OPINION

This appeal, set in the -high-stakes world of furniture sales, runs the gamut of trademark infringement issues. The facts are somewhat unusual: the alleged infringer, leather furniture manufacturer Omnia Italian Design, Inc. (“Omnia”), admits that it blatantly copied and began selling the same goods branded with the mark of its (now ex) business partner, retail furniture company Stone Creek, Inc. (“Stone Creek”).

The first question we confront is whether Omnia’s use of Stone Creek’s mark was likely to cause confusion. Placing an identical mark on identical goods creates a strong likelihood of confusion, especially when the mark is fanciful. Because Stone Creek also sells in overlapping marketing channels and the law dictates that other factors heighten the likelihood that consumers will be confused as to the origin of the furniture, we reverse the district court’s contrary determination.

We also reject Omnia’s invocation of a common-law defense — known as the Tea Rose-Rectanus doctrine — that protects use of a mark in a remote geographic area when the use is in good faith. Omnia’s knowledge of Stone Creek’s prior use defeats any claim of good faith. Finally, we confirm that a 1999 amendment to the trademark statutes does not sweep away our precedent requiring that a plaintiff prove willfulness to justify an award of the defendant’s profits. A remand is necessary to determine whether Stone Creek can make that showing here.

Background Stone Creek, which manufactures furniture and sells directly to customers, has five showrooms in the Phoenix, Arizona area. Around 1990, Stone Creek adopted and began using the STONE CREEK mark:

[[Image here]]

In 1992, Stone Creek obtained state trademark protection. Twenty years later, in 2012, Stone Creek federally registered its mark. As described in the federal registration, the STONE CREEK mark is a red oval circling the words “Stone Creek” for various types of furniture.

In 2003, Stone Creek met representatives of Omnia — a manufacturer of leather furniture — at a California trade show. Dazzled by Omnia’s pitch, Stone Creek agreed to buy Omnia’s furniture. The two companies entered into an agreement under which Omnia manufactured leather furniture branded with the STONE CREEK mark. The business relationship stayed strong through 2012, but in 2013, Stone Creek discovered that Omnia had been using the STONE CREEK mark on competing furniture.

Omnia’s unauthorized use began in 2008 when Omnia was trying to woo a big client. For many years before that, Omnia had worked with retailer Bon-Ton Stores, Inc. (“Bon-Ton”), but Bon-Ton became a “significant” customer in 2008. Bon-Ton signed on for Omnia to supply Bon-Ton’s leather furniture. However, Bon-Ton did not want to sell under the Omnia name; instead, Bon-Ton preferred a label that sounded “American.” Although Omnia offered multiple options, Bon-Ton opted for STONE CREEK. According to Omnia, part of the allure of selecting the STONE CREEK mark was that marketing materials and a logo were already prepared.

Omnia copied the logo directly from Stone Creek’s materials. Omnia’s team used old documents with Stone Creek’s logo to digitally recreate the identical logo because they could not achieve sharp resolution by scanning. Then Omnia plastered the mark onto a host of items, including binders, leather samples, and color boards for display in Bon-Ton stores. Most relevant here, Omnia designed warranty cards with the STONE CREEK mark, and, from 2008 to 2013, sold leather furniture to Bon-Ton branded with the STONE CREEK mark. Near the end of this period, Stone Creek — still unaware of Omnia’s misdoings — obtained its federal trademark registration.

The STONE CREEK-labeled furniture produced by Omnia was shipped to Bon-Ton and sold to customers at Bon-Ton’s various furniture galleries in the Midwest. Omnia’s products reached purchasers living within 200 miles of a Bon-Ton gallery, including portions of Illinois, Indiana, Iowa, Michigan, Ohio, Pennsylvania, and Wisconsin. The claimed infringing sales all occurred within that area.

Stone Creek caught a whiff of what Om-nia was doing by 2013. Multiple customers contacted Stone Creek to ask about product options and store locations in the Midwest. Another customer called about a warranty issue with a leather sofa purchased at a Bon-Ton store in Chicago. The warranty card contained the STONE CREEK mark, which led the customer to Stone Creek’s website. At that point, Stone Creek’s president discovered that the mark was being used on furniture sold on Bon-Ton’s website, and Stone Creek asked Omnia if it was selling products with the STONE CREEK mark to other companies.

To its credit, Omnia was candid. In an email from the Vice President of Sales, Omnia unequivocally admitted to selling furniture under the STONE CREEK mark. In a move not recommended when litigation is certainly impending, the email observed: “In this day of internet shopping and surfing, it is unfortunate and probably a nuisance for you that your stores are receiving inquiries regarding these products due to the similar name.”

Stone Creek filed suit in the District of Arizona, alleging federal and common-law trademark infringement and unfair competition. After a bench trial, the district court held that Omnia did not infringe Stone Creek’s trademark because there was no likelihood of confusion. The court also ruled at summary judgment that disgorgement of Omnia’s profits required a showing of willful infringement, but that determination was never made because Omnia was not found to infringe. During the course of the case, the court sanctioned Stone Creek’s attorneys on two occasions for filings related to Omnia’s profits and damages.

Analysis

I. Likelihood of Confusion

Under the Lanham Act, infringement lies for both registered and unregistered trademarks when the alleged infringer’s use “is likely to cause confusion, or to cause mistake, or to deceive.” 15 U.S.C. §§ 1114 (1)(a), 1125(a)(1)(A). The touchstone for trademark infringement is likelihood of confusion, which asks whether a “reasonably prudent” marketplace consumer is “likely to be confused as to the origin of the good or service bearing one of the marks.” Rearden LLC v. Rearden Commerce, Inc., 683 F.3d 1190, 1214 (9th Cir. 2012) (citation omitted). Although we review the district court’s findings and determination of no likelihood of confusion for clear error, we address legal error de novo. See Gilman v. Brown, 814 F.3d 1007, 1017 (9th Cir. 2016) (“Because the district court applied the wrong standard, it committed legal error, and the resulting factual findings are clearly erroneous.”), cert. denied sub nom. Madden v. Brown, — U.S.-, 137 S.Ct. 650 , 196 L.Ed.2d 544 (2017); Reno Air Racing Ass’n, Inc. v. McCord, 452 F.3d 1126, 1135 (9th Cir. 2006); see also Dorr-Oliver, Inc. v. Fluid-Quip, Inc., 94 F.3d 376, 380 (7th Cir. 1996) (noting that courts “review the district court’s statement of the law de novo for legal error” in likelihood of confusion cases).

The district court properly cited the well-established factors (the Sleekcraft factors) that guide the likelihood of confusion inquiry. See AMF Inc. v. Sleekcraft Boats, 599 F.2d 341, 348-49 (9th Cir. 1979), abrogated on other grounds by Mattel, Inc. v. Walking Mountain Prods., 353 F.3d 792 (9th Cir. 2003). In conducting the analysis, courts do not merely count beans or tally points. See Dreamwerks Prod. Grp., Inc. v. SKG Studio, 142 F.3d 1127, 1129 (9th Cir. 1998). Not all factors are created equal, and their relative weight varies based on the context of a particular case. See Network Automation, Inc. v. Advanced Sys. Concepts, Inc., 638 F.3d 1137, 1145 (9th Cir. 2011).

Two particularly probative factors are the similarity of the marks and the proximity of the goods. See Lindy Pen Co. v. Bic Pen Corp., 796 F.2d 254, 256-57 (9th Cir. 1986). Other potentially relevant factors include the strength of the protected mark, evidence of actual confusion, the use of a common marketing channel, the defendant’s intent in selecting the allegedly infringing mark, the type of goods and the degree of consumer care, and the likelihood of product expansion. Sleekcraft, 599 F.2d at 348-49 . After examining the district court’s consideration of the factors in this case, we conclude that the court legally erred in framing many of the Sleekcraft factors, leading us to reverse the finding of no likelihood of confusion. The indistinguishable marks and goods, coupled with a fanciful mark, evidence of actual confusion, convergent marketing channels, and blatant copying, tell the real story.

A. Similarity of the Marks and Proximity of the Goods

The parties do not dispute — nor could they — the district court’s finding that the factors examining similarity of the marks and the proximity of the goods favor Stone Creek. But simply acknowledging the identity of the marks and goods understates the importance of these factors and our precedent’s conclusion that identical marks paired with identical goods can be case-dispositive: “In light of the virtual identity of marks, if they were used with identical products or services likelihood of confusion would follow as a matter of course.” Brookfield Commc’ns, Inc. v. W. Coast Entm’t Corp., 174 F.3d 1036, 1056 (9th Cir. 1999); Opticians Ass’n of Am. v. Indep. Opticians of Am., 920 F.2d 187, 195 (3d Cir. 1990) (“[LJikelihood of confusion is inevitable, when, as in this case, the identical mark is used concurrently by unrelated entities.”). The case for confusion based on identical marks and identical goods could hardly be stronger than here.

Omnia’s mark is an exact replica of Stone Creek’s logo that Omnia copied from materials given to Omnia by Stone Creek. And the goods on which the marks appeared are identical: not only do Stone Creek and Bon-Ton sell leather furniture, but they both sell leather furniture that is manufactured by Omnia. As McCarthy, a leading trademark scholar, has observed, cases involving identical marks on competitive goods are rare and “hardly ever find their way into the appellate reports” because liability is “open and shut.” 4 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 23:20 (4th ed. 2017); Wynn Oil Co. v. Thomas, 839 F.2d 1183, 1190-91 (6th Cir. 1988). We examine the remaining factors to determine whether they displace the powerful case of likelihood of confusion.

B. Strength of the Protected Mark

The strength of the mark is a key factor with two components: the mark’s recognition in the market (i.e., its commercial strength) and the mark’s inherent distinctiveness (i.e., its conceptual strength). See Lahoti v. Vericheck, Inc., 636 F.3d 501, 508 (9th Cir. 2011). The district court analyzed only half of the equation when it said that “[t]he STONE CREEK mark is strong in Arizona, but it is not recognized in the [Midwest] for its relationship to Stone Creek.” As we explain below, this crabbed view of commercial strength is misplaced. But the court’s omission of the inherent distinctiveness of the mark is even more significant because, on the spectrum of conceptual strength, Stone Creek’s mark falls at the high end as a fanciful or arbitrary mark. See Entrepreneur Media, Inc. v. Smith, 279 F.3d 1135, 1141 (9th Cir. 2002). Its mark is not a suggestive or descriptive mark that is less likely to act as a unique identifier of the source in a consumer’s mind. Id. at 1141-42 . The district court committed legal error in not adducing and evaluating this aspect of the strength factor. See id. at 1141 (describing the importance of the strength factor in determining the scope of trademark protection).

C. Evidence of Actual Confusion

Given the other factors favoring likelihood of confusion, it is not surprising that Stone Creek has put forward several instances of actual confusion. Such evidence is not necessary for a finding of likelihood of confusion, but it bears on the inquiry and is particularly potent. See Sleekcraft, 599 F.2d at 352-53 . The district court disregarded Stone Creek’s evidence because the court focused on whether there was “actual confusion by any consumer in the [Midwest] who purchased Omnia furniture believing it was manufactured or sold by Stone Creek.” This approach misapprehends the breadth of likelihood of confusion, which can exist even when consumers are not “confused as to the source of a product they actually purchase.” Brookfield Commc’ns, 174 F.3d at 1057 .

When we widen the lens to embrace the full scope of qualifying actual confusion evidence, we credit the examples of customers seeking to purchase furniture or having already purchased Bon-Ton furniture who misdirected their queries to Stone Creek. Cf. Nutri/Sys., Inc. v. Constan Indus., Inc., 809 F.2d 601, 606 (9th Cir. 1987). These consumers did not ask whether Stone Creek and Bon-Ton are affiliated; they were actually confused as to the source. See Cohn v. Petsmart, Inc., 281 F.3d 837 , 842 n.7 (9th Cir. 2002) (per curiam). The customers requested information about product options and store locations — issues likely to affect their buying habits and their view of the company associated with the mark. See Rearden LLC, 683 F.3d at 1214 (characterizing trademark infringement as concerned with consumers’ purchasing decisions). They also contacted Stone Creek about warranty problems with Omnia furniture from Bon-Ton. These occasions of actual confusion cannot be dismissed out of hand but must be considered in context and in light of the other evidence of likelihood of confusion. See 4 McCarthy, supra, § 23:14; see also Streamline Prod. Sys., Inc. v. Streamline Mfg., Inc., 851 F.3d 440, 457 (5th Cir. 2017) (“[C]ourts may not ignore competent evidence of actual confusion.” (citation omitted)); Daddy's Junky Music Stores, Inc. v. Big Daddy’s Family Music Ctr., 109 F.3d 275, 284 (6th Cir. 1997) (explaining that one instance of actual confusion “favors plaintiff at least to some extent”).

D. Convergence of Marketing Channels

The district court’s consideration of overlapping marketing channels rested on the faulty legal assumption that geographic separation automatically means no intersection in marketing channels. This misconception led the court to unduly weigh this factor against Stone Creek. Even when parties “operate in different geographical areas,” they may still “provide [goods] in convergent marketing channels.” Park ’N Fly, Inc. v. Dollar Park & Fly, Inc., 782 F.2d 1508, 1509 (9th Cir. 1986). This principle is highlighted here because Stone Creek uses its website as a substantial channel to market its furniture beyond its physical stores in Arizona.

The parties tacitly agree that the analysis of marketing channels should look to Bon-Ton’s sales to customers rather than Omnia’s sales to Bon-Ton. Thus, we focus on Stone Creek and Bon-Ton. Both are retail furniture stores, and their products are identical pieces of leather furniture manufactured by Omnia. Selling similar, let alone interchangeable, products suggests that Stone Creek and Bon-Ton share the same general class of customers. See Pom Wonderful LLCv. Hubbard, 775 F.3d 1118, 1130 (9th Cir. 2014).

That Stone Creek does most of its business in the Phoenix area does not foreclose overlapping marketing channels with Om-nia. The district court found that Stone Creek placed its mark on its website, sto-necreekfurniture.com, as early as 2000 and optimized search engine searches so that its website appears early in Internet search results. During the period of alleged infringement, Stone Creek also advertised in a nationwide magazine with readership in the Midwest. Since its inception, Stone Creek has made more than $200,000,000 in sales; $610,384 of those sales occurred in the Midwest.

The district court’s factual findings confirm that Stone Creek’s print and online advertising reached the Midwest. Stone Creek’s marketing traction in the Midwest is buttressed by the finding that Stone Creek had customers and made sales in that region at the same time that Bon-Ton was selling the STONE CREEK-labeled furniture sourced from Omnia. To be sure, those sales represent a small fraction of Stone Creek’s total sales, but, in light of the particular facts of this case, the small volume of the overall sales does not undercut Stone Creek’s distribution of furniture in the Midwest.

What appears to have led the district court astray in analyzing the marketing channels factor (and permeates much of the court’s discussion of other factors) is a myopic focus on the considerable distance between Stone Creek’s physical showrooms in Arizona and Bon-Ton’s in the Midwest. As the court explained, Stone Creek follows the retail furniture business model, where sales are generally limited to customers living within a drivable distance from the brick-and-mortar stores. But the territorial separation is not so well-delineated or physically defined: Stone Creek and Bon-Ton were simultaneously advertising and selling under the STONE CREEK mark in the Midwest. Thus, the likelihood of confusion is not so diminished that Stone Creek’s nationwide right to enforce its mark in the alleged infringer’s area has not yet ripened into a remedy. See Mister Donut of Am., Inc. v. Mr. Donut, Inc., 418 F.2d 838, 844 (9th Cir. 1969) (finding that likelihood of confusion was not precluded because the parties were “no longer confined to separate and distinct market areas”).

E. Intent in Selecting the Allegedly Infringing Mark

Omnia’s reason for adopting the STONE CREEK mark also plays a critical role: when the alleged infringer intended to deceive customers, we infer that its conscious attempt to confuse did in fact result in confusion. Playboy Enters., Inc. v. Netscape Commc’ns Corp., 354 F.3d 1020, 1028 (9th Cir. 2004). Recognizing the difficulty of collecting evidence of a party’s motive, we have held that choosing a designation with knowledge that it is another’s trademark permits a presumption of intent to deceive. Hokto Kinoko Co. v. Concord Farms, Inc., 738 F.3d 1085, 1096 (9th Cir. 2013).

The district court found that “Omnia adopted and used the STONE CREEK mark with full knowledge of Stone Creek’s senior use,” an explicit finding establishing the factual predicate to apply the presumption. Yet the court nowhere assesses or acknowledges the presumption, and that error colored its analysis and conclusion. Omnia’s bare assertion that the mark was picked for its “American” sound does not counteract the intent to deceive, especially when Omnia had endless options for suitably “Ameriean”-sounding names to offer to Bon-Ton. This rationale is no more convincing than if a contemporary marketeer decided to appropriate the long-standing Haagen-Dazs mark believing that a foreign-sounding name would appeal to customers. And Omnia’s replication of the mark from Stone Creek’s materials and unauthorized use on identical furniture contribute to an intentional appropriation. See Au-Tomotive Gold, Inc. v. Volkswagen of Am., Inc., 457 F.3d 1062, 1076 . (9th Cir. 2006).

Additionally, conjectural statements by Omnia’s president about the scope of Stone Creek’s business do not amount to a “good faith belief that there [would be] no conflict between the marks as used on [Omnia’s] goods.” 4 McCarthy, supra, § 23:115. Although the president “understood” that Stone Creek sold in Phoenix, he never researched where or how the furniture was sold. Nor did Omnia ask or investigate where Stone Creek’s customers were located before using the mark. Omnia has not unseated the presumption, and its deceptive intent is “entitled to great weight” in the ultimate determination of likelihood of confusion. Kendall-Jackson Winery, Ltd. v. E. & J. Gallo Winery, 150 F.3d 1042, 1048 (9th Cir. 1998) (citation omitted).

F. Degree of Consumer Care and Likelihood of Product Expansion

The remaining two factors — the degree of consumer care based on the type of goods and the likelihood of product expansion — do not support either party; at best, they weakly support Omnia. As the district court found, since furniture is an expensive good, likelihood of confusion is diminished because we justifiably expect that consumers will make purchases with more research and closer scrutiny. Multi Time Mach., Inc. v. Amazon.com, Inc., 804 F.3d 930, 937 (9th Cir. 2015), cert. denied, — U.S.-, 136 S.Ct. 1231 , 194 L.Ed.2d 185 (2016). But this factor is less instructive in cases like this one where the marks and goods are identical for the simple reason that even the trained eye will not be able to discern any difference. McGregor-Doniger Inc. v. Drizzle Inc., 599 F.2d 1126, 1137 (2d Cir. 1979), superseded on other grounds as recognized by Cross Commerce Media, Inc. v. Collective, Inc., 841 F.3d 155 (2d Cir. 2016); 4 McCarthy, supra, § 23.96. Similarly, while the district court did not clearly err in finding that Stone Creek has not demonstrated non-speculative plans to expand, that fact has minimal legal significance because Stone Creek established an overlap in goods and marketing channels. See Playboy Enters., 354 F.3d at 1029 .

G. Conclusion

We reverse the district court’s finding of no likelihood of confusion because it is based on faulty legal foundations. We credit the court’s factual findings, but its circumscribed view of the legal landscape left the court with an incomplete picture. The slam-dunk evidence of a conceptually strong mark together with the use of identical marks on identical goods is difficult to surmount. Viewing the facts through the correct legal lens, there is no substantial argument that the other factors and evidence overcome the robust case that Om-nia’s use of the STONE CREEK mark is likely to cause confusion.

II. The Tea Rose-Rectanus Doctrine

Our determination of a likelihood of confusion with respect to the STONE CREEK mark does not end the infringement analysis. Omnia asserts that its use of Stone Creek’s mark is protected under the Tea Rose-Rectanus doctrine and argues that we may affirm the district court’s judgment of no liability on this alternative basis. The district court rejected this defense, and so do we.

The Tea Rose-Rectanus doctrine has its roots in the common law: it is named for a pair of Supreme Court cases, Hanover Star Milling Co. v. Metcalf, 240 U.S. 403 , 36 S.Ct. 357 , 60 L.Ed. 713 (1916) (“Tea Rose”), and United Drug Co. v. Theodore Rectanus Co., 248 U.S. 90 , 39 S.Ct. 48 , 63 L.Ed. 141 (1918). The central proposition underlying the two cases is that common-law trademark rights extend only to the territory where a mark is known and recognized, so a later user may sometimes acquire rights in pockets geographically remote from the first user’s territory. The question we address is whether Omnia acquired common-law rights in the Midwest under the Tea Rose-Rectanus doctrine.

Omnia’s common-law rights, if they exist, are not wiped out merely because Stone Creek later filed a federal registration. Although federal registration presumptively entitles the senior user to nationwide protection, 15 U.S.C. § 1057 (b), the Lanham Act preserves legal and equitable defenses that could have been asserted prior to registration, id. § 1115(a). Under this rule, already-established common-law rights are carved out of the registrant’s scope of protection. Id. § 1115(b)(5); Johnny Blastoff, Inc. v. L.A. Rams Football Co., 188 F.3d 427, 435 (7th Cir. 1999). In other words, the geographic scope of a senior user’s rights in a registered trademark looks like Swiss cheese: it stretches throughout the United States with holes cut out where others acquired common-law rights prior to the registration. 5 McCarthy, supra, § 26:31. Because Omnia began using the mark in 2008, well before Stone Creek’s federal registration in 2012, the Tea Rose-Rectanus defense is available to Omnia if it is applicable.

To take advantage of the Tea Rose-Rectanus doctrine, the junior user must establish good faith use in a, geographically remote area. See Rectanus, 248 U.S. at 100 , 39 S.Ct. 48 ; cf. Grupo Gigante SA De CV v. Dallo & Co., 391 F.3d 1088 , 1096 & n.26 (9th Cir. 2004). Like the district court, we limit our discussion to the question of good faith because it is disposi-tive.

The varying descriptions of good faith in the leading Supreme Court cases have spawned a circuit split, and our circuit has not yet weighed in. See Grupo Gigante, 391 F.3d at 1096 n.26. On one side, some circuits have held that the junior user’s knowledge of the senior user’s prior use of the mark destroys good faith. See, e.g., Nat’l Ass’n for Healthcare Commc’ns, Inc. v. Cent. Ark. Area Agency on Aging, Inc., 257 F.3d 732, 735 (8th Cir. 2001); Money Store v. Harriscorp Fin., Inc., 689 F.2d 666 , 674-75 (7th Cir. 1982). In contrast, other circuits have held that knowledge is a factor informing good faith, but the “focus is on whether the [junior] user had the intent to benefit from the reputation or goodwill of the [senior] user.” GTE Corp. v. Williams, 904 F.2d 536, 541 (10th Cir. 1990); see C.P. Interests, Inc. v. Cal. Pools, Inc., 238 F.3d 690, 700 (5th Cir. 2001). We conclude that the better view is that there is no good faith if the junior user had knowledge of the senior user’s prior use.

Looking back to the origins of the Tea Rose-Rectanus doctrine informs why knowledge defeats a claim of good faith use. In Tea Rose, the senior user began selling “Tea Rose” flour in approximately 1872; many years later, the junior user began selling “Tea Rose” flour without any knowledge of the senior user’s prior use. 240 U.S. at 407-08 , 36 S.Ct. 357 . At the time that the trademark infringement action was filed, the senior user had made sales in Massachusetts, Ohio, and Pennsylvania, while the junior user’s sales had reached Mississippi, Alabama, Georgia, and Florida. Id. at 408-10 , 36 S.Ct. 357 . Rectanus arose on similar facts: the senior user began selling “Rex” drugs around 1877 and operated in New England, while the junior user began selling “Rex” drugs around 1883 and operated in Kentucky, with neither party being aware of the other’s use of the “Rex” mark for more than twenty years. 248 U.S. at 94-96 , 39 S.Ct. 48 . In both cases, the Supreme Court held that the senior user could not enjoin the junior user’s use of the same mark because the junior user adopted the mark in good faith and had developed a local reputation in an area where the mark was not recognized as designating the senior user. See id. at 103-04 , 39 S.Ct. 48 ; Tea Rose, 240 U.S. at 415-16 , 36 S.Ct. 357 .

When describing good faith, the Supreme Court emphasized that the junior user had no awareness of the senior user’s use of the mark. The Court in Tea Rose states that the junior user “adopted and used [the trademark] in good faith without knowledge or notice that the name ‘Tea Rose’ had been adopted or used ... by anybody else.” 240 U.S. at 410 , 36 S.Ct. 357 . The Court also refers to the situation as one where the two parties “independently” employ the same mark. Id. at 415 , 36 S.Ct. 357 . And the Court’s reasoning concentrates on knowledge:

Id. at 420 , 36 S.Ct. 357 (emphasis added).

The same focus on notice emerges in Rectanus , which grants protection for an “innocent” junior user who has “hit upon” the same mark and avers that the parties acted “in perfect good faith; neither side having any knowledge or notice of what was being done by the other.” 248 U.S. at 96, 103 , 39 S.Ct. 48 . The Court also relies on a case that says that the defendants there acted in good faith because they “believfed] [their] use to be original with them.” Richter v. Anchor Remedy Co, 52 F. 455, 455 (C.C.W.D. Pa. 1892), aff'd sub nom. Richter v. Reynolds, 59 F. 577 (3d Cir. 1893). Seventy years later, Justice Brennan stressed that application of the Tea Rose-Rectanus doctrine requires an absence of knowledge. See K Mart Corp. v. Cartier, Inc., 486 U.S. 281 , 314 n.8, 108 S.Ct. 1811 , 100 L.Ed.2d 313 (1988) (Brennan, J., concurring in part and dissenting in part) (“[A] firm can develop a trademark that is identical to a trademark already in use in a geographically distinct and remote area if the firm is unaware of the identity.”).

The Seventh and Eighth Circuits and the Trademark Trial and Appeal Board (“TTAB”) agree with this reading. The Seventh Circuit put it explicitly: “A good faith junior user is one who begins using a mark with no knowledge that someone else is already using it.” Money Store, 689 F.2d at 674. The court went on to analyze whether the junior user in that case had constructive or actual knowledge of the senior user’s use. Id. at 675. The Eighth Circuit follows the same approach, parroting the language from Tea Rose and Rectanus. See Nat’l Ass’n for Healthcare Commc’ns, 257 F.3d at 735 (“adopted the [mark] in good faith, without knowledge of [the] prior use”). And the TTAB, the administrative board charged with deciding certain trademark disputes and appeals, similarly holds that “appropriat[ing] a mark with knowledge that it is actually being used by another” means “that use is not believed to be a good faith use.” Woman’s World Shops Inc. v. Lane Bryant Inc., 5 U.S.P.Q.2d 1985 , 1988 (T.T.A.B. 1988).

The courts that have ruled the other way have latched on to one line in the Tea Rose case which reads:

240 U.S. at 415 , 36 S.Ct. 357 (emphasis added). But this brief reference to “design inimical” does not override the central focus on knowledge; it is not without significance that “design inimical” does not appear anywhere else in the opinion. The Court in Rectanus repeats the “design inimical” language as a direct quote of the language from the Tea Rose case and mentions offhand that the junior user did not have a “sinister purpose.” 248 U.S. at 101 , 39 S.Ct. 48 . More salient are the various points in the leading opinions that draw a close connection between “good faith” and “knowledge” or “notice.” See, e.g., id. at 96, 39 S.Ct. 48 (“in perfect good faith; neither side having any knowledge or notice of what was being done by the other”); id. at 103 , 39 S.Ct. 48 (“in good faith, and without notice of any prior use by others, selected and used the ‘Rex’ mark”); Tea Rose, 240 U.S. at 410 , 36 S.Ct. 357 (“trademark was adopted and used [by the junior user] in good faith without knowledge or notice that the name ‘Tea Rose’ had been adopted or used by the [senior user]”); id. at 419 , 36 S.Ct. 357 (“in good faith and without notice of the [senior user’s] mark”).

Tying good faith to knowledge makes sense in light of the policy underlying the doctrinal framework. As the Supreme Court explained, the Tea Rose-Rectanus doctrine operates to protect a junior user who unwittingly adopted the same mark and invested time and resources into building a business with that mark. Rectanus, 248 U.S. at 103 , 39 S.Ct. 48 ; Tea Rose, 240 U.S. at 419 , 36 S.Ct. 357 . A junior user like Omnia who has affirmative knowledge of the senior user’s mark has not serendipitously chosen the same mark and independently built up its own brand. Instead, a user like Omnia knows that its actions come directly at the expense of the senior user, potentially blocking the senior user from entering into the new market. Viewed in this light, the junior user has acted in bad faith, which “serve[s] as evidence that the [senior] user’s mark, at least in reputation, has extended to the new area.” Developments in the Law Trade-Marks and Unfair Competition, 68 Harv. L. Rev. 814 , 859 (1955); 5 McCarthy, supra, § 26:12.

The knowledge standard also better comports with the Lanham Act. The statutory section preserving the Tea Rose-Rec-tanus defense for junior users acting preregistration requires that the junior user’s mark “was adopted without knowledge of the registrant’s prior use.” 15 U.S.C. § 1115 (b)(5) (emphasis added). More broadly, one major change effected by the Lanham Act is that securing federal registration affords nationwide rights regardless of where the registrant has used the mark, a result accomplished by a provision that puts would-be users on constructive notice. See id. §§ 1057(b), 1072; 5 McCarthy, supra, § 26:32. In other words, the Lanham Act displaces the Tea Rose-Rectanus defense by charging later users with knowledge of a mark listed on the federal register. If constructive notice is sufficient to defeat good faith, it follows that actual notice should be enough too.

Once knowledge is accepted as a determinative factor in deciding good faith, the Tea Rose-Rectanus doctrine has no applicability here. The district court found that “[Omnia] was a non-innocent remote user” who “acquired no common law trademark rights in the [Midwest].” That conclusion flows from the parties’ agreement that Omnia adopted Stone Creek’s mark with knowledge of Stone Creek’s previous use. The Tea Rose-Rectanus doctrine provides no shelter to Omnia for infringement of Stone Creek’s mark.

III. Willfulness and Disgorgement of Profits

Under the remedies provision of the Lanham Act, a court may award (1) the defendant’s profits, (2) the damages sustained by the plaintiff, and (3) the costs of the action. 15 U.S.C. § 1117 (a). At issue here is the applicable standard to award disgorgement of profits. In an effort to shape its trial strategy, Stone Creek filed a motion for summary judgment asking the district court to rule that willfulness is not required for such an award. The court denied the motion and sanctioned Stone Creek for filing it.

Because we conclude that Omnia infringed Stone Creek’s mark, we address the standard for awarding disgorgement of profits as it will be front and center on remand. Historically, we have imposed a willfulness requirement with respect to disgorgement of profits. Now we must decide what effect, if any, a 1999 amendment to the Lanham Act’s remedies provision has on our precedent regarding awarding of profits. We agree with the district court that the 1999 amendment has not changed the state of the law on disgorgement and that willfulness is still required.

The evolution of the remedies provision-including the ever-persisting circuit split — is key to understanding the impact of the 1999 amendment, so we recount the history and the current state of affairs. Before 1999, the trademark remedies provision — § 1117(a) — stated:

15 U.S.C. § 1117 (a) (1996). The spirited debate among the circuits has been reserved for how the phrase “subject to the principles of equity” applies to an award of the defendant’s profits.

Our circuit fell in line with the camp that requires a showing of willfulness. We held that an award of the defendant’s profits “is not automatic and must be granted in light of equitable considerations”; equity dictates that the plaintiff must show that the defendant’s infringing acts were accompanied by some form of intent. Lindy Pen Co. v. Bic Pen Corp., 982 F.2d 1400, 1405-06 (9th Cir. 1993), abrogated on other grounds by SunEarth, Inc. v. Sun Earth Solar Power Co., 839 F.3d 1179 (9th Cir. 2016) (per curiam). At that time, the Third Circuit was in accord that “a plaintiff must prove that an infringer acted willfully before the infringer’s profits are recoverable.” SecuraComm Consulting Inc. v. Securacom Inc., 166 F.3d 182, 190 (3d Cir. 1999), superseded by statute as stated in Banjo Buddies, Inc. v. Renosky, 399 F.3d 168 (3d Cir. 2005). Other circuits agreed. See, e.g., George Basch Co. v. Blue Coral, Inc., 968 F.2d 1532, 1540 (2d Cir. 1992); ALPO Petfoods, Inc. v. Ralston Purina Co., 913 F.2d 958, 968 (D.C. Cir. 1990). In the other camp were circuits who viewed willfulness as one factor in the overall determination of whether an award of profits is appropriate. See, e.g., Pebble Beach Co. v. Tour 18 I Ltd., 155 F.3d 526, 554-55 (5th Cir. 1998), abrogated on other grounds by TrafFix Devices, Inc. v. Marketing Displays, Inc., 532 U.S. 23 , 121 S.Ct. 1255 , 149 L.Ed.2d 164 (2001); Roulo v. Russ Berrie & Co., 886 F.2d 931, 941 (7th Cir. 1989). These decisions all predate the 1999 amendment.

The amendment became necessary after Congress made a substantive change to the Lanham Act in 1996. That year, Congress added § 1125(c), which created a federal cause of action for trademark dilution. See H.R. Rep. No. 104-374, at 2 (1995), as reprinted in 1996 U.S.C.C.A.N. 1029, 1029. The provision expressly allowed holders of famous trademarks to enjoin uses that diluted the distinctive quality of their marks. See 15 U.S.C. § 1125 (c) (1996). Section 1125(c) also purported to provide monetary relief under the remedies provision, § 1117(a), when dilution was “willfully intended.” But Con gress failed to make the requisite cross-reference in § 1117(a) to harmonize that section with the amendment and soon discovered the missing link between the two statutory provisions.

That statutory mismatch spurred the 1999 amendment. Congress revised the remedies section, § 1117(a), to include reference to a “willful violation under section 1125(c).” The amendment thus made clear that a plaintiff with a dilution claim could recover money damages. The current version of § 1117(a) reads:

15 U.S.C. § 1117 (a) (emphasis added). To put it in plain English, if there is

then, “subject to the principles of equity,” the plaintiff is entitled to relief. Critically, Congress did not modify the “subject to the principles of equity” language.

The contrast in language between clause [1], which does not reference willfulness, and newly inserted clause [2], which does, has caused ripples through the circuit courts, which remain divided on the role of willfulness in awarding profits. The Federal Circuit, interpreting Second Circuit jurisprudence, held that “nothing in the 1999 amendment ... allows us to depart from ... precedent requiring willfulness for the recovery of profits in infringement cases.” Romag Fasteners, Inc. v. Fossil, Inc., 817 F.3d 782, 791 (Fed. Cir. 2016), cert. granted, judgment vacated, — U.S.-, 137 S.Ct. 1373 , 197 L.Ed.2d 550 (2017), opinion reinstated in relevant part, 686 Fed.Appx. 889 , 2017 WL 1906904 (Fed. Cir. May 3, 2017) (per curiam). Other circuits have adhered to or adopted the rule that willfulness is one piece of the puzzle. See Synergistic Int’l, LLC v. Korman, 470 F.3d 162 , 175 & n.13 (4th Cir. 2006); Quick Techs., Inc. v. Sage Grp. PLC, 313 F.3d 338, 347-49 (5th Cir. 2002). The Third Circuit switched sides, concluding that the 1999 amendment upended its precedent requiring willfulness. Banjo Buddies, 399 F.3d at 175 .

We have not yet addressed this question. See Fifty-Six Hope Rd. Music, Ltd. v. A.V.E.L.A., Inc., 778 F.3d 1059, 1073-74 (9th Cir.) (omitting substantive analysis of the effect of the 1999 amendment but finding sufficient evidence of willfulness), cert. denied, — U.S.-, 136 S.Ct. 410 , 193 L.Ed.2d 317 (2015). We now decide that the 1999 amendment does not change the foundation of Ninth Circuit precedent— willfulness remains a prerequisite for awarding a defendant’s profits.

Our conclusion is consistent with the Federal Circuit’s analysis. See Romag Fasteners, 817 F.3d at 791 . We agree with its approach to start with the history of the amendment and thoroughly examine the context in which the amendment came to be. See id. at 785-91 . Several circuits have ruled the other way without looking at the backstory of the remedies provision. See Synergistic Int’l, 470 F.3d at 175 & n.13; Banjo Buddies, 399 F.3d at 173-75 ; Quick Techs., - 313 F.3d at 346-49 . That history is illuminating and reveals why the 1999 amendment does not upend our prior interpretation of the remaining language in § 1117(a).

The history of enactment convincingly shows that the 1999 amendment was intended only to correct a conspicuous drafting error in the 1996 version of the remedies provision. The legislative history bolsters the view that Congress’s sole purpose was to correct the mistaken omission of willful violations of the dilution statute, § 1125(c), from the remedies provision, § 1117(a). H.R. Rep. No. 106-250, at 4 (1999) (“[The amendment] seeks to clarify that ... Congress did intend to allow for ... damages against a defendant found to have wilfully [sic] intended to engage in commercial activity that would cause dilution of a famous mark.”).

Equally important is what Congress changed. Congress created a new predicate — namely, a willful violation of § 1025(c) — that permits monetary recovery. But it did not touch the other language in § 1117(a), which has consistently provided for an award of defendant’s profits under the “principles of equity.” Our holding in Lindy Pen — that a plaintiff can secure the defendant’s profits only after establishing willfulness — is based entirely on an interpretation of that unaltered language. 982 F.2d at 1405-06 .

Thus, it would be a mistake to draw a negative implication from the unrelated and later-introduced language that the amendment somehow negated our circuit’s well-settled willfulness requirement. See Russello v. United States, 464 U.S. 16, 23 , 104 S.Ct. 296 , 78 L.Ed.2d 17 (1983); Anto-nin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts 331 (2012) (“A clear, authoritative judicial holding on the meaning of a particular provision should not be cast in doubt and subjected to challenge whenever a related though not utterly inconsistent provision is adopted in the same statute....”). As McCarthy has explained, it would be improper to “leverage[ ] this statutory change beyond its intended scope in order to adjust the equities in ordinary infringement cases.” 5 McCarthy, supra, § 30:62. This conclusion has added force because we see no indication that the legislature meant to take sides in the entrenched circuit split on willfulness. See Jama v. Immigration & Customs Enf't, 543 U.S. 335, 349 , 125 S.Ct. 694 , 160 L.Ed.2d 708 (2005) (rejecting congressional ratification where there was no “judicial consensus so broad and unquestioned that we must presume Congress knew of and endorsed it”).

For these reasons, the district court properly ruled that Stone Creek must show intentional or willful infringement before disgorgement of Omnia’s profits could be awarded. Because the court denied summary judgment based on a triable issue of fact about whether Omnia infringed, the court had no need to reach the question of willfulness. We note that many of the factual findings that the court has already made — including those on Omnia’s intent in selecting and using the STONE CREEK mark — may be relevant to willful ness. See 4 McCarthy, supra, § 23:112. However, we decline Stone Creek’s invitation to rule that Omnia’s infringement was willful as a matter of law and instead remand for the district court to make this determination.

IY. Sanctions Orders

The district court, relying on 28 U.S.C. § 1927 , twice sanctioned Stone Creek’s attorneys. Section 1927 permits sanctions against an attorney who “multiplies the proceedings in any case unreasonably and vexatiously” and tailors the amount awarded to the costs and fees “reasonably incurred because of such conduct.” Without more, reckless, but nonfrivolous, filings may not be sanctioned. B.K.B. v. Maui Police Dep’t, 276 F.3d 1091, 1107 (9th Cir. 2002). The district court’s first sanctions order runs afoul of that rule, but the second order falls well within the court’s discretion. See Pac. Harbor Capital, Inc. v. Carnival Air Lines, Inc., 210 F.3d 1112, 1117 (9th Cir. 2000) (reviewing for abuse of discretion).

The court sanctioned Stone Creek’s attorneys for filing a summary judgment motion on willfulness, reasoning that summary judgment was not an appropriate vehicle for Stone Creek to request a legal ruling that willfulness is not required for a disgorgement of profits. The court stated that it could not rule on willfulness until it had ruled on infringement and held that Stone Creek’s arguments were frivolous on the merits because Fifty-Six Hope forecloses Stone Creek’s argument that, after the 1999 amendment, a showing of willfulness is not necessary for disgorgement.

On the latter point, the district court was incorrect as a matter of law. Fifty-Six Hope does not address the effect of the 1999 amendment on the continuing vitality of the willfulness requirement. 778 F.3d at 1073-74 . Although the 1999 amendment is referenced in a citation, there is no analysis or determination about the import of the amendment. Id. at 1073 . Instead, the court in Fifty-Six Hope took willfulness as a given and did not need to go further because willfulness was adequately established as a factual matter. Id. at 1074 . Importantly, at the time that Stone Creek filed its motion, the interplay between the amendment and the prior version of the statute remained an open question. This point is underscored in the Ninth Circuit’s Model Jury Instructions: “The Ninth Circuit has not addressed ... whether willfulness remained a prerequisite to disgorgement of a defendant’s profits as a result of the Trademark Amendments Act of 1999.” Manual of Model Civil Jury Instructions for the Ninth Circuit 15.29 cmt. (2017).

Although we ultimately disagree with Stone Creek on the merits of this issue, its arguments were not frivolous. The unsettled nature of the question in our circuit provided Stone Creek with a legitimate basis to ask the district court for a legal ruling — namely, to determine whether to present evidence of willfulness and Omnia’s profits, in addition to Stone Creek’s damages, at trial. The unresolved legal issue, combined with the fact that another circuit had accepted the argument that the 1999 amendment did away with the willfulness requirement, see Banjo Buddies, 399 F.3d at 175 , legitimizes Stone Creek’s arguments. See W. Sys., Inc. v. Ulloa, 958 F.2d 864, 873-74 (9th Cir. 1992). We reverse the sanctions order related to the willfulness issue.

On the other hand, the court’s second sanctions order reflects a discretionary judgment adequately grounded in the law and record. The court sanctioned Stone Creek for not earlier dropping its actual damages claim when it intended to pursue only Omnia’s profits. As the court described, Stone Creek had no evidence to support an actual damages claim and, with the knowledge that “its actual damages claim was meritless,” Stone Creek failed to withdraw the claim and opposed Omnia’s motion to strike the claim.

It is true that actual damages and defendant’s profits are two distinct and well-recognized remedies available to the plaintiff. 15 U.S.C. § 1117 (a)(1) — (2); 5 McCarthy, swpra, § 30:57. The district court’s order does not offend that principle. The court did not question Stone Creek’s right to pursue actual damages as an appropriate avenue of recovery; instead, the court determined that Stone Creek had not actually done so.

As the district court explained, while Stone Creek’s expert finally determined that proving actual damages would be “too difficult,” Stone Creek never provided any analysis from its expert on actual damages or identified what information it was seeking during discovery to shed light on the issue. Stone Creek’s expert acknowledged in his deposition that he had not been asked to perform a damages analysis or calculation. With the actual damages claim still on the table, Omnia was forced to expend time and resources defending against the claim by, for example, taking depositions and having its expert prepare a report. When confronted with Omnia’s motion to strike the actual damages claim before trial, Stone Creek opposed that effort even though it could point to no evidence to support the claim. The district court acted within its discretion in concluding that Stone Creek’s attorneys “unreasonably and vexatiously” “multiplie[d] the proceedings” and awarding Omnia the resulting attorneys’ fees.

Conclusion

We reverse the district court’s finding that there is no likelihood of confusion and reject the application of the Tea Rose-Rectanus defense. Thus, we hold that Om-nia is liable for infringement of the STONE CREEK mark. We affirm the district court’s conclusion that willfulness remains a necessary condition for a disgorgement of profits but remand for a determination on whether Omnia had the requisite intent. Finally, we reverse the district court’s sanctions order with respect to Stone Creek’s summary judgment on willfulness but uphold the sanctions order with respect to Stone Creek’s continued assertion of the actual damages claim.

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.

Costs on appeal shall be awarded to Stone Creek, Inc.

. We refer to the Midwest for ease of reference. The relevant states covered are Illinois, Michigan, Ohio, Pennsylvania, and Wisconsin. . Earlier in the case, the district court rejected Omnia’s Tea Rose-Rectanus defense based on Omnia’s lack of good faith. . To reach its conclusion, the court relied on survey evidence indicating limited awareness of the Stone Creek brand in the Midwest. Because we ultimately conclude that Stone Creek has established a likelihood of confusion, we need not reach Stone Creek’s challenges to the admission of Omnia’s expert and surveys. . We see ho reason to modify the applicability of the presumption simply because Stone Creek had not yet federally registered its mark at the time that Omnia adopted the mark. Omnia’s understanding of the scope of Stone Creek’s rights is relevant to rebutting the presumption, but it does not undercut the rationale for applying the presumption in the first place. . The irony, of course, is that the creators, who started the business in New York, invented the fanciful mark because they thought there would be cache in the foreignness of the mark. As it turned out, they were prescient about market identification because customers' thought the ice cream came from Scandinavia. See Alison Spiegel, Haagen-Dazs Doesn’t Come From Where You Think It Comes From, HuffPost (May 13, 2015, 7:00 AM), http://www.huffingtonpost.com/2015/05/13/ haagen-dazs-comes-from_n_7266208.html. . In 1999, Congress also passed legislation directed at preventing cyberpiracy related to trademarks and domain names. 15 U.S.C. § 1125 (d). This subsection of the statute— often referred to as the cybersquatting provision — provides further evidence that Congress did not intend a wholesale revision of the remedies provision. In particular, while the remedies provision was amended to list violations of the cybersquatting provision (§ 1125(d)) and willful violations of the dilution provision (§ 1125(c)), the former -requires "bad faith intent” for all forms of relief, whereas the latter requires “willfull[] inten[t]” only for monetary relief. Compare 15 U.S.C. § 1125 (d)(1)(A)(i), with id. § 1125(c)(1), (c)(5)(B).