Inc., 100 Cal. App. 5th 825, 835–37 (Cal. Ct. App. 2024)

(identifying when a car manufacturer may use an arbitration agreement in a contract drafted by

the car dealer); Cooper Indus., LLC v. Pepsi-Cola Metro. Bottling Co., 475 S.W.3d 436, 442

(Tex. App. 2015) (same). That creates an individualized endeavor. The district court must

interpret each arbitration agreement and identify which ones warrant extension to the class’s

claims against GM.

Because the district court has not answered either question, we vacate for further

proceedings over whether questions of arbitrability bar class certification. The record leaves a

mere sample of 20 arbitration agreements, some of which bind the consumer in claims against

“third parties who do not sign this Lease,” R.245-58 at 16, and some of which do not. This

twenty-contract document does not permit us to estimate the number of different arbitration

agreements in this 800,000-car class. The district court should decide in the first instance

whether arbitration questions overwhelm the common ones in this class.

We disagree with what the district court did say—that GM waived arbitration as to the

entire class when it waived arbitration as to the named plaintiffs. Federal courts may not resolve

the rights of “third persons not parties to” suit. Singleton v. Wulff, 428 U.S. 106, 113 (1976)

(plurality op.). A party who proposes a class action may not tie the hands of absent class

members until the court certifies and appoints them the class representative. See Smith v. Bayer

Corp., 564 U.S. 299, 315–16 (2011). That is why pre-certification stipulations do not bind a

No. 23-1940 Speerly et al. v. General Motors, LLC Page 37

class. Standard Fire Ins. v. Knowles, 568 U.S. 588, 593 (2013). A person who makes a

stipulation makes an “express waiver” that he concedes some fact. 9 John Henry Wigmore, A

Treatise on the System of Evidence in Trials at Common Law § 2588, at 821 (J. Chadbourn rev.

1981). But that precertification stipulation “does not bind anyone but himself.” Knowles, 568

U.S. at 593. For the same reason, waiver, or GM’s “voluntary relinquishment” of the right to

resolve the named plaintiffs’ claims by trial, Waiver, Black’s Law Dictionary (12th ed. 2024),

does not speak for the unnamed class members.

We vacate the order of the district court certifying the class and remand for further

proceedings consistent with this opinion.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 38

_________________

CONCURRENCE

_________________

THAPAR, Circuit Judge, concurring. I join the majority’s excellent opinion in full.

I write separately to address an undertheorized area of law: standing in class actions.

This case presents two issues related to standing in class actions. First, who must suffer

an injury in fact? Everyone agrees that the class representatives must.

Second, what happens if one of the aspiring class representatives seeks to certify a class

when he’s suffered a harm different than some of the potential class? That is true here. The

named plaintiffs’ cars experienced a shudder or hard shift while driving, but some of the cars of

the proposed class did not. Instead, the named plaintiffs alleged that the threat of the shudder or

hard shift makes these class members’ cars worth less. This disparity between the named

plaintiffs’ injuries and those of some prospective class members is called the “disjuncture

problem.”

Courts have struggled with the disjuncture problem for decades. Some courts have

applied Article III standing principles to determine whether a plaintiff may sue on behalf of class

members who suffered different injuries or who are perhaps not injured at all, in the Article III

sense. But other courts have said that a “disjuncture” doesn’t eliminate a court’s ability to hear a

given case. Rather, it raises a question best addressed during Rule 23’s certification inquiry.

The latter view is the better one. Courts should look to Rule 23, not Article III, in

determining whether a named plaintiff may represent a class of members who assert different

injuries—or no injury at all.

I.

The issues in this case are easier to think about in bite-sized chunks. Here there are two

issues: standing and Rule 23. Addressing these concepts separately reveals how each functions

and ultimately works together.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 39

Start with standing. Anytime a plaintiff tries to sue in federal court, he must be able to

answer the question, “What’s it to you?” Antonin Scalia, The Doctrine of Standing as an

Essential Element of the Separation of Powers, 17 Suffolk U. L. Rev. 881, 882 (1983). If a

plaintiff can’t answer that question, then he doesn’t have standing.

In many disputes, it’s easy to determine whether a plaintiff has standing. A party to a

contract can say his counterpart breached the terms of a deal and thus caused him harm. Or an

assault victim can say a tortfeasor harmed him. Or, relevant here, a plaintiff who bought a

defective product can say that he was injured when that product stopped working as it should.

But in the class action context, the inquiry is murky. Does only the class representative

need to have standing? What about the unnamed class members? If unnamed members don’t

have standing, can the case proceed? These questions have troubled courts and commentators

for years.

In answering these questions, courts must focus on the distinction between Article III

standing and the requirements of Rule 23. For one, they “spring from different sources.”

1 W. Rubenstein, Newberg on Class Actions § 2:6 (6th ed. 2025). Standing doctrine comes from

Article III of the Constitution, which confines the jurisdiction of federal courts to “Cases” and

“Controversies.” To make out a case or controversy, a plaintiff must have a “personal stake” in

any dispute that he seeks to litigate. TransUnion LLC v. Ramirez, 594 U.S. 413, 423 (2021).

In early analogues to class actions called bills of peace, the original plaintiff (the

representative party) had to establish his personal stake at the outset of the proceeding. Mayor of

York v. Pilkington (1737) 25 Eng. Rep. 946, 947 (Ch.). And, if other parties (the unnamed

parties) were involved on the plaintiff’s side, they didn’t have to prove their entitlement to relief

until later in the proceeding. Cf. Story, Commentaries on Equity Pleadings § 99, at 99 (2d ed.

1840). The idea was that joining such parties was impracticable, so the original plaintiff wasn’t

punished for not doing so at the start (he could always join them later). Id. § 72, at 74‒78. And,

as this court explained, “[n]o one doubted that bills of peace created justiciable cases.” Fox v.

Saginaw County, 67 F.4th 284, 299 (6th Cir. 2023); cf. Smith v. Swormstedt, 57 U.S. (16 How.)

288, 298‒303 (1853). Why? One reason is that while unnamed parties could participate, they

No. 23-1940 Speerly et al. v. General Motors, LLC Page 40

weren’t entitled to relief until the representative party established the unnamed parties’ stake in

the proceeding. Story, Commentaries, § 99. So courts weren’t awarding judgment to uninjured

parties.

It’s true that the original plaintiff in a bill of peace often had to have the same kind of

claim as the absent parties. Cf. Smith, 57 U.S. (16 How.) at 298. This requirement sometimes

arose when one of several creditors wanted to collect from a deceased debtor. Story,

Commentaries, § 99. So a particular creditor could bring a representative suit on behalf of all

creditors, since they all had the same “interest” in getting paid. Id. But this common interest

requirement doesn’t change the jurisdictional analysis—instead, it was a prudential protection to

protect unnamed parties. How? By making sure the named plaintiff wouldn’t resolve the rights

of other absent parties in ways that unfairly harmed them. Id. In the creditor example, the

common claim or interest requirement ensured that one creditor wouldn’t unfairly prejudice the

others by taking too much money. Id. This determination—the adequacy of one plaintiff

representing the unnamed parties’ interests—resembles the analysis courts conduct under Rule

23. After all, Rule 23 directs courts to consider, among other things, whether there are common

questions and whether a named party can adequately represent the unnamed plaintiffs.

This history reveals that Article III isn’t a restraint on mass representation; in the class

action context, it’s a restraint on giving relief to uninjured parties. 1 Indeed, from the early days

of the Republic, courts explained that their jurisdiction—the “power to hear and determine the

subject matter in controversy”—meant the power “to render a judgment or decree upon the rights

of the litigant parties.” Rhode Island v. Massachusetts, 37 U.S. (12 Pet.) 657, 718 (1838); cf. 2

The Records of the Federal Convention of 1787, 430 (M. Farrand ed. 1966) (James Madison

stating that courts could only take actions that were of “a Judiciary Nature”). And in Mills v.

Duryee, Justice Story analogized between jurisdiction and judgment: If a court had jurisdiction

1It’s true that a court also needs jurisdiction before it can declare the law.

See Ex parte McCardle, 74 U.S.

(7 Wall.) 506, 514 (1868). But in class actions, once the court has jurisdiction over the named parties, it may

declare the law as to those parties. Then, once the case proceeds to the judgment stage, the unnamed parties must

demonstrate they have standing, too. Cf. Murthy v. Missouri, 603 U.S. 43, 58 (2024) (emphasizing that any party

invoking the jurisdiction of a federal court “must support each element of standing ‘with the manner and degree of

evidence required at the successive stages of the litigation.’” (citation omitted)); Biden v. Nebraska, 600 U.S. 477,

489 (2023) (explaining that if at least one plaintiff has standing, the suit may proceed).

No. 23-1940 Speerly et al. v. General Motors, LLC Page 41

over the parties, the judgment could be valid. 11 U.S. (7 Cranch) 481, 484‒85 (1813). If not, the

judgment wasn’t legitimate. See id. Courts have followed this framework in the centuries since.

Cf. Tyson Foods, Inc. v. Bouaphakeo, 577 U.S. 442, 466 (2016) (Roberts, C.J., concurring)

(explaining that “Article III does not give federal courts the power to order relief to any

uninjured plaintiff, class action or not” (emphasis added)).

The fact that some jurisdictional analysis now happens to fall under the label “standing

doctrine” doesn’t change that principle. After all, standing is a “jurisdictional doctrine.” Acheson

Hotels, LLC v. Laufer, 601 U.S. 1, 8 (2023) (Thomas, J., concurring). It exists to ensure courts

can decide disputes historically resolved through the judicial process, while forbidding courts

from exercising power over people who never fell within the sweep of a court’s authority. Susan

B. Anthony List v. Driehaus, 573 U.S. 149, 158 (2014). In other words, it protects core

jurisdictional power, as it was historically defined. See Rhode Island, 37 U.S. at 718. And, as

the history of the bills of peace shows, mass representation didn’t offend jurisdiction. In a

proceeding brought under a bill of peace, if there was a valid controversy between one plaintiff

and a defendant, courts could add unnamed parties. And that didn’t create a jurisdictional

problem until the court entered judgment for someone without a legitimate interest in the

dispute.2

All told, history tells us that (1) a named plaintiff must have a personal stake in a dispute

from the beginning, and (2) unnamed parties can be a part of a suit without establishing their

entitlement to relief at the outset, but those parties must do so before a court may order relief on

their behalf.

While standing doctrine governs who can bring a suit and what relief courts can grant,

Rule 23 governs what happens during a suit. Like all the Federal Rules of Civil Procedure, Rule

23 is a procedural tool regulating how cases flow through courts. It was designed to organize

disparate procedural mechanisms into a “coherent set of statements which would govern the

2This historical inquiry is relevant because “history and tradition offer a meaningful guide to the types of

cases that Article III empowers federal courts to consider.” Sprint Commc’ns Co. v. APCC Servs., Inc., 554 U.S.

269, 274 (2008). If certain types of cases fell within a court’s traditional powers, courts shouldn’t shut their doors to

those cases today. Cf. TransUnion, 594 U.S. at 423–25.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 42

conduct of all civil litigation in the federal courts.” John G. Harkins, Jr., Federal Rule 23—The

Early Years, 39 Ariz. L. Rev. 705, 705 (1997). Rule 23, after all, stems from a prudential

judgment that it’s better to handle some cases en masse than in discrete chunks. Cf. Sosna v.

Iowa, 419 U.S. 393, 403 (1975).

Rule 23’s history confirms this purpose. When older procedural tools struggled with

joining large numbers of claims, the body responsible for suggesting new rules sought to replace

the old rule with a more flexible version. David Marcus, Response: Making Adequacy More

Adequate, 88 Tex. L. Rev. 137, 146 n.46 (2009). So it came up with Rule 23. This change was a

watershed: Opponents criticized Rule 23 classes as a “form of ‘legalized blackmail’” that could

force defendants to the table. Arthur R. Miller, Of Frankenstein Monsters and Shining Knights:

Myth, Reality, and the “Class Action Problem”, 92 Harv. L. Rev. 664, 665 (1979) (citation

omitted). But proponents saw a tool that would allow for the joinder of many parties, thereby

creating a “panacea for a myriad of social ills.” Id.; cf. Brian T. Fitzpatrick, The Conservative

Case for Class Actions 9 (2019). The proponents won out, creating the Rule 23 courts know

today. If plaintiffs show numerosity, typicality, commonality, and adequacy, their claims may be

best heard together (assuming they meet Rule 23’s other requirements). If not, the claims should

stand alone.

All told, looking at these requirements one-by-one reveals some simple principles about

how they fit together. A named plaintiff, like any plaintiff, must allege an injury in fact if he

wants to get into court. The fact that a case is a class action doesn’t change that result. After all,

that a “suit may be a class action . . . adds nothing to the question of standing.” Simon v. E. Ky.

Welfare Rts. Org., 426 U.S. 26, 40 n. 20 (1976). A joinder rule like Rule 23 can’t create

standing when no standing exists. 1 W. Rubenstein, Newberg on Class Actions § 2:5 (6th ed.

2025). Instead, Rule 23 operates as a guidepost for whether it’s efficient to bring a particular suit

en masse.

Thus, at the beginning of the suit, courts should consider the same question that they

always do: has the plaintiff, the would-be class representative, alleged an injury in fact? If so,

the plaintiff has provided a satisfactory answer to Justice Scalia’s famous question: “What’s it to

you?” Scalia, supra, at 882. A case or controversy exists.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 43

II.

Here, there’s no question that the named plaintiffs have standing. The named plaintiffs

had a car that experienced a shudder and hard shift, and thus, overpaid for a defective product.

Id. That combination is sufficient to create an injury in fact. Id.

But what happens when the named representatives experience an injury in fact, while

some of the potential class members may have experienced a different injury or may not have

experienced an injury at all? This situation is where the “disjuncture problem” kicks in. The

name comes from the gap—the disjuncture—between what the named plaintiff says he suffered

and what the class plaintiffs may have experienced. Any court considering this difficult issue

faces a confusing situation. On the one hand, plaintiffs can cite Supreme Court and circuit court

caselaw saying that only the named plaintiff needs standing. Thus, courts often certify classes

when the named plaintiff suffered a different injury than some of the potential class members.

On the other hand, defendants can cite caselaw saying that the class should not even proceed to

the certification stage because the named plaintiff doesn’t have standing to sue on behalf of other

litigants with different harms. So what’s a court to do?

A.

Courts facing a disjuncture have adopted two conflicting approaches. Some courts

simply find that the class representative lacks standing to pursue the class members’ claims

because he did not suffer their potential injuries, if any. See, e.g., Prado-Steiman ex rel. Prado v.

Bush, 221 F.3d 1266 (11th Cir. 2000); cf. Barrows v. Becerra, 24 F.4th 116, 129 (2d Cir. 2022).

Other courts will say that the class representative has standing to pursue his own claims, and then

discuss any disjuncture as an issue for class certification. See, e.g., Fallick v. Nationwide Mut.

Ins. Co., 162 F.3d 410, 424 (6th Cir. 1998). This first approach is called the “standing

approach,” and the second is the “class certification approach.” 1 W. Rubenstein, Newberg on

Class Actions § 2:6 (6th ed. 2025).

These doctrines are not well-defined. Both approaches have some grounding in Supreme

Court dicta and circuit court caselaw. That said, courts often pursue one of these approaches

without discussing or acknowledging the other’s existence.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 44

1.

Start with the standing approach. Under the standing approach, a named plaintiff may

litigate only on behalf of those whose harms are a close match for the harms he suffered.

This framework dates to a Supreme Court case called Blum v. Yaretsky, 457 U.S. 991

(1982). There, the Supreme Court considered a suit brought by nursing home residents. Id. at

993–95. The named plaintiffs claimed that they hadn’t been given adequate notice of decisions

to transfer them to a lower level of care. Id. But they also wanted to represent unnamed

plaintiffs, some of whom were transferred to a higher level of care. Id. at 1001. The Supreme

Court explained that the named plaintiffs didn’t have standing to bring the claims on behalf of

the higher-level transfers. Why? A plaintiff “who has been subject to injurious conduct of one

kind” does not “possess by virtue of that injury the necessary stake in litigating conduct of

another kind, although similar, to which he has not been subject.” Id. at 999 (citation omitted).

In holding that the plaintiffs lacked standing to bring these representative claims, the Court said

that the “conditions under which such transfers occur are sufficiently different from those which

[the plaintiffs] do have standing to challenge that any judicial assessment of their procedural

adequacy would by wholly gratuitous and advisory.” Id. at 1001. The Blum Court focused on

the “kind” of injury and whether that injury placed the potential representative “within the class

of persons who will be concretely affected.” Id. at 999.

This approach is no relic of the 1980s. The Court has revisited the standing approach in

more recent years, starting with Lewis v. Casey, 518 U.S. 343, 346 (1996). In Lewis, the Court

explained that the “actual-injury requirement would hardly serve [its] purpose . . . if once a

plaintiff demonstrated harm from one particular [harm], the court were authorized to remedy all

[similar harms].” Id. at 357 (emphasis in original).

Applying this logic to the facts here, a court would consider the harm the named plaintiffs

allege (manifest defect leading to their cars’ diminished value) and see whether it matched what

some of the would-be unnamed plaintiffs claimed (a defective car with diminished value). These

harms are similar, but they’re not identical.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 45

Thus, there’s a key wrinkle. How close of a match must the named plaintiff’s injury be

to the unnamed class members’ injuries to make out an Article III case or controversy? Is it a

mirror image? Or something less?

Unsurprisingly, courts splinter when faced with that question. Some courts take a strict

approach, explaining that the named plaintiff must have exactly the same “set of concerns” as the

potential unnamed members. See, e.g., Barrows, 24 F.4th at 129 (quotation omitted); Fox v.

Ritz-Carlton Hotel Co., 977 F.3d 1039, 1046 (11th Cir. 2020). Other courts adopt a more

flexible approach, reading Lewis to require only a general “similarity of injury between the

named plaintiffs and passive class members.” 1 W. Rubenstein, Newberg on Class Actions § 2:6

(6th ed. 2025); see, e.g., Armstrong v. Davis, 275 F.3d 849, 867 (9th Cir. 2001), abrogated on

other grounds by Johnson v. California, 543 U.S. 499, 504–05 (2005). Courts applying this

latter view are “careful not to employ too narrow or technical an approach. Rather, [they] must

examine the questions realistically . . . [and] reject the temptation to parse too finely, and

consider instead the context of the inquiry.” Armstrong, 275 F.3d at 867.

Applying those various tests here yields mixed results. Under a strict framework, the

same standing analysis outlined above applies: The harms between the would-be named

plaintiffs and potential unnamed members aren’t the same, because the named plaintiffs’ cars

manifested the defect, but some of the unnamed plaintiffs’ cars didn’t. Thus, under this

framework, Article III bars the suit from proceeding in its current form. But under the more

flexible test, perhaps Article III doesn’t stand in the way of this suit, as a court could find that

both the named and unnamed members suffered a similar injury because their cars all had

diminished value.

No matter the approach, these tests fall short. Both tests, whether they use flexible

language like “parse too finely” and “context,” or strict language like “same set of concerns,” are

vague. Id. These abstract standards make for “a terribly confusing way to decide something as

essential as the . . . jurisdiction of the court.” Barry v. O’Grady, 895 F.3d 440, 448 (6th Cir.

2018) (Sutton, J., dissenting). What’s more, it’s not clear what either concern has anything to do

with Article III’s case or controversy requirement. After all, the named plaintiff has suffered an

No. 23-1940 Speerly et al. v. General Motors, LLC Page 46

injury, even if it’s not identical to the unnamed would-be plaintiffs’ injuries. Thus, there is a

case or controversy between the named plaintiff and the defendant.

All told, while lacking a robust theoretical basis, some circuits read the Supreme Court’s

caselaw as suggesting that the standing inquiry requires an examination of how the harms that

the named plaintiff suffered compare to the harms that the unnamed class members suffered.

This framework is thus called the standing approach.

2.

Next, consider the class certification approach. This framework says that once a named

plaintiff shows he has standing to bring a claim, the standing inquiry is at an end, and the court

should next examine if the named plaintiff can show that the proposed class meets the Rule 23

prerequisites. Like the standing approach, this framework asks courts to compare the harms of

the named representative and the would-be unnamed class members. But it does so through the

lens of Rule 23, not justiciability.

As with the standing approach, the class certification approach is grounded in Supreme

Court caselaw. In Sosna, the Supreme Court considered a named plaintiff who was denied a

divorce after failing to live in Iowa for one year (the minimum period required to get a divorce in

the state). 419 U.S. at 395. The Court explained that standing “does not automatically establish

that [a named plaintiff] is entitled to litigate the interests of the class she seeks to represent.” Id.

at 403. Rather, concluding that the named plaintiff has standing “shift[s] the focus of

examination from the elements of justiciability to the ability of the named representative to

‘fairly and adequately protect the interests of the class.’” Id. (quoting Fed. R. Civ. P. 23(a)).

Thus, Sosna indicates that the Article III standing analysis, like any other justiciability inquiry,

precedes questions of class certification under Rule 23. According to Sosna, a court considering

a class with a potential disjuncture should first look at whether the individual named plaintiff has

standing and then focus on how that individual’s injury relates to the class at issue through the

lens of Rule 23. The Court followed a similar path in subsequent cases. See, e.g., Gen. Tel. Co.

of Sw. v. Falcon, 457 U.S. 147, 157–60 (1982).

No. 23-1940 Speerly et al. v. General Motors, LLC Page 47

Applying that logic here, we’d look to whether the named plaintiff has standing. And

here, as the majority opinion correctly notes, he does. Why? He alleged a diminished value

injury when his car manifested a shudder or hard shift. Next, we’d look to whether he is suited

to represent the class, running through Rule 23’s familiar requirements. As the majority opinion

explains, the class at issue would fail—not because there’s no case or controversy, but because

the plaintiff’s proposed class lacks some of these necessary characteristics.

Five circuits, including ours, follow this approach. See, e.g., Fallick, 162 F.3d at 424.3

*

In sum, there are two plausible ways to resolve classes that feature a disjuncture between

the harms the named plaintiff says he suffered and the harms that some of the members of the

prospective class may have suffered. Both approaches are grounded in Supreme Court precedent

and various circuits have adopted each of the two tests.

B.

Reasoning from first principles, the class certification approach is superior to the standing

framework. As outlined above, Article III standing is a constitutional doctrine that asks whether

a definite case or controversy exists. Lujan v. Defs. of Wildlife, 504 U.S. 555, 560–61 (1992).

Once the named plaintiffs establish they have standing, there is a live case or controversy.

As to the other prospective class members’ claims, a court must decide whether the

would-be unnamed plaintiffs’ injures are similar to the named plaintiff’s valid case or

controversy. That’s a textbook prudential determination. And such determinations are the

touchstone of Rule 23, which gives courts a framework to consider when parties can bring class

actions in the first place. Thus, Rule 23 is the correct tool to use in evaluating the relationship

between the named plaintiff and the unnamed parties in a particular lawsuit. See Lab’y Corp. of

Am. Holdings v. Davis, __ S. Ct. __, 2025 WL 1583302, at *5–6 (U.S. June 5, 2025)

(Kavanaugh, J., dissenting) (applying Rule 23’s predominance requirement to classes with

3In re Asacol Antitrust Litig., 907 F.3d 42, 49 (1st Cir. 2018); Boley v. Universal Health Servs., Inc., 36

F.4th 124, 133 (3d Cir. 2022); Chavez v. Plan Benefit Servs., Inc., 108 F.4th 297, 312 (5th Cir. 2024); B.K. ex rel.

Tinsley v. Snyder, 922 F.3d 957, 967 (9th Cir. 2019).

No. 23-1940 Speerly et al. v. General Motors, LLC Page 48

injured and uninjured members). Such a determination is not a constitutional question. Instead,

it revolves around how a given plaintiff relates to others.

The class certification approach aligns with the broader framework of federal civil

procedure. Rule 23 is a device for aggregating claims. It has nothing to say about jurisdiction.

Thus, satisfying Rule 23 doesn’t create standing when standing doesn’t exist. Simon, 426 U.S. at

40 n.20. And failing to satisfy Rule 23 doesn’t destroy a valid claim when such a claim exists.

Indeed, this principle governs federal court practice throughout the federal rules. Cf. Charles

Alan Wright, Arthur R. Miller, Mary Kay Kane, & Howard M. Erichson, Federal Practice and

Procedure § 1681 (3d ed. 2024) (hereinafter “Wright & Miller”). In various contexts, if a

plaintiff joins a non-justiciable claim with a justiciable claim, or if a party with standing joins a

party without standing, the court doesn’t dismiss the action for want of jurisdiction. See, e.g.,

Fed. R. Civ. P. 13, 14, 18, 21; Frank v. Gaos, 586 U.S. 485, 493 (2019) (explaining that what

matters is whether any named plaintiff has standing). Instead, it applies the relevant joinder rule

and dismisses the improperly joined claim or the party without standing. The proper party then

proceeds in the normal course. See Fed. R. Civ. P. 21.

On the other hand, adopting the standing approach would return courts to an era when

misjoinder ended an action. See Benjamin J. Shipman, Handbook of Common Law Pleading

§§ 226–28 (3d ed. 1923). Indeed, at common law, the “perils of misjoinder” were “like the

perils of war and contagious disease.” Edson R. Sunderland, Joinder of Actions, 18 Mich. L.

Rev. 571, 575 (1920). Why? They were deadly, ending actions immediately. Id.

But it would be a mistake to return to that rule out of a desire to protect Article III’s case

or controversy requirement. Why? The harsh common law approach did not safeguard courts

from hearing matters that sat outside Article III. Instead, the rule prevented courts from hearing

cases that they actually should have heard. This rule resulted from two historic procedural quirks

that courts gradually abandoned.

First, there were common law causes of action (like assumpsit and trover), each of which

demanded its own form of process. Id. at 574‒75. And common law jurists didn’t want to allow

No. 23-1940 Speerly et al. v. General Motors, LLC Page 49

one action—with a certain form of process—to collapse into another action, with a separate form

of process. Id. So courts punished parties who joined one action to another.

Second, courts used to issue different judgments (judgments quod sit in misericordia and

judgments quod capiatur) for different causes of action. Id. at 578. And they didn’t want to

allow a cause of action that merited one kind of judgment to join with a cause of action that

deserved another. Id. All told, these procedural matters became irrelevant when the civil actions

and the judgments we know today replaced the historic causes of action and judgments. Id. So,

as the nineteenth century progressed, equity courts developed rules to allow an “action [to]

proceed on its merits despite an initial misjoinder or nonjoinder whenever the error could be

corrected without adversely affecting the parties to the action.” Wright & Miller, supra § 1681.

When equity merged into law, the Federal Rules of Civil Procedure adopted this

forgiving approach to misjoinder. See id.; Fed. R. Civ. P. 21 advisory committee’s note. And

when the advisory committee adopted Federal Rule of Civil Procedure 23, it incorporated this

equitable principle, noting “an order that grants or denies class certification may be altered or

amended before final judgment.” Fed. R. Civ. P. 23(c)(1)(C). What’s more, Rule 23 is not only

“derived from” equity—it's an expansion of English equitable rules. Wright & Miller, supra

1681. How? At equity, parties had to apply to the court to amend a class. Fed. R. Eq. 43 (1912)

reprinted in The New Federal Equity Rules Promulgated by the United States Supreme Court

172 (1912). And Rule 23 doesn’t require such an application. Thus, a class need not be

dismissed just because there’s misjoinder.

This principle holds true even when a claim with standing is joined to claims without

standing, thereby creating a disjuncture problem. Again, courts facing a class with both injured

and potentially uninjured parties are making a determination about joinder. As far back as 1952,

the Supreme Court wrote that the proper solution to cases when a party without standing was

accidentally joined to a party with standing was to use Rule 21—which provides that parties who

are misjoined should be removed from the case. Mullaney v. Anderson, 342 U.S. 415, 417

(1952); Riley T. Keenan, Minimal Justiciability, 109 Minn. L. Rev. 1, 30 (2025). And under

Rule 23, federal courts can remove claims from absent class members who don’t have standing

No. 23-1940 Speerly et al. v. General Motors, LLC Page 50

and allow the justiciable claims to proceed, so long as Rule 23’s other requirements are satisfied.

Cf. 1 W. Rubenstein, Newberg on Class Actions §§ 2:2; 2:8 (6th ed. 2025).

That makes sense. Otherwise, courts would have to adopt an eggshell class principle,

under which the presence of a single unnamed plaintiff without standing (or with a different

harm than the class representative) would defeat a class. After all, standing is a constitutional

requirement. So if courts determine that a class is defeated when plaintiffs don’t have standing,

the presence of even one such unnamed plaintiff would mean the court can’t certify a class. But

that can’t be right. The Supreme Court has repeatedly explained that “standing is not dispensed

in gross.” Lewis, 518 U.S. at 358 n.6; Wright & Miller, supra § 3531.16. Thus, what matters is

not the standing of an entire class but the standing of the individual members in a given class.

Whether a given member has standing implicates Article III. Whether the class should exist

hinges on Rule 23: numerosity, typicality, commonality, and adequacy.

This distinction coheres with the Supreme Court’s recent standing caselaw. In

TransUnion, the Supreme Court held that “[e]very class member must have Article III standing

in order to recover individual damages.” 594 U.S. at 431 (emphasis added). That conclusion

necessarily follows from the fact that Article III doesn’t give courts the power to “order relief to

any uninjured plaintiffs, class action or not.” Id. (quotation omitted) (emphasis added).

TransUnion teaches that when courts exercise their judicial power by entering judgment, they

can’t award damages to uninjured parties. Thus, current doctrine already prevents uninjured

parties from recovering damages.

All told, there are several reasons to think the class certification approach is preferable to

the standing approach.

III.

So how should a court confronted with a disjuncture—in which the named plaintiff has

standing but members of the class don’t—proceed? In brief, it should do what courts always do.

If the named plaintiff is seeking to certify a class where none of the would-be unnamed parties

has standing, the court should apply Rule 23 as an analogue to traditional joinder rules, which

say such people shouldn’t join the suit.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 51

But what happens when the potential class merely could contain individuals who don’t

have standing to sue? This situation doesn’t change the analysis because courts should still

conduct a standard Rule 23 predominance inquiry. After all, Rule 23 governs which parties

courts can join in class actions. And it requires looking at whether removing non-justiciable

claims and class members involve individual questions, such as whether a particular buyer’s car

really was defective. See, e.g., Fed. R. Civ. P. 23(b)(3). If so, then the court faces headwinds if

it wants to certify the class because the individual assessments would predominate over the

common questions. But if there aren’t individual inquiries, the class should proceed. Why?

Assuming the class meets the other Rule 23 requirements, like numerosity, commonality, and

adequacy, and there’s a predominant question of law or fact, Rule 23 is satisfied, too. Of course,

if at any point the district court determines that certain plaintiffs don’t have standing, the district

court must dismiss them and can’t award them relief.

One response is that this approach gives Rule 23 classes a great deal of power. If a court

were to certify a class that has uninjured members (i.e., where the only determination that must

be made is whether the person purchased the product), the logic goes, it could force corporations

to the table and lead to settlements. That’s true. And it’s a large part of why people objected

when Rule 23 became part of the rules governing civil procedure. Cf. Miller, supra, at 665. But

the solution to that pragmatic concern isn’t to “twist[] the law by the tail” by saying there’s a

constitutional issue, namely no Article III case or controversy. Am. Newspaper Publishers Ass’n

v. NLRB, 345 U.S. 100, 113 (Clark, J., dissenting). Instead, courts should either accept that Rule

23 is a powerful and consequential addition to our civil procedure framework or change the rule.

Cf. 28 U.S.C. § 2072. Until then, judges must apply the rules as written.

What’s more, while Rule 23 classes do have power, their might isn’t an automatic burden

on corporations or society. Why? Rule 23’s framers recognized the potential for abuse and

provided a protective mechanism: A district court’s decision certifying a class is immediately

appealable. Fed. Rule Civ. P. 23(f). Indeed, the defendants used that very mechanism to defeat

class certification here. Future parties can follow their lead when district courts go too far and

don’t conduct the rigorous analysis necessary to ensure class certification is proper.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 52

*

Put Rule 23 aside for a moment. Imagine this were a case in which four friends, who

each bought the same truck, sued after they had a transmission problem. A fifth friend then

shows up to court to get in on the action. What happens? The court looks to joinder rules to see

if it can let the fifth person into the suit. Under those rules, it’d be a mistake to join the

newcomer if the court knew he didn’t have standing. But if he had standing, the court could join

him to the case. And, if it later became clear the friend didn’t really have standing, the district

court should remove him from the suit but let the original four continue.

The principle doesn’t change just because the suit gets bigger. If there are four thousand

buyers instead of four friends, a court would still look to joinder rules (once it determines the

named plaintiffs have standing). And this time, the court would use Rule 23’s joinder principles

because the case is so large. Just as in a normal case, the court can’t join a bunch of consumers

who it knows don’t have standing, as that violates traditional joinder rules. But if the group were

a mixed bag of people as it almost always is, the court would simply do what any court does

when determining whether to certify a class action: apply Rule 23. That means looking at

whether there are common questions of law or fact and what predominates—individual

questions, like whether the buyer’s particular car manifested a defect, or a common question. If

the individual concerns predominate, the court can’t join the parties and thus can’t certify the

class. But if the common question predominates, the court can proceed with certification. At

bottom, the history of jurisdiction, bills of peace, Rule 23, and joinder all point towards this

result.

In conclusion, when the named parties have standing, there’s a live case or controversy.

If other parties want to join in, the court considers whether joinder rules allow their participation.

Rule 23 simply sets out how joinder works in the class action context. Since the majority

opinion correctly applies Rule 23, I concur.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 53

_________________

CONCURRENCE

_________________

NALBANDIAN, Circuit Judge, concurring. I join the majority opinion in full, but I write

separately to address two points. First, I believe this court should have resolved the standing

question by holding that a class cannot be certified with members who have not suffered an

injury in fact. Second, a benefit-of-the-bargain theory of injury is inherently speculative and

insufficient to permit Article III standing for class members who never experienced either

alleged defect. So I would hold that the district court erred in certifying a class in which many

plaintiffs lack standing.

I.

Begin with whether this court can—or indeed must—resolve the standing question in this

appeal. As the majority notes, the unnamed plaintiffs in a putative class are not parties to the suit

before certification. And so the district court “need not worry about their standing until it

certifies the class.” Maj. Op. at 7 (quoting Fox v. Saginaw County, 67 F.4th 284, 296 (6th Cir.

2023)). And yet that is exactly what happened here. The district court first certified the

proposed classes, but then reserved the standing question as to the unnamed plaintiffs, holding

that “the appropriate opportunity to address claims of absent class members whose vehicles

never have manifested any defect is a Rule 56 motion for summary judgment.” Speerly v. Gen.

Motors, LLC, 343 F.R.D. 493, 522 (E.D. Mich. 2023). The court then noted that at the

certification stage it was enough “that the plaintiffs have alleged that every class member

suffered a loss due to overpaying for defective vehicles.” Id. at 523.

It’s true that, despite holding that all unnamed plaintiffs must have standing to recover

individual damages under a class judgment, the Supreme Court has so far declined to resolve

whether they also need standing at the class certification stage. See TransUnion LLC v. Ramirez,

594 U.S. 413, 408 n.4 (2021). And yet leaving this question unanswered imposes real costs, as

“the consequences of overbroad and incorrectly certified damages class actions can be

widespread and significant.” Lab’y Corp. of Am. Holdings v. Davis, 145 S. Ct. 1608, 1612

No. 23-1940 Speerly et al. v. General Motors, LLC Page 54

(2025) (mem.) (Kavanaugh, J., dissenting). So the Supreme Court’s failure to find the perfect

vehicle to resolve class-action standing doesn’t detract from the question’s importance. Nor does

it prevent us from resolving this issue when it’s squarely presented to us.1 See Mr. Dee’s Inc. v.

Inmar, Inc., 127 F.4th 925, 934 (4th Cir. 2025) (“Whatever the resolution of the question posed

in Laboratory Corp., the presence of 32% of uninjured members in a proposed class strikes us as

much too high”). In any event, this court has an obligation to provide guidance to the district

court on remand. See United States v. Campbell, 168 F.3d 263, 268 (6th Cir. 1999). With the

benefit of extensive briefing, the en banc court is well situated to discharge this obligation. And

so I would address the question and hold that a district court cannot certify a putative class when

a significant number of unnamed plaintiffs lack standing.

The district court’s contrary holding was—in a word—error. Any party invoking the

jurisdiction of a federal court “must support each element of standing ‘with the manner and

degree of evidence required at the successive stages of the litigation.’” Murthy v. Missouri, 603

U.S. 43, 58 (2024) (quoting Lujan v. Defs. of Wildlife, 504 U.S. 555, 561 (1992)). And while

unnamed plaintiffs are not parties subject to the court’s jurisdiction before class certification,

they are after. How else could the court issue a binding judgment as to their claims? As Justice

Scalia noted, the class action form, like “traditional joinder (of which it is a species), merely

enables a federal court to adjudicate claims of multiple parties at once, instead of in separate

suits.” Shady Grove Orthopedic Assocs., P.A. v. Allstate Ins., 559 U.S. 393, 408 (2010)

(plurality). And since it is only a tool for aggregating claims, the class action “leaves the parties’

legal rights and duties intact and the rules of decision unchanged.” Id. Because Rule 23 didn’t

alter the fundamental nature of our judicial system, class action requirements “must be

interpreted in keeping with Article III constraints.” Amchem Prods., Inc. v. Windsor, 521 U.S.

591, 613 (1997). For that reason, the district court cannot defer the question of standing for these

unnamed plaintiffs until summary judgment.

1The Supreme Court has stated that the resolution of class-certification issues is “logically antecedent to the

existence of any Article III issues.” Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 612 (1997). As a result, issues

of class certification “may properly be treated before Article III standing.” See Ortiz v. Fibreboard Corp., 527 U.S.

815, 831 (1999). This proposition, though, is prudential and does not preclude the reviewing court from reaching

the substance of a properly presented standing challenge. So we need not blind ourselves to a constitutional error

even if there are alternative grounds for reversing certification.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 55

At the core of standing doctrine lies the separation of powers. The separation of powers

“was not simply an abstract generalization in the minds of the Framers: it was woven into the

document that they drafted in Philadelphia in the summer of 1787.” TransUnion, 594 U.S. at

422–23 (quoting INS v. Chadha, 462 U.S. 919, 946 (1983)). That the judicial power is limited to

specific cases and controversies “serves to prevent the judicial process from being used to usurp

the powers of the political branches.” Clapper v. Amnesty Int’l USA, 568 U.S. 398, 408 (2013).

“Relaxation of standing requirements is directly related to the expansion of judicial power,” such

that even small deviations from Article III’s limits can upset the delicate balance inherent to our

system of government. Id. at 408–09 (quoting United States v. Richardson, 418 U.S. 166, 188

(1974) (Powell, J., concurring)). For that reason, these requirements cannot be waived, forfeited,

or excused—regardless of the expediency of doing so. Va. House of Delegates v. Bethune-Hill,

587 U.S. 658, 662–63 (2019). And standing remains “a bedrock constitutional requirement”

necessary to the exercise of federal judicial power. United States v. Texas, 599 U.S. 670, 675

(2023). As a result, “any person invoking the power of a federal court must demonstrate

standing to do so.” Hollingsworth v. Perry, 570 U.S. 693, 704 (2013). Without that showing,

the federal court lacks jurisdiction over the claim.

History also shows that the federal judicial power encompasses not just “the power to

hear and determine the subject matter in controversy” but also the power “to adjudicate or

exercise any judicial power over” those “parties to a suit.” Rhode Island v. Massachusetts,

37 U.S. (12 Pet.) 657, 718 (1838). Even before the entry of final judgment, the exercise of

federal judicial power can “profoundly affect the lives, liberty, and property of those to whom it

extends.” 2 Valley Forge Christian Coll. v. Ams. United for Separation of Church & State, Inc.,

2The judicial power of the United States includes—but is not limited to—the power to issue binding

judgments. After all, the Supreme Court has long recognized that “[c]ertain implied powers must necessarily result

to our Courts of justice from the nature of their institution.” United States v. Hudson, 11 U.S. (7 Cranch) 32, 34

(1812). Foremost among these is “the power to fine and imprison for contempt . . . inherent in all courts of record.”

United States v. Barnett, 376 U.S. 681, 699–700 (1964) (quoting Watson v. Williams, 36 Miss. 331, 341 (1858)).

And courts also have “the power to issue subpoenas as to witnesses and documents.” U.S. Cath. Conf. v. Abortion

Rts. Mobilization, Inc., 487 U.S. 72, 76 (1988). These powers, like all judicial power vested in the federal courts,

are “jealously confined” by Article III’s requirements. United States v. Morton Salt Co., 338 U.S. 632, 641–42

(1950). So a federal court may only exercise them when considering a case or controversy properly within its

jurisdiction, which in turn requires that the parties have standing to sue. This limitation is important. Our

Constitution does not authorize federal judges to be either knights errant or inquisitors—pursuing our own ideals or

No. 23-1940 Speerly et al. v. General Motors, LLC Page 56

454 U.S. 464, 473 (1982). Which is why a court must assure itself of its jurisdiction as to all

parties at every stage of the litigation. See FBI v. Fikre, 601 U.S. 234, 244 (2024). “Without

jurisdiction the court cannot proceed at all in any cause” such that when it is lacking, “the only

function remaining to the court is that of announcing the fact and dismissing the cause.” Ex

parte McCardle, 74 U.S. (7 Wall.) 506, 514 (1868). Rule 23 didn’t—and couldn’t—alter this

fundamental limitation on the federal judicial power. See Fed. R. Civ. P. 82; see also Amchem

Prods., 521 U.S. at 613. And so the class action form doesn’t exempt unnamed plaintiffs from

Article III’s essential requirement that they have a “personal stake” in the case. Moore v.

Harper, 600 U.S. 1, 14 (2023) (quoting Baker v. Carr, 369 U.S. 186, 204 (1962)).

Nor is it too onerous for the district court to consider the standing of the unnamed class

members prior to asserting jurisdiction over their claims. The added burden of this inquiry is not

enough to require the wholesale disregard of this “irreducible constitutional minimum of

standing.” Dep’t of Educ. v. Brown, 600 U.S. 551, 561 (2023) (quoting Lujan, 504 U.S. at 560).

Because “standing is not dispensed in gross,” each plaintiff “must demonstrate standing for each

claim that they press.” TransUnion, 594 U.S. at 431. And each unnamed plaintiff in a class

action joins his own claim or claims to the litigation. See United States v. Sanchez-Gomez, 584

U.S. 381, 387–88 (2018). After all, the class action is a tool of convenience, simply “provid[ing]

a procedure by which the court may exercise . . . jurisdiction over the various individual claims

in a single proceeding.” Id. (quoting Califano v. Yamasaki, 442 U.S. 682, 700–01 (1979)

(alteration in original)). And once certified, the class has “important consequences for the

unnamed members”—binding them to the ultimate judgment and limiting their ability to settle

their individual claims. Id. (quoting Sosna v. Iowa, 419 U.S. 393, 399 n.8 (1975)). Which is

why unnamed plaintiffs “are considered parties to the litigation in many important respects.” Id.;

see e.g., Zahn v. Int’l Paper Co., 414 U.S. 291, 301 (1973) (holding that absent class members

are parties that must independently satisfy the amount-in-controversy requirement), superseded

by statute, Judicial Improvements Act of 1990, Pub. L. No. 101-650, Title III, § 310(a), 104 Stat.

5089 (codified at 28 U.S.C. § 1367), as recognized in Exxon Mobil Corp. v. Allapattah Servs.,

Inc., 545 U.S. 546, 566 (2005).

rooting out evil wherever it is found. Our exercise of the judicial power entrusted to us is circumscribed. And we

must respect these limitations at all points in the litigation—not just before entry of final judgment.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 57

That’s not to say that Article III requires each unnamed plaintiff to provide individualized

proof of standing at certification. But it does prevent a district court from certifying a class when

the record and class definition make clear that a significant number of the unnamed parties lack

standing. See Denney v. Deutsche Bank AG, 443 F.3d 253, 263–64 (2d Cir. 2006). As a result, a

class must “be defined in such a way that anyone within it would have standing.” Id. at 264.

Such a requirement ensures that federal courts don’t assert jurisdiction over claims that fall

outside the constitutional limitations imposed on the judiciary by Article III.

This standing requirement for unnamed plaintiffs is also consistent with the historical

practice that gave rise to the modern class action form. Though the history of Anglo-American

representative litigation is complex, the bill of peace provides the clearest analogue to the

modern class action. See Stephen C. Yeazell, From Medieval Group Litigation to the Modern

Class Action 24–26, 214–20 (1987). This was an equitable practice that allowed a representative

to sue on behalf of those similarly situated. It arose in response to the restrictive necessary-

parties rule, which required “all persons materially interested . . . in the subject matter of the

bill . . . to be made parties to the suit.” Ortiz v. Fibreboard Corp., 527 U.S. 815, 832 (1999)

(quoting West v. Randall, 29 F. Cas. 718, 721 (C.C.D.R.I. 1820) (No. 17,424) (Story, J.)). And

this necessary-parties requirement in turn developed to prevent a multiplicity of suits addressing

a common question.3 1 John Norton Pomeroy, A Treatise on Equity Jurisprudence §§ 245–46, at

257–58 (1881). But the joinder of necessary parties could create large and unwieldy suits when

many individuals had a common interest in the litigation. So the bill of peace was especially

useful when the number of interested parties made traditional methods of joinder impracticable.

Joseph Story, Commentaries on Equity Pleadings §§ 95–96, at 94–96 (2d ed. 1840). And once

joined under the bill, the unnamed parties were, “in a sense, deemed to be before the Court.” Id.

§ 99, at 102.

3This necessary-parties requirement was more than just a prudential protection for unnamed parties. The

doctrine hinged on the idea that a court of equity “in all cases delights to do complete justice, and not by halves.”

Knight v. Knight, 24 Eng. Rep. 1088, 1089 (Ch. 1734). And so the inability to join a necessary party rendered a suit

“unavoidably defective.” Joseph Story, Commentaries on Equity Pleadings § 81, at 81 (2d ed. 1840). Though the

bill of peace softened this rule’s astringency, it extended only to those who were necessary parties due to their

common interest. This is relevant to our modern jurisdictional inquiry because “history and tradition offer a

meaningful guide to the types of cases that Article III empowers federal courts to consider.” Sprint Commc’ns Co.

v. APCC Servs., Inc., 554 U.S. 269, 274 (2008). And so when the history speaks with one voice to show that the bill

of peace was limited to parties with a common interest, we should listen.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 58

That is not the same as saying that the bill of peace allowed a court of equity to join

necessary parties over whom it lacked jurisdiction. Indeed, lack of jurisdiction over a necessary

party was often “fatal to the whole suit” as a court had “no ground for proceeding to any degree

against them or their rights or interests.” Id. § 81, at 81–83. The bill of peace was merely a tool

of convenience, allowing representatives to join a multitude of necessary parties over whom the

court could already assert jurisdiction.

Review of the English chancery courts’ approach to these suits reveals that “in every case

in which one person can be permitted to sue on behalf of himself and others . . . he and all for

whom he appears, shall have an interest not merely in the property in question, but also in the

object of the suit.” See Frederic Calvert, A Treatise upon the Law Respecting Parties to Suits in

Equity 20 (1837); see also Zechariah Chafee, Jr., Some Problems of Equity 201 (1950) (“In such

situations, each member of the multitude had the same interests at stake as every other

member.”). For example, a group of tenants could represent all the tenants of an estate because

they all were united in interest since the lord’s assertion of “free warren” raised a question of

“whether there were sufficient common left for the tenants.” How v. Tenants of Bromsgrove, 23

Eng. Rep. 277, 277 (Ch. 1681). And certain named ore workers and lead mine owners could

represent all others within a parish because they all shared a common interest in the amount of

the tithe due on lead ore. See Brown v. Vermuden, 22 Eng. Rep. 802, 802 (Ch. 1676). Because

of this unity of interest, “[t]he common questions were the only questions in these old bills of

peace.” Chafee, supra, at 158.

Even as the bill of peace expanded beyond cases concerning manorial rights and

obligations, the chancery courts emphasized the continued requirement of a unity of interest

among the parties. For instance, a group of fishermen could stand in for all those who might

claim fishing rights in a river because the city of York’s claim of an exclusive right in those

waters “extends to all the defendants.” Mayor of York v. Pilkington, 26 Eng. Rep. 180, 181 (Ch.

1737). And in a similar vein, a group of partners in a brasswork mill could represent the entire

partnership in a suit against embezzling managers because the shared interest meant that the

unnamed partners “were in effect parties” to the litigation. Chancey v. May, 24 Eng. Rep. 265,

265 (Ch. 1722). For that reason two sailors could represent the shared interest of the crew in

No. 23-1940 Speerly et al. v. General Motors, LLC Page 59

seeking an accounting of prize money. See Leigh v. Thomas, 28 Eng. Rep. 201, 201 (Ch. 1751)

(upholding the demurrer because the two sailors had not brought the suit on behalf of the entire

crew, which raised the possibility of a multiplicity of suits). So despite the doctrine’s expansion,

it remained necessary that the absent members in a bill of peace “be interested in the subject

matter of controversy in the same way as their representatives.” Chafee, supra, at 164.

And this element persisted into the American equitable tradition. The first case in which

the United States Supreme Court addressed this issue concerned the schism between the northern

and southern branches of the Methodist church—a schism brought on by the issue of slavery.

See Smith v. Swormstedt, 57 U.S. (16 How.) 288, 298 (1853); see also Geoffrey C. Hazard, Jr. et

al., An Historical Analysis of the Binding Effect of Class Suits, 146 U. Pa. L. Rev. 1849, 1896

(1998). As part of the separation, representatives from the southern branch sought an equitable

accounting for their portion of the profits in the church’s Cincinnati publishing business, which

funded pensions for traveling ministers and their widows. Smith, 57 U.S. at 301. In response,

the northern branch challenged the bill on the grounds that the southern representatives could not

seek relief due to a lack of necessary parties. Id. at 302.

As a result, the case squarely presented the Court an opportunity to clarify the

requirements for a bill of peace. Drawing from Justice Story’s Commentaries on Equity

Pleadings, the Court emphasized that in cases where “a few are permitted to sue and defend on

behalf of the many, by representation, care must be taken that persons are brought on the record

fairly representing the interest or right involved.” Id. at 303. And the existence of the

representative suit depended on “a common interest or a common right” that implicates each of

the unnamed parties. Id. at 302. This requirement of common interest or right was understood to

limit the bill of peace. See Ayres v. Carver, 58 U.S. (17 How.) 591, 594 (1854) (noting in dicta

the apparent absence of an “interest or estate in common . . . that would authorize the rights of

the absent parties to be represented in the litigation”). And so the history shows that this

equitable practice didn’t give the Chancellor free rein to ignore the normal strictures of

jurisdiction.

Despite this history, it is tempting to ask whether any of this matters. Couldn’t we wait

until summary judgment to do the dirty work of separating the sheep from the goats and

No. 23-1940 Speerly et al. v. General Motors, LLC Page 60

determining who among the unnamed plaintiffs was injured? But this argument tiptoes around

the elephant in the room. For many damages class actions, certification is the whole ballgame.

The aggregation of hundreds of thousands of claims into a single suit is an immensely powerful

tool that raises the “risk of ‘in terrorem’ settlements that class actions entail.” Viking River

Cruises, Inc. v. Moriana, 596 U.S. 639, 662 (2022) (quoting AT&T Mobility LLC v. Concepcion,

563 U.S. 333, 350 (2011)). After all, “[e]ven in the mine-run case, a class action can result in

‘potentially ruinous liability’” such that the district “court’s decision to certify a class . . . places

pressure on the defendant to settle even unmeritorious claims.” See Shady Grove, 559 U.S. at

445 n.3 (Ginsburg, J., dissenting) (quoting Fed. R. Civ. P. 23(f) advisory committee’s note to

1998 amendment). This is the crux of the “procedural unfairness to which class actions are

uniquely susceptible.” In re Ford Motor Co., 86 F.4th 723, 729 (6th Cir. 2023) (per curiam).

And yet it’s not only the defendant’s ox that is gored. The district court’s improper

assertion of jurisdiction can just as easily harm the interests of an unnamed plaintiff. Before

certification, a district court’s judgment “cannot bind proposed class members.” See Smith v.

Bayer Corp., 564 U.S. 299, 316 n.11 (2011). But once certified, the court’s “judgments bind

absentees with respect to their individual claims for relief.” Viking River Cruises, 596 U.S. at

654 (citing Cooper v. Fed. Rsrv. Bank of Richmond, 467 U.S. 867, 874 (1984)). So when a

district court certifies a class containing individuals who lack Article III standing, an adverse

judgment could purport to bind parties over whom the court lacks jurisdiction. This would be,

“by very definition,” an ultra vires act. Steel Co. v. Citizens for Better Env’t, 523 U.S. 83, 101–

02 (1998). And such action severely prejudices the unnamed defendant by robbing them of the

chance to have their claim adjudicated by a court with proper jurisdiction. After all, Article III

only limits the federal judicial power, not that of the state courts. ASARCO Inc. v. Kadish, 490

U.S. 605, 617 (1989). And the existence of federal standing is not a requirement of alternative

dispute resolution—such as arbitration. Since Rule 23(c)(2) presumes the inclusion of all

unnamed class members who don’t opt out, there is a risk of procedural unfairness if the court

doesn’t assure itself of jurisdiction at certification.

No. 23-1940 Speerly et al. v. General Motors, LLC Page 61

The latent procedural unfairness of the class action device counsels against a special rule

exempting these suits from the injury-in-fact requirement until the final stages of the litigation.

Contrary to Judge Thapar’s suggestion that such a holding would presage the return of the

traditional perils of misjoinder, a lack of standing by some unnamed plaintiffs wouldn’t always

end the case. When a proposed class has a significant number of unnamed members without a

concrete injury, those individuals lacking Article III standing must be excluded—just like in any

other case before a federal court. And yet that doesn’t mean that the case is over. It would be up

to the plaintiffs and the court to redefine the class to exclude the parties shown to lack standing.

And so I would hold that Article III’s requirements apply with equal force to all suits before the

federal courts, class action or not.

II.

Since a district court cannot certify a class with a significant number of uninjured

individuals, I would apply that standard to this case. And a review of the record shows that

many unnamed class members lack standing because their claimed harm isn’t an injury in fact.

To sue in federal court, a litigant must show that they have suffered or will suffer injury

in fact—the “‘[f]irst and foremost’ of standing’s three elements.” Spokeo, Inc. v. Robins, 578

U.S. 330, 338 (2016) (alteration in original) (quoting Steel Co., 523 U.S. at 103). Injury in fact

requires that the complained-of harm be concrete and particularized—not abstract or

generalized—as well as being actual or imminent—not speculative. FDA v. All. for Hippocratic

Med., 602 U.S. 367, 381 (2024) (citing Lujan, 504 U.S. at 560–61). These elements ensure that

“a federal court may resolve only ‘a real controversy with real impact on real persons.’”

TransUnion, 594 U.S. at 424 (quoting Am. Legion v. Am. Humanist Ass’n, 588 U.S. 29, 87

(2019) (Gorsuch, J., concurring in judgment)). Since this is a constitutional requirement,

Congress cannot sidestep injury in fact “by statutorily granting the right to sue to a plaintiff who

would not otherwise have standing.” Spokeo, 578 U.S. at 339 (quoting Raines v. Byrd, 521 U.S.

811, 820 n.3 (1997)).

When analyzing whether a harm is sufficiently concrete, we look to whether the claimed

injury “has a ‘close relationship’ to a harm ‘traditionally’ recognized as providing a basis for a

No. 23-1940 Speerly et al. v. General Motors, LLC Page 62

lawsuit in American courts.” TransUnion, 594 U.S. at 424 (quoting Spokeo, 578 U.S. at 341).

Because of their historical pedigree, certain “physical harms and monetary harms” can “readily

qualify as concrete injuries under Article III.” Id. at 425. But this doesn’t mean that every harm

measured in dollars and cents automatic

This text is long and has been trimmed here. Open the source document for the complete record.