Full Opinion

NOT FOR PUBLICATION FILED UNITED STATES COURT OF APPEALS NOV 12 2026 MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS FOR THE NINTH CIRCUIT ALEX LOPEZ; TRAVELL WOODS, No. 26-6306 D.C. No. Plaintiffs - Appellees, 2:25-cv-07808-FWS-JDE v. MEMORANDUM* SUZUKI MOTOR OF AMERICA, INC., Defendant - Appellant. Appeal from the United States District Court for the Central District of California Fred W. Slaughter, District Judge, Presiding Argued and Submitted October 26, 2026 Pasadena, California Before: R. NELSON and H.A. THOMAS, Circuit Judges, and MARTÍNEZ- OLGUÍN, District Judge.** Defendant Suzuki Motor of America, Inc., appeals the district court’s order remanding Plaintiffs’ class action to state court based on the mandatory home state exception to the Class Action Fairness Act (“CAFA”). That exception requires * This disposition is not appropriate for publication and is not precedent except as provided by Ninth Circuit Rule 36-3. ** The Honorable Araceli Martínez-Olguín, United States District Judge for the Northern District of California, sitting by designation. remand if at least 2/3 of class members and all “primary defendants” “are citizens of the State in which the action was originally filed.” 28 U.S.C. § 1332(d)(4)(B). We have jurisdiction under 28 U.S.C. § 1453(c)(1). We review the district court’s decision to remand under the mandatory home state exception de novo and the district court’s factual findings for clear error. Adams v. W. Marine Prods., Inc., 958 F.3d 1216, 1219–20 (9th Cir. 2020). We affirm the district court’s order. 1. For the mandatory home state exception to apply, “the moving party must provide some facts in evidence from which the district court may make findings regarding class members’ citizenship.” Adams, 958 F.3d at 1221 (cleaned up). Plaintiffs’ burden of proof is preponderance of the evidence. Id. “Although such a finding must be based on more than mere guesswork, we have repeatedly cautioned that the burden . . . should not be exceptionally difficult to bear.” Id. (cleaned up). Here, it was not clear error for the district court to conclude that at least 2/3 of the proposed class are likely California citizens. The district court reviewed Plaintiffs’ expert analysis, which ran Suzuki’s customer registration data through TransUnion’s address database to estimate the citizenship of proposed class members. This analysis of class members’ purchase data revealed that (1) 60.7% “were tracked to the same California address they provided at the time of purchase or a new California address,” (2) 21% “were not tracked to any new address and provided a California address at the time of purchase,” and (3) 17.6% “were matched 2 26-6306 to an address outside California.” As a result, the district court found that “about 81.7% of the class ha[ve] a last-known mailing address in California.” The district court acknowledged it was a close question, but determined that Plaintiffs’ data were “sufficient to show that at least two-thirds of the class were California citizens at the time of removal.” In making this finding, the district court held that (1) individuals “matched to a California address twice—both at the time of purchase and at [the time of the expert’s analysis]—” were likely to be California citizens, and (2) people are “more likely to purchase a motorcycle in their home state.” The district court also reasoned that the 15% excess of class members above the 2/3 threshold covered any non-citizens. Based on this factual record and the absence of contrary evidence from Suzuki, the district court’s conclusion “that at least two-thirds of the class were California citizens at the time of removal” was not clearly erroneous. Suzuki makes essentially six arguments in an attempt to avoid this conclusion. First, Suzuki argues that the district court erred because Plaintiffs’ purported evidence of last-known mailing addresses cannot establish U.S. citizenship. But in Mondragon v. Capital One Auto Finance, we rejected the notion “that evidence of residency can never establish citizenship” and required only that “a [district] court . . . consider the entire record to determine whether evidence of residency can properly establish citizenship.” 736 F.3d 880, 886 (9th Cir. 2013) (cleaned up). And 3 26-6306 in Adams, we held that evidence of last known mailing addresses was sufficient, in light of the entire record, to establish the citizenship of class members. 958 F.3d at 1221–23. Second, Suzuki contends that, even if 81.7% of class members have a last- known mailing address in California, this fact is insufficient to establish that 2/3 of class members are California citizens. In Adams, however, we held that the district court reasonably inferred that the plaintiff had established that 1/3 of class members were California citizens because over 90% had California last-known mailing addresses and the excess accounted for non-citizens. Id. at 1221–23. While the difference between 2/3 and 81.7% is not as great, Suzuki offered no evidence below that would render the district court’s opinion clearly erroneous. Third, Suzuki argues the district court erred by assuming the data represented (1) purchasers rather than motorcycles and (2) individuals rather than businesses. But as the district court explained, Suzuki agreed in its discovery responses “to provide data about ‘unique persons.’” And the Confidentiality Agreement under which Suzuki shared its customer data with Plaintiffs’ expert described the data as “personally identifiable information of individuals.” The district court did not err by holding Suzuki to its representations. Fourth, Suzuki argues that the district court wrongly assumed that the addresses provided to Suzuki at the time of purchase were last known mailing 4 26-6306 addresses and that these reflected customers’ domiciles at the time of the complaint. But the TransUnion analysis supports the district court’s conclusion. It showed that 60.7% of class members were associated with a California address twice. And it found no record for another 21% of class members who provided a California address at the time of purchase. As Plaintiffs’ expert testified, “TransUnion aggregates data from thousands of public and proprietary sources” and “provide[s] an updated address if there is a single matching record,” which makes TransUnion’s data “highly accurate.” If an individual relocated elsewhere, TransUnion likely would have found the individual. It was therefore reasonable for the district court to base its conclusion on 81.7% of class members having a last-known mailing address in California. The TransUnion data is even more compelling given that some class members were associated with a California address for up to 20 years because long-term association strongly suggests domicile. See Adams, 958 F.3d at 1221. Any effect from a 10-month gap between the complaint and the analysis is likely negligible. Fifth, Suzuki argues that the district court wrongly assumed that the 21% of class members who were not matched by TransUnion remained California citizens. But we recognize a “presumption of continuing domicile, which provides that, once established, a person’s state of domicile continues unless rebutted with sufficient evidence of change.” Adams, 958 F.3d at 1221 (cleaned up). 5 26-6306 Sixth, Suzuki faults the district court for “‘infer[ring] that people are more likely to purchase a motorcycle in their home state.’” But as Plaintiffs explain, “[t]hat 17,412 purchasers changed addresses and are still located in California is consistent with [this] inference.” Even if we accepted the reductions Suzuki advances on appeal, Suzuki’s best estimate of class members who are likely California citizens is about 64.6%.1 If Suzuki’s best estimate is so close to 2/3, there cannot be “a definite and firm conviction that a mistake has been made.” Husain v. Olympic Airways, 316 F.3d 829, 835 (9th Cir. 2002) (cleaned up), aff’d on other grounds, 540 U.S. 644 (2004). 2. We also reject Suzuki’s argument that the mandatory home state exception does not apply because Suzuki Motor Corporation (“SMC”), its Japanese corporate relative, is a “primary defendant” disguised as a Doe Defendant. First, a primary defendant is one that is “liable” or “sued.” Singh v. Am. Honda Fin. Corp., 925 F.3d 1053, 1068 (9th Cir. 2019). This describes only parties to a lawsuit. See Lightfoot v. Cendant Mortg. Corp., 580 U.S. 82, 95 (2017); Cox v. CoinMarketCap OPCO, LLC, 112 F.4th 822, 829 (9th Cir. 2024), cert. denied sub nom. CoinMarketCap OpCo v. Cox, 145 S. Ct. 2846 (2025). “Courts do grant an exception to this rule for ‘John Doe’ defendants, but only in situations where the 1 This number is based on compounding the rate of migration to other states over 20 years. But as Plaintiffs note, not all class members bought their motorcycles 20 years ago. 6 26-6306 otherwise unavailable identity of the defendant will eventually be made known through discovery.” Newdow v. Roberts, 603 F.3d 1002, 1010–11 (D.C. Cir. 2010) (footnote omitted) (citing Gillespie v. Civiletti, 629 F.2d 637, 642 (9th Cir. 1980)). Second, a plaintiff is “the master of the complaint.” Royal Canin U. S. A., Inc. v. Wullschleger, 604 U.S. 22, 35 (2025) (cleaned up). If SMC is treated as a defendant, Plaintiffs would lose their right to sue only those whom they choose. Third, our conclusion will not create a “loophole” that “allow[s] plaintiffs to sue diverse defendants in state courts.” A class action that becomes removable under CAFA after a remand can be removed again at any time. See 28 U.S.C. § 1453(b). So, if Suzuki adds SMC as a party, Suzuki can simply remove again. If SMC is not added as a party, the state court cannot find SMC liable because it also ultimately lacks jurisdiction over unserved defendants. See Farris v. Merritt, 63 Cal. 118, 119 (1883); Cal. Civ. Proc. Code § 474. AFFIRMED. 7 26-6306