DUNCAN, Circuit Judge:
The United States Patent and Trademark Office (the "USPTO") and Booking.com ("Booking.com") both appeal the district court's summary judgment ruling regarding the protectability of the proposed trademark BOOKING.COM.
Before we recount the facts of this case, we briefly discuss the legal trademark context in which it arises. Trademark law protects the goodwill represented by particular marks and serves the twin objectives of preventing consumer confusion between products and the sources of those products, on the one hand, and protecting the linguistic commons by preventing exclusive use of terms that represent their common meaning, on the other. OBX-Stock, Inc. v. Bicast, Inc., 558 F.3d 334, 339-40 (4th Cir. 2009).
In order to be protectable, marks must be "distinctive." To determine whether a proposed mark is protectable, courts ascertain the strength of the mark by placing it into one of four categories of distinctiveness, in ascending order: (1) generic,
A term is generic if it is the "common name of a product" or "the genus of which the particular product is a species," such as LITE BEER for light beer, or CONVENIENT STORE for convenience stores. OBX-Stock, Inc., 558 F.3d at 340. Generic terms do not contain source-identifying significance—they do not distinguish the particular product or service from other products or services on the market. George & Co., 575 F.3d at 394. Accordingly, generic terms can never obtain trademark protection, as trademarking a generic term effectively grants the owner a monopoly over a term in common coinage. If protection were allowed, a competitor could not describe his goods or services as what they are. CES Publ'g Corp. v. St. Regis Publ'ns, Inc., 531 F.2d 11, 13 (2d Cir. 1975).
In contrast, descriptive terms, which may be protectable, describe a "function, use, characteristic, size, or intended purpose of the product," such as 5 MINUTE GLUE or KING SIZE MEN'S CLOTHING. Sara Lee Corp. v. Kayser-Roth Corp., 81 F.3d 455, 464 (4th Cir. 1996). In order to be protected, a descriptive term must have acquired secondary meaning. Hunt Masters, Inc. v. Landry's Seafood Rest., Inc., 240 F.3d 251, 254 (4th Cir. 2001). Secondary meaning indicates that a term has become sufficiently distinctive to establish a mental association in the relevant public's minds between the proposed mark and the source of the product or service. George & Co., 575 F.3d at 394.
Against this background, we consider the facts before us.
Booking.com operates a website on which customers can book travel and hotel accommodations. It has used the name BOOKING.COM since at least 2006. In 2011 and 2012, Booking.com filed four trademark applications for the use of BOOKING.COM as a word mark and for stylized versions of the mark with the USPTO. Booking.com sought registration for, inter alia, Class 43 services, which include online hotel reservation services.
The USPTO examiner rejected Booking.com's applications, finding that the marks were not protectable because BOOKING.COM was generic as applied to the relevant services. In the alternative, the USPTO concluded that the marks were merely descriptive and that Booking.com had failed to establish that they had acquired secondary meaning as required for trademark protection. After the examiner denied Booking.com's motion for reconsideration, Booking.com appealed to
The TTAB affirmed the USPTO's four refusals of registration in three separate opinions. These opinions all concluded that BOOKING.COM was a generic term for the services offered, and therefore ineligible for trademark protection, because "booking" generically refers to "a reservation or arrangement to buy a travel ticket or stay in a hotel room" or "the act of reserving such travel or accommodation"; ".com" indicates a commercial website; and consumers would understand the resulting composite BOOKING.COM to primarily refer to an online reservation service for travel, tours, and lodging, which are the services proposed in Booking.com's applications. Booking.com B.V. v. Matal, 278 F.Supp.3d 891, 896 (E.D. Va. 2017) (summarizing the TTAB's findings). In the alternative, the TTAB concluded that BOOKING.COM is merely descriptive of Booking.com's services and that Booking.com had failed to demonstrate that the mark had acquired secondary meaning, as required for trademark protection.
Booking.com appealed the TTAB's decisions by filing this civil action under 15 U.S.C. § 1071(b) against the USPTO and the USPTO's director in the Eastern District of Virginia in April 2016.
The district court held that although "booking" was a generic term for the services identified, BOOKING.COM as a whole was nevertheless a descriptive mark. The district court further determined that Booking.com had met its burden of demonstrating that the proposed mark had acquired secondary meaning, and therefore was protectable, as to the hotel reservation services described in Class 43. The court therefore partially granted Booking.com's motion for summary judgment, ordering the USPTO to register two of the marks and remanded for further administrative proceedings as to the other two.
The USPTO subsequently filed two motions. Pursuant to Federal Rule of Civil Procedure 59(e), the USPTO sought to amend the court's order requiring the USPTO to register the two trademarks, requesting instead that the court remand for further administrative proceedings. It also filed a motion for expenses pursuant to 15 U.S.C. § 1071(b)(3), which would require Booking.com to pay $76,873.61 of the USPTO's expenses under 15 U.S.C. § 1071(b)(3). These expenses included the salaries of the PTO's attorneys and paralegals that worked on the defense action. The district court denied the USPTO's motion to amend as to the two marks, reasoning that they were registerable as trademarks and that no further administrative proceedings were necessary. However, the district court granted the USPTO's motion for expenses. Both the USPTO and Booking.com appealed. The USPTO and Booking.com challenge, respectively, whether BOOKING.COM is protectable, and
We turn first to the USPTO's contention that the district court erred in concluding that BOOKING.COM is a protectable trademark. According to the USPTO, BOOKING.COM is a generic, not a descriptive, term that can never be protected.
We review a district court's grant of summary judgment based on the conclusion that a mark is sufficiently distinctive to warrant trademark protection de novo. Retail Servs. Inc. v. Freebies Publ'g, 364 F.3d 535, 541-42 (4th Cir. 2004). The question of whether a proposed mark is generic is a question of fact that is subject to deferential review. See Swatch AG v. Beehive Wholesale, LLC, 739 F.3d 150, 155 (4th Cir. 2014) (citing Pizzeria Uno Corp. v. Temple, 747 F.2d 1522, 1533 (4th Cir. 1984)).
Critically to our analysis, the USPTO concedes that if BOOKING.COM may properly be deemed descriptive, the district court's finding that it has acquired secondary meaning was warranted. Rather, the USPTO only challenges the district court's determination that BOOKING.COM is not generic. Therefore, the limited question on appeal is whether the district court erred in finding that BOOKING.COM is not generic. For the reasons that follow, we conclude that the district court did not err in finding that there was no genuine issue of material fact on the issue of genericness and that, on these facts, BOOKING.COM is a protectable trademark.
Before undertaking our analysis, two issues pertinent to the genericness inquiry bear further elaboration: first, who bears the burden of proving genericness, and second, the framework for determining whether a proposed mark is generic.
We have never directly addressed the issue of which party bears the burden of proving genericness on appeal when registration of a mark is denied.
We therefore hold here that the USPTO bears the burden of proving that
We next discuss the framework for determining whether a mark is generic. As we have discussed, generic terms are the "common name of a product or service itself." Sara Lee, 81 F.3d at 464. To determine whether a term is generic, we follow a three-step test: (1) identify the class of product or service to which use of the mark is relevant; (2) identify the relevant consuming public; and (3) determine whether the primary significance of the mark to the relevant public is as an indication of the nature of the class of the product or services to which the mark relates, which suggests that it is generic, or an indication of the source or brand, which suggests that it is not generic.
Once a term is deemed generic, it cannot subsequently become non-generic. A term may be generic if, for example, it was previously determined to be generic by a court. Kellogg Co. v. Nat'l Biscuit Co., 305 U.S. 111, 117, 59 S.Ct. 109, 83 S.Ct. 73 (1938) (finding that "shredded wheat" was generic because a court had already deemed it to be so). A term may also be deemed generic where evidence suggests that a term was "commonly used prior to its association with the products [or services] at issue." Hunt Masters, 240 F.3d at 254-55. In such cases of common usage, a court may find that a term is generic even without looking to evidence of consumer recognition. Id. For example, in Hunt Masters, we found that the term "crab house" was commonly used, as there were many restaurants called "crab houses" across the country, id. at 254 n.1, and concluded, therefore, that the district court did not err in declining to consider consumer survey evidence.
If a term is deemed generic, subsequent consumer recognition of the term as brand-specific cannot change that determination. See Retail Servs., Inc., 364 F.3d at 547. Indeed, courts have explained that "no matter how much money and effort the user of a generic term has poured into promoting the sale of its merchandise and what success it has achieved in securing public identification," that user cannot claim the exclusive right through trademark protection to call the product or service by its common name. Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 9 (2d Cir. 1976).
In this case, neither party disputes the district court's finding at the first step: that the mark is used to identify the class or product to which it belongs—here, making hotel reservations for others. Nor do
To ascertain the public's understanding of a term, courts may look to "purchaser testimony, consumer surveys, listings and dictionaries, trade journals, newspapers, and other publications." Glover, 74 F.3d at 59. For example, in determining that the term "ale house" was generic for a facility that serves both food and beer, we considered newspaper articles and restaurant reviews that referred to such facilities as "ale houses," as well as the lack of evidence suggesting that it was not a generic term for such institutions. Ale House Mgmt., Inc. v. Raleigh Ale House, Inc., 205 F.3d 137, 140-41 (4th Cir. 2000).
In assessing whether a term is understood by the relevant public to primarily refer to the service or the source, we look to the proposed mark as a whole, see Hunt Masters, 240 F.3d at 254 ("[A] mark must be considered as a whole to determine its validity."). Even where a proposed mark is a phrase or a compound term, such as "crab house" or "ale house," the relevant inquiry is the public's understanding of the entire mark, not its understanding of the mark's separate components independently. Estate of P.D. Beckwith, Inc. v. Comm'r of Patents, 252 U.S. 538, 545-46, 40 S.Ct. 414, 64 S.Ct. 705 (1920).
With this framework for genericness in mind, we turn to the USPTO's contention on appeal that the district court erred in finding that BOOKING.COM is a descriptive, rather than a generic, mark.
We hold that the district court, in weighing the evidence before it, did not err in finding that the USPTO failed to satisfy its burden of proving that the relevant public understood BOOKING.COM, taken as a whole, to refer to general online hotel reservation services rather than Booking.com the company. Because the USPTO concedes that, if the mark is descriptive, it is protectable, this ends our inquiry. In affirming the district court's finding, we reject the USPTO's contention that adding the top-level domain (a "TLD") .com to a generic second-level domain (an "SLD") like booking can never yield a non-generic mark. We turn first to the district court's finding on genericness before addressing the USPTO's proposed rule.
Genericness is a question of fact to which the district court, as the trier of fact, is accorded great deference.
First, the district court found "highly relevant" the absence of evidence by the USPTO that consumers commonly refer to online hotel reservation services as "bookings.com." Booking.com B.V., 278 F.Supp.3d at 914. Instead, the court determined that the USPTO's evidence demonstrated that such services are referred to as "booking website(s)," or "booking site(s)." Id.
While the USPTO identified other domain names that contain "booking.com"— such as "hotelbooking.com" and "ebooking.com" —to support its argument that the relevant public understands BOOKING.COM to refer to online hotel booking services, the district court did not err in finding this evidence less probative of common usage. It is true that some courts have found the use of a proposed mark in longer domain names to be evidence in support of finding that term generic. See, e.g., Advertise.com, Inc. v. AOL Advertising, Inc., 616 F.3d 974, 980-81 (9th Cir. 2010) (finding that the way in which ADVERTISING.COM was used in other domain names was evidence of genericness); In re Hotels.com, L.P., 573 F.3d 1300, 1304 (Fed. Cir. 2009) (same for HOTELS.COM); In re Reed Elsevier Props., Inc., 482 F.3d 1376, 1380 (Fed. Cir. 2007) (same for LAWYERS.COM). We note, however, that although those courts recognized that the inclusion of the proposed mark in longer domain names was strong evidence of genericness, they nonetheless remained open to considering consumer surveys to determine the public's understanding of the proposed mark. See, e.g., In re Hotels.com, 573 F.3d at 1304-05 (finding that the TTAB did not err in determining that the term was generic, citing in part concerns arising from the methodology of the applicant's consumer survey). Moreover, using the characters "booking.com" or "bookings.com" in a longer domain name does not necessarily mean that BOOKING.COM is generic. Unlike "hotels" or "lawyers," "booking" is used to describe a plethora of reservation services, including, for example, theatrical or musical engagements. See Booking.com B.V., 278 F.Supp.3d at 904 (quoting Random House: Unabridged Dictionary (2d ed. 1993)). Including booking.com in a longer domain name therefore does not necessarily demonstrate that consumers would understand BOOKING.COM to identify any website that provides hotel reservation services. In fact, the record evidence demonstrates the opposite.
The USPTO challenges the court's weighing of this evidence, contending that the district court erred in emphasizing that the public does not use "booking.com" to refer to the relevant services, and instead should have evaluated whether the public would understand the term to refer to those services. We agree with the USPTO that the ultimate inquiry in determining whether a term is generic is what the public understands the proposed mark to mean. Glover, 74 F.3d at 59. Nonetheless, courts have considered usage to be probative of the public's understanding. Compare In re Dial-A-Mattress, 240 F.3d 1341, 1346 (Fed. Cir. 2001) (finding that a mark was not generic where there was no evidence that the relevant public referred to the class of shop-at-home mattress retailers as "1-888-M-A-T-R-E-S-S"), with Frito-Lay N. Am. Inc. v. Princeton Vanguard, LLC, 124 U.S.P.Q.2d 1184 at *6-10 (TTAB 2017) (finding "pretzel crisp" to be generic where the record evidence, including newspaper articles and food blogs, used the term to refer to the genus of snack products rather than a particular brand), and Ale House Mgmt., 205 F.3d at
Second, the district court also considered Booking.com's Teflon survey, which demonstrates that 74.8% of respondents identified BOOKING.COM as a brand name, rather than as a general reference to hotel reservation websites. Such consumer surveys are the "preferred method of proving genericness." Princeton Vanguard, LLC v. Frito-Lay N. Am., Inc., 786 F.3d 960, 970 (Fed. Cir. 2015) (citation omitted). Indeed, courts have recognized that "[c]onsumer surveys have become almost de rigueur in litigation over genericness." Berner Int'l Corp. v. Mars Sales Co., 987 F.2d 975, 982-83 (3d Cir. 1993) (internal quotation marks and citation omitted). Accordingly, where, as here, the district court found that the survey was methodologically sound, the survey is strong evidence that the public does not understand BOOKING.COM to refer to the proposed mark's generic meaning.
On appeal, the USPTO does not contest the validity of the survey or its methodology. Instead, it relies on dicta in Hunt Masters to argue that the district court erred in considering the survey at all. 240 F.3d at 254-55. Its reliance is misplaced; our reasoning in that case does not apply here. In Hunt Masters, we considered whether the owners of "the Charleston Crab House," who sought to enjoin a competitor from using the name "The Crab House," had a protected proprietary interest in the term "crab house." We declined to find such an interest because we determined that "crab house" was a generic term referring to a class of restaurants that serve crabs. Id. at 254. In so determining, we held that the district court did not err in declining to consider the plaintiff's consumer survey. We explained that there are two ways in which terms may be classified as generic—"(1) where the term began life as a `coined term'" that had become generic through common usage, and "(2) where the term was commonly used prior to its association with the products at issue"—and that while consumer surveys are relevant to determining whether a term is generic in the former scenario, they are not in the latter. Id. at 254-55. Contrary to the USPTO's contention, Hunt Masters does not control where, as here, the district court determined based on the dearth of evidence in the record that the proposed mark was not commonly used. As such, the proposed mark does not fall within the category of terms for which survey evidence is irrelevant.
Weighing the evidence before it, the district court did not err in finding that the USPTO did not satisfy its burden of showing that BOOKING.COM is generic. It is axiomatic that determinations regarding the relative weight of evidence are left for the trier of fact. See In re Hotels.com, 573 F.3d at 1305-06 (finding that the trier of fact, the TTAB, could reasonably have given controlling weight to dictionary definitions
We therefore conclude that the district court did not err in finding that BOOKING.COM is a descriptive, rather than generic, mark.
The USPTO nevertheless contends that adding the top-level domain ".com" to a generic second-level domain like "booking" is necessarily generic, and that the district court therefore erred in finding that BOOKING.COM was non-generic. The USPTO advances two theories as reasons for adopting a per se rule against protecting terms like BOOKING.COM. For the reasons that follow, we decline to adopt such an approach under either theory.
First, the USPTO relies on an 1888 Supreme Court case to argue that, as a matter of law, adding .com to a generic SLD like booking can never be nongeneric. In Goodyear's Rubber Mfg. Co. v. Goodyear Rubber Co., 128 U.S. 598, 602-03, 9 S.Ct. 166, 32 S.Ct. 535 (1888), the Court held that the addition of commercial indicators such as "Company" to terms that merely describe classes of goods could not be trademarked, like "Grain Company" or, as the Dissent provides, "The Grocery Store." According to the USPTO, ".com" is analytically indistinct from "company," as it is a generic identifier for an entity operating a commercial website, and therefore its addition to a generic term can never be protected. However, Goodyear was decided almost sixty years before the Lanham Act and, crucially, did not apply the primary significance test. No circuit has adopted the bright line rule for which the USPTO advocates—indeed, sister circuits have found that when ".com" is added to a generic TLD, the mark may be protectable upon a sufficient showing of the public's understanding through consumer surveys or other evidence. See, e.g., Advertise.com, Inc., 616 F.3d at 982; In re Hotels.com, 573 F.3d at 1304-05. We similarly decline to do so here.
Second, the USPTO argues that the proposed mark is per se generic because it is nothing more than the sum of its component parts. It contends that "booking" is a generic term for hotel reservation services, that ".com" is generic for an online company, and that when combined the resulting composite is generic for the online booking services at issue here because a member of the relevant public would understand BOOKING.COM to name an online booking website. Therefore, the USPTO contends, BOOKING.COM is generic. We disagree that it is necessarily so.
We begin by discussing the genericness inquiry as it applies to compound terms. When confronted with a compound term like PRETZEL CRISPS, courts may consider as a first step the meaning of each of the term's component marks; but as we explained in Hunt Masters, the ultimate inquiry examines what the public primarily perceives the term as a whole to refer to. 240 F.3d at 254. For example, in determining whether PRETZEL CRISPS is generic, a court may first determine based on dictionary definitions and other competent sources that PRETZEL is primarily understood to refer to the genus of
Where the proposed mark is a composite that includes .com, we clarify that, contrary to the district court's suggestion,.com does not itself have source-identifying significance when added to an SLD like booking.
For the same reason, neither is it the case, as the USPTO would have it, that assuming booking and .com are each generic terms according to their respective dictionary definitions, and that together they describe the service provided, this necessarily ends the genericness inquiry. Within this inquiry, dictionary definitions, though "relevant and sometimes persuasive" to the genericness inquiry based on the assumption that such definitions generally reflect the public's perception of a word's meaning, are not necessarily dispositive or controlling. Retail Servs., Inc., 364 F.3d at 544-45. Instead, where, as here, the court found that the term was not previously commonly used, it may consider additional evidence like consumer surveys in making its genericness determination.
We therefore decline to adopt a per se rule and conclude that when ".com" is combined with an SLD, even a generic SLD, the resulting composite may be non-generic where evidence demonstrates that the mark's primary significance to the public as a whole is the source, not the product.
This approach comports with that taken by our sister circuits, who have similarly declined to adopt a per se rule against protecting domain names, even where they are formed by combining generic terms with TLDs. See, e.g., Advertise.com Inc., 616 F.3d at 978-79; In re Steelbuilding.com, 415 F.3d 1293, 1299 (Fed. Cir. 2005). These courts have left open the possibility that in "rare circumstances" a TLD may render a term sufficiently distinctive to be protected as a trademark. See In re Steelbuilding.com, 415 F.3d at 1299.
Tellingly, even where courts have found that the individual components of a domain name mark are independently generic, and that when added together the resulting composite merely describes the genus of the service provided, courts still considered other evidence such as consumer surveys in determining whether the mark was generic. For instance, in determining whether ADVERTISING.COM was generic, the Ninth Circuit explained that even though both "advertising" and ".com" were generic, and that ADVERTISING.COM conveyed only the genus of the services offered, it was possible "that consumer surveys or other evidence might ultimately demonstrate that [the] mark is valid and protectable." Advertise.com, Inc., 616 F.3d at 982 (emphasis added); see In re Hotels.com, 573 F.3d at 1304-05 (considering a consumer survey regarding the public's understanding of HOTELS.COM even though it determined that "hotels" and ".com" were independently generic and that the combination did not produce new meaning). While these courts have generally found the resulting composite of adding ".com" to certain SLDs to be generic,
We are not unsympathetic to the USPTO's concerns that granting trademark protection over BOOKING.COM may prevent other companies
In sum, adding ".com" to an SLD can result in a non-generic, descriptive mark upon a showing of primary significance to the relevant public. This is one such case. Based on the record before it, the district court properly found that the USPTO did not meet its burden of proving that "booking.com" is generic. We therefore affirm the court's finding that BOOKING.COM is descriptive. Because the USPTO does not challenge the district court's finding that BOOKING.COM has acquired secondary meaning where the mark is deemed descriptive, we affirm the district court's partial grant of summary judgment finding that BOOKING.COM is protectable as a trademark.
We turn now to Booking.com's contention that it should not be required to pay the USPTO's attorneys fees under 15 U.S.C. § 1071(b)(3). Under the Lanham Act, a dissatisfied trademark applicant may seek review of an adverse ruling on his trademark application either by appealing the USPTO's ruling to the Federal Circuit, 15 U.S.C. § 1071(a)(1), or by commencing a de novo action in a federal district court, id. § 1071(b)(1). If the applicant chooses to appeal to the Federal Circuit, the appeal is taken "on the record" before the USPTO, id. § 1071(a)(4), and the court defers to the USPTO's factual findings unless they are unsupported by substantial evidence. Shammas v. Focarino, 784 F.3d 219, 225 (4th Cir. 2015). In contrast, if he chooses to appeal in a district court, the parties may conduct discovery and submit evidence beyond the record before the USPTO, which the district court reviews de novo as the trier of fact. Id. Crucially, if the applicant decides to challenge the USPTO's ruling in the district court, the applicant must pay "all the expenses of the proceeding ... whether the final decision is in favor of such party or not." 15 U.S.C. § 1071(b)(3) (emphasis added).
Pursuant to this statute, the district court granted the USPTO's motion requiring
In so holding in Shammas, we first concluded that the "American Rule"—the bedrock principle that each litigant pays his own attorneys fees unless Congress has specifically and explicitly provided otherwise —was inapplicable to the provision because the rule applies "only where the award of attorneys fees turns on whether a party seeking fees has prevailed to at least some degree." 784 F.3d at 223. Accordingly, we interpreted the phrase "all the expenses of the proceeding" for "its ordinary meaning without regard to the American Rule," and concluded that it included attorneys fees. Id. at 224.
Whether the American Rule applies to § 1071(b)(3), however, has since been called into question. Relying on our decision in Shammas, the Federal Circuit previously held that a nearly identical provision of the Patent Act, 35 U.S.C. § 145, included attorneys fees. Nantkwest, Inc. v. Matal, 860 F.3d 1352, 1355 (Fed. Cir. 2017). Subsequently, however, the Federal Circuit reversed its decision en banc, squarely rejecting our reasoning in Shammas; it now holds that attorneys fees are not covered under that provision. Nantkwest, Inc. v. Iancu, 898 F.3d 1177, 1185 (Fed. Cir. 2018) (en banc). Moreover, the year after we decided Shammas, the Supreme Court applied the American Rule to a bankruptcy statute that did not mention a prevailing party. See Baker Botts L.L.P. v. ASARCO LLC, ___ U.S. ___, 135 S.Ct. 2158, 2165, 192 L.Ed.2d 208 (2015).
These subsequent developments suggest that the American Rule's requirement that Congress "clearly and directly" express an intent to deviate from that rule may apply to § 1071(b)(3)—a statute that, if read to include attorneys fees, anomalously requires an appealing party to pay the prorated salaries of government attorneys.
Nonetheless, Shammas remains the law in this circuit, and as long as we continue to be bound by that precedent we must affirm the district court's grant of attorneys fees.
For the foregoing reasons, we affirm the district court's partial grant of summary judgment to Booking.com as to the protectability of its trademark applications, and we affirm the district court's grant of the USPTO's motion for expenses.
WYNN, Circuit Judge, concurring in part and dissenting in part:
This case addresses a problem that Booking.com chose to bring upon itself. Because trademark law does not protect generic terms, an online business, like Booking.com, has two options in choosing its domain name. On the one hand, it can choose to operate under a generic domain
Booking.com chose the former approach —to operate under a generic domain name and forego the ability to exclude competitors from using close variants of its domain name. But in the face of recognizing that "booking" is a generic term, the district court creatively decided that combining the generic term "booking" with the generic top-level domain ".com" rendered it non-generic. In doing so, the district court's judgment—which the majority opinion concedes was grounded in legal error, but nonetheless declines to set aside— allows Booking.com to have its cake and eat it too. Booking.com gets to operate under a domain that merely describes the nature of its business and exclude its competitors from doing the same.
Booking.com maintains that such a result is warranted to prevent "unscrupulous competitors [from] prey[ing] on its millions of loyal consumers," Appellee's Br. at 38. But to the extent Booking.com fears that its competitors are using the terms "booking" and ".com" in ways that might confuse its customers, "this is the peril of attempting to build a brand around a generic term." Advertise.com, Inc. v. AOL Advert., Inc., 616 F.3d 974, 980 n.6 (9th Cir. 2010). Accordingly, although I agree with much of the analysis in the majority opinion,
That decision rests upon my colleagues' determination that the district court's factual findings pertaining to genericness should be "accorded great deference," Ante at 181, regarding whether a proposed
Here, the district court rendered the legal "conclu[sion]" that "when combined with a[ ] [Secondary Level Domain], a [Top Level Domain] generally has source identifying significance and the combination of a generic [Secondary Level Domain] and a [Top Level Domain] is generally a descriptive mark that is protectable upon a showing of acquired distinctiveness." Booking.com B.V. v. Matal, 278 F.Supp.3d 891, 909 (E.D. Va. 2017). Put differently, the district court concluded that, as a matter of law, "the combination of a generic [Secondary Level Domain] and a [Top Level Domain]" is presumptively descriptive and protectable upon a showing of acquired distinctiveness. Id.
Notably, in adopting this presumption, the district court expressly rejected the approach taken by the Federal Circuit in cases, like the instant case, involving a proposed trademark that combines a generic Secondary Level Domain with a Top Level Domain. See id. at 908-10. The majority opinion rejects that legal presumption but nevertheless defers to the district court's factual finding that BOOKING.COM is descriptive, and therefore protectable, on the theory that that finding was not tainted by the district court's legal error. See Ante at 185 n.9. But a close examination of the district court's opinion reveals that the district court's legal error did play a role in the court's ultimate determination that BOOKING.COM is descriptive.
In particular, because the district court presumed that "the combination of a generic [Secondary Level Domain] and a [Top Level Domain]" is descriptive, Booking.com, 278 F.Supp.3d at 909, it subjected the registrant to a less onerous evidentiary burden for establishing descriptiveness than the law demands. This is evident for two reasons.
First, the district court's ultimate determination —that the proposed mark BOOKING.COM is descriptive—conflicts with the determination that every other court has reached in cases, like the instant case, involving the registration or enforcement of a proposed mark composed of a generic Secondary Level Domain and a Top Level Domain.
Likewise, in In re Reed Elsevier Properties, the Federal Circuit found it to be "abundantly clear" that the proposed mark in question, "LAWYERS.COM," was generic because "the relevant public would readily understand the term to identify a commercial web site providing access to and information about lawyers." 482 F.3d 1376, 1379-80 (Fed. Cir. 2007) (internal quotation marks omitted).
Similarly, in In re 1800Mattress.com IP, LLC, the Federal Circuit affirmed the Trademark Board's decision that the proposed mark MATTRESS.COM was generic because customers would naturally recognize the mark as referring to a commercial website providing retail services featuring mattresses. 586 F.3d 1359, 1364 (Fed. Cir. 2009). Notably, the Federal Circuit rejected the applicant's argument —which Booking.com also advances in this case and on which the district court relied, see Booking.com, 278 F.Supp.3d at 914—that the term could not be generic because consumers did not refer to the stores as "mattresses.com[s]," id. (quoting In re 1800Mattress.com IP, LLC, 586 F.3d at 1362). Instead, the court agreed with the Trademark Board that the relevant public would understand MATTRESS.COM to be "no more than the sum of its constituent parts"—an online provider of mattresses. Id. at 1363.
Like the Federal Circuit, the Ninth Circuit has found that a generic Secondary Level Domain combined with a Top Level Domain does not generally amount to a protectable mark. In Advertise.com, the court considered whether AOL's claimed mark ADVERTISING.COM was protectable. 616 F.3d at 977. The district court below enjoined Advertise.com, a putative competitor of AOL's Advertising.com, from
Lower courts have followed Hotels.com, Reed-Elsevier Properties, 1800Mattress.com, and Advertising.com and refused to award trademark significance to proposed marks, like BOOKING.COM, combining a generic Secondary Level Domain with a Top Level Domain. See, e.g., Borescopes R U.S. v. 1800Endoscope.com, LLC, 728 F.Supp.2d 938, 952 (M.D. Tenn. 2010) (finding that the use of the term "borescopes" in companies' domain names "generically describes the class of product each sells"); Image Online Design, Inc. v. Core Ass'n, 120 F.Supp.2d 870, 878 (C.D. Cal. 2000) (noting that "a [Top Level Domain] and other non-distinctive modifiers of a URL like `http://www' have no trademark significance"); In Re Eddie Zs Blinds & Drapery, Inc., 74 U.S.P.Q.2d 1037 at *6 (T.T.A.B. 2005) (determining that BLINDSANDDRAPERY.COM was not one of the "exceptional circumstances" that should forestall a finding of genericness); In Re Martin Container, Inc., 65 U.S.P.Q.2d 1058 at *5 (T.T.A.B. 2002), appeal dismissed, 56 Fed. App'x. 491 (Fed. Cir. 2003) (finding CONTAINER.COM generic because the combination of "container" and a Top Level Domain "does not result in a compound term that has somehow acquired" the ability to function as an indication of source).
A second indication that the district court's incorrect legal framework tainted its ultimate determination that BOOKING.COM is protectable is that BOOKING.COM is not like the types of marks courts have recognized as among the "rare" occasions when adding a generic Secondary Level Domain to a Top Level Domain does not create a generic mark. Because Top Level Domains generally convey only that a business is web-based, it is "only in rare circumstances" that the combination of a generic Secondary Level Domain, on its own incapable of source identification, and ".com" will produce a composite mark that "expand[s] the meaning of the mark." See, e.g., Advertise.com, 616 F.3d at 979 (quoting In re Steelbuilding.com, 415 F.3d 1293, 1299 (Fed. Cir. 2005)).
As an example of the "rare circumstance" in which a Top Level Domain could provide additional non-functional information, the Federal Circuit considered the possibility of a brick-and-mortar company that sold tennis equipment and operated under the name "tennis.net." In re Oppedahl & Larson, LLP, 373 F.3d 1171, 1175 (Fed. Cir. 2004). The court noted the "witty double entendre related to tennis nets" created by the combination of "tennis" and ".net" rendered such a usage distinguishable from a standard proposed mark combining a generic Secondary Level Domain with a Top Level Domain. Id.
Following Oppedahl's reasoning, if we were presented with a situation in which the ".com" was used in a way that played upon or expanded the meaning of its ordinary use as a Top Level Domain, registration might be appropriate. For instance, a
The tennis.net and rom.com examples are readily distinguishable from the instant case. The district court did not find— nor has Booking.com ever argued—that the combination of "booking"—a generic term—and ".com" amounts to a "witty double entendre" that expands the ordinary meanings of the proposed mark's component parts. Rather, BOOKING.COM is a run-of-the-mill combination of a generic term with a Top Level Domain that creates a composite mark concerning the subject or business encompassed by the generic term—precisely the type of mark that the courts in Hotels.com, Reed Elsevier Properties, 1800Mattress.com, and Advertise.com found did not amount to the "rare circumstance" that warranted affording the domain name trademark protection.
BOOKING.COM's mark differs substantially from the types of proposed marks fitting into the "rare circumstances" in which a generic Secondary Level Domain and a Top Level Domain are protectable. This provides further evidence that the district court's incorrect legal test tainted its ultimate factual determination. Because the district court erroneously believed that marks combining a generic Secondary Level Domain and Top Level Domain are presumptively protectable, it never examined —as it should have—whether BOOKING.COM amounted to one of the "rare circumstances" when such marks are protectable.
The district court's incorrect legal test— the application of which my colleagues leave in place—upsets the careful balance the law has struck between assisting consumers to identify the source of goods and preserving the linguistic commons. Presumptively allowing protection of domain names composed of a generic Secondary Level Domain and Top Level Domain conflicts with the law's longstanding refusal to permit registration of generic terms as trademarks.
Fundamentally, the proscription against allowing generic terms to be trademarked stems from considerations regarding the monopolization of language. To permit generic terms to be trademarked "would grant the owner of the mark a monopoly, since a competitor could not describe his goods as what they are." CES Publ'g Corp. v. St. Regis Publ'ns, Inc., 531 F.2d 11, 13 (2d Cir. 1975). This Court has long sought to foreclose such a result, holding that no single competitor has the right to "corner the market" on ordinary words and phrases, thereby enclosing the "public linguistic commons." Am. Online, Inc. v. AT&T Corp., 243 F.3d 812, 821 (4th Cir. 2001); see, e.g., Ashley Furniture Indus. v. Sangiacomo N.A., 187 F.3d 363, 369 (4th Cir. 1999) (noting that the trade name "Pet
Importantly, the law forbids trademarking generic terms, even when a putative mark holder engages in successful efforts to establish consumer recognition of an otherwise generic term. "[N]o matter how much money and effort the user of a generic term has poured into promoting the sale of its merchandise and what success it has achieved in securing public identification, it cannot deprive competing manufacturers of the product of the right to call an article by its name." Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 9 (2d Cir. 1976) (emphasis added). Therefore, even advertising, repeated use, and consumer association will not warrant affording trademark protection to a generic term. See Am. Online, 243 F.3d at 821 ("[T]he repeated use of ordinary words ... cannot give [a single company] a proprietary right over those words, even if an association develops between the words and [that company]."); see also Surgicenters of Am., Inc. v. Med. Dental Surgeries, Co., 601 F.2d 1011, 1017 (9th Cir. 1979) (finding that even if a generic term becomes identified with a first user, trademark protection will not be available in the generic term).
The policy considerations underlying trademark law's refusal to protect generic terms apply regardless of whether the putative mark holder is a traditional brick-and-mortar business or located in cyberspace. As the PTO argues, a grocery business called The Grocery Store would—and should—never receive trademark protection because the name is generic —regardless of whether consumers associated the name with a particular entity. However, under the district court's erroneous approach—the consequences of which the majority leaves in place—if enough consumers recognized an online grocery business called "grocerystore.com," that business would be entitled to trademark protection. There is no basis in law or policy for drawing such a distinction. Just as competing brick-and-mortar grocery stores need to make use of the term "grocery" to inform customers of the nature of their businesses, so too do competing internet grocery providers need to make use of the term "grocery" in their domain names.
The district court's treatment of marks combining generic Secondary Level Domains and Top Level Domains as presumptively protectable also conflicts with precedent foreclosing protection of marks that combine a generic term with a generic commercial designation (e.g., "Company," "Corp.," or "Inc."). More than a hundred years ago, the Supreme Court held that the addition of commercial designations like these does not transform otherwise generic terms into protectable marks. See Goodyear's India Rubber Glove Mfg. Co. v. Goodyear Rubber Co., 128 U.S. 598, 602-03, 9 S.Ct. 166, 32 S.Ct. 535 (1888).
Under Goodyear's, a brick-and-mortar reservation service operating under the term The Booking Company would not be able to receive a trademark in that name, regardless of the degree of brand recognition it generated or the number of consumers who identified it as a brand. See Am. Online, 243 F.3d at 821; Abercrombie, 537 F.2d at 9. Yet, under the district court's approach, the term BOOKING.COM is presumptively protectable. Compare Advertise.com, 616 F.3d at 982 ("That `.com,' when added to a generic term, `indicates a commercial entity' does not suffice to establish that the composite is distinctive, much as AOL would not have created a
The district court concluded that "Goodyear's reasoning regarding corporate designators does not apply with equal force to domain names" because "adding a [Top Level Domain] such as `.com' to a generic [Secondary Level Domain] does more than indicate that a company offers services via the internet; it indicates a unique domain name that can only be owned by one entity." Booking.com, 278 F.Supp.3d at 910. But as the Federal Circuit has explained, "[a]lthough not a perfect analogy, the comparison of [Top Level Domains] (i.e., `.com,' `.org,' etc.) to entity designations such as `Corp.' and `Inc.' has merit." Oppedahl, 373 F.3d at 1175. "The commercial impression created by `.com' is similar to the impression created by `Corp.' and `Co.', that is, the association of a commercial entity with the mark." Id.
Additionally, the difference between an entity designation such as "Corp." and a Top Level Domain recognized by the district court is attributable to the functional nature of the internet. Although most Top Level Domains do suggest a relationship with the internet, a domain name "serves the purely technological function of locating a Web site in cyberspace." McCarthy on Trademarks, § 7:17.50. Functional features, however, cannot be the basis for trademark protection: "[E]ven if a functional feature has achieved consumer recognition (secondary meaning) of that feature as an indication of origin, the feature cannot serve as a legally protectable symbol." Am. Online, Inc., 243 F.3d at 822-23. Yet that is precisely what the district court's legal test—the application of which my colleagues in the majority leave in place—does.
It is particularly important that we ensure that the district court's ultimate finding that BOOKING.COM is descriptive, and therefore protectable, was not tainted by its erroneous legal test because, as the PTO argues, trademark registration will provide Booking.com with a weapon to freeze out potential competitors. As explained above, trademark law's proscription on the registration of generic terms prevents firms from monopolizing language and allows competitors to "describe [their] goods as what they are." CES Publ'g, 531 F.2d at 13.
I believe that this Court's affirmance of the district court's judgment—even as it rejects the district court's legal analysis— unjustifiably empowers Booking.com to monopolize language, thereby enclosing the linguistic commons and adversely affecting competitors in precisely the manner that trademark law seeks to forestall. Put simply, the majority opinion's judgment will directly disadvantage Booking.com's competitors by taking away their ability to use the term "booking" in their own website domain names. Indeed, any competitors that attempt to use the term "booking" will face the risk of a costly, protracted, and uncertain infringement lawsuit.
Booking.com and the majority opinion assert that the potential harm to competitors —and therefore consumers—is minimal because: (1) the doctrine of descriptive fair use will prevent Booking.com from monopolizing the term; (2) the protection of BOOKING.COM extends only to hotel reservations, not to other reservation services; and (3) domain names are unique. I find none of these assertions persuasive.
Descriptive fair use is an affirmative defense that allows a competitor to use words contained in a trademark in their ordinary sense to describe the competitor's own goods or services to consumers. See
For several reasons, I fear that the doctrine of descriptive fair use will provide Booking.com's competitors cold comfort. I am not convinced that the descriptive fair use defense will provide significant protection to competitors using the term "booking" in their own domain names. As the Ninth Circuit has explained, "granting trademark rights over a domain name composed of a generic term and a [Top Level Domain] grants the trademark holder rights over far more intellectual property than the domain name itself." Advertise.com, 616 F.3d at 980. Such trademark protection "would potentially reach almost any use of the generic term in a domain name." Id. at 981 (noting that any one of the thirty-two other domain names containing some form of the word "advertise" would be placed at risk of a lawsuit). Thus, notwithstanding the doctrine of descriptive fair use, a firm that obtains a trademark in a domain name that describes the service the firm provides, like Booking.com, may have the power "to foreclose competitors from using a vast array of simple, easy to remember domain names and designations that describe [such] services." Id.
Additionally, descriptive fair use is an affirmative defense, not an immunity from suit. Any online reservation business that chooses to include "booking" in its domain name will face the risk of defending an expensive infringement lawsuit. As a result, commercial competitors seeking to avoid litigation risk and expenses—even if they might ultimately prevail on descriptive fair use grounds—will be chilled from using the term.
My colleagues in the majority also emphasize that the district court awarded Booking.com protection for the mark only for Class 43 hotel services, not Class 39 travel agency services, meaning that "protection over BOOKING.COM would not necessarily preclude another company from using, for example, carbooking.com or flightbooking.com." Ante at 187 (emphasis added). But even assuming my colleagues' optimism that Booking.com could not preclude competitors from using domain names containing the word "booking" is well-founded—a decidedly uncertain question, see Advertise.com, 616 F.3d at 980-81—I see no reason why Booking.com should be entitled to monopolize the generic term "booking" in the online hotel reservation industry by precluding competitors from using domain names like hotelbooking.com or ehotelbooking.com,
Finally, my colleagues in the majority seek to assuage concerns that granting trademark protection over BOOKING.COM will prevent other companies from using variants of the mark by emphasizing that, to enforce the mark, Booking.com will have to prove that there is a "likelihood of confusion" with the allegedly infringing mark. Ante at 187. According to the majority, "[g]iven that domain names are unique by nature and that the public may understand a domain name as indicating a single site, it may be more difficult for domain name plaintiffs to demonstrate a likelihood of confusion." Id. Even accepting my colleagues' assumption that Booking.com will have difficulty proving likelihood of confusion—again, an uncertain question—affording protection to BOOKING.COM would still likely chill competition in the online booking space. Put simply, putative competitors may—and likely will—choose not to operate under domain names that include the word "booking"— even if that term best describes the service they offer—because they do not want to incur the expense and risk of defending an infringement action.
In sum, the district court's opinion reveals that its incorrect understanding of the governing legal framework likely tainted its finding that BOOKING.COM is distinctive, and therefore protectable. Because the district court's erroneous legal test factored into its ultimate factual determination as to descriptiveness, I cannot concur in my colleagues' decision to affirm the district court's judgment. Accordingly, with great respect for my good colleagues in the majority, I dissent.