BORINQUEN BISCUIT CORP., Plaintiff, Appellee, v. M.V. TRADING CORP., Defendant, Appellant, 443 F.3d 112

SELYA, Circuit Judge. After an evidentiary hearing, the district court preliminarily enjoined defendant-appellant M.V. Trading Corp. (M.V.) from advertising, distributing, or selling cookies or crackers in Puerto Rico under the trade name “Ricas” on the ground that such activities would likely infringe a registered trademark held by plaintiff-appellee Borin-quen Biscuit Corp. (Borinquen). M.V. appeals. See 28 U.S.C. § 1292(a)(1). We affirm. I.

BACKGROUND Borinquen is a manufacturer and distributor of “galletas.” There is no precise English equivalent of this Spanish word; “galleta” (or, in the plural, “galletas”) encompasses all types of crackers, cookies, and biscuits. Since 1976, Borinquen has sold a round, yellowish, semi-sweet galleta in Puerto Rico under the federally registered trademark “RICA.” This galleta resembles a cookie. Borinquen acquired the recipe for the product and the rights to the mark from Sunland Biscuit Company (Sun-land), which had sold the galleta in Puerto Rico under that mark since 1962.

Sunland officially registered the mark on the principal register of the Patent and Trademark Office (PTO) in 1969. The federal registration states that “the Spanish word ‘Rica’ may be translated as ‘rich.’ ” Borinquen’s “RICA” has always borne a logo that consists of a red circle encompassing the white-lettered phrase “Galletas RICA Sunland.” Borinquen registered both the mark “RICA” and the product’s logo with the Puerto Rico Department of State in 2000.

It currently sells the product in predominately red-and-white packaging, with the circular logo centered against a background consisting of rows of the galletas. While other firms have registered “rica” trademarks for different kinds of products (e.g., bananas, brown sugar, and tobacco), Borinquen’s “RICA” is the only cookie, cracker, or biscuit registered under that name in the United States. Moreover, until the events at issue here, Borinquen was the only company to use the word “rica” in connection with the marketing or distribution of galletas in Puerto Rico.

The tectonic plates shifted in April of 2003, when M.V. began selling a round, yellowish, salty galleta bearing the name “Nestlé Ricas.” This galleta, which resembles a cracker, is manufactured by Nestlé Ecuador and imported by M.V. The product logo consists of a white oval with the name “Ricas” centered in red letters and with a red square in the upper right-hand corner of the oval bearing the white-lettered brand name “Nestlé.” M.V.’s packaging is mostly red and white, albeit with some yellow and blue design. The logo is centered in the upper half of the box against a background of scattered galletas.

In or around the summer of 2004, Borin-quen learned that M.V. was marketing Nestlé Ricas in earnest. Since both parties’ galletas were being sold in Puerto Rican supermarkets and convenience stores, this was a matter of considerable concern. Borinquen informed M.V. that it believed M.V.’s distribution of galletas under the name “Ricas” infringed its registered trademark and asked M.V. to cease and desist. When M.V. refused, Borinquen filed suit for damages and injunctive relief in the federal district court.

In its complaint, Borinquen alleged that M.V.’s use of the “Ricas” mark, coupled with similarities in trade dress, infringed its trademark and trade dress rights under federal law. See 15 U.S.C. §§ 1114(1), 1125(a). M.V. denied the essential allegations of the complaint and counterclaimed for cancellation of Borinquen’s “RICA” mark. Coincident with the institution of suit, Borinquen moved for a preliminary injunction. M.V. opposed that motion. The district court held an evidentiary hearing on May 31 and June 7, 2005. The president of each company testified about the advertising and sales history of his product.

In addition, each side presented expert testimony anent the likelihood of confusion. The expert witnesses approached the problem from different directions. Borin-quen’s witness (an expert in advertising and brand recognition) polled others in his field and concluded that confusion was probable. M.V.’s witness (an expert in market research) conducted a survey of 100 consumers in metropolitan areas and concluded that consumers could recognize differences in the packaging of the two products. On that basis, she hypothesized that the marketing of Nestlé Ricas created no likelihood of confusion.

On September 6, 2005, the district court issued a preliminary injunction that enjoined M.V. from advertising, distributing, or selling any cookies or crackers in Puer-to Rico under the name “Ricas.” Borinquen Biscuit Corp. v. M.V. Trading Corp., No. 04-2070, slip op. at 12 (D.P.R. Sept. 6, 2005) (D.Ct.Op.). The court concluded, inter alia, that Borinquen was likely to succeed in establishing both that its “RICA” mark was entitled to federal trademark protection and that M.V.’s use of the “Ri-cas” mark was likely to foment consumer confusion. This interlocutory appeal ensued. II.

ANALYSIS On appeal, M.V. advances two principal assignments of error. First, it charges that the district court blundered in not requiring Borinquen to establish that the “RICA” mark had acquired secondary meaning. Second, it maintains that the district court mistakenly concluded that M.V.’s product was likely to cause consumer confusion. After delineating the legal standards applicable to grants of preliminary injunctive relief, we address each of these claims. A. The Preliminary Injunction Standard. The preliminary injunction standard is familiar.

Before granting this type of relief, a nisi prius court must consider (1) the likelihood of the movant’s success on the merits; (2) the anticipated incidence of irreparable harm if the injunction is denied; (3) the balance of relevant equities (i.e., the hardship that will befall the nonmovant if the injunction issues contrasted with the hardship that will befall the movant if the injunction does not issue); and (4) the impact, if any, of the court’s action on the public interest. Ross-Simons of Warwick, Inc. v. Baccarat, Inc., 102 F.3d 12, 15 (1st Cir.1996); Narragansett Indian Tribe v.

Guilbert, 934 F.2d 4, 5 (1st Cir. 1991). While all these factors must be weighed, the cynosure of this four-part test is more often than not the movant’s likelihood of success on the merits. See Weaver v. Henderson, 984 F.2d 11, 12 (1st Cir.1993) (“The sine qua non of [the four-factor] formulation is whether the plaintiffs are likely to succeed on the merits.”). The importance of that inquiry is magnified in trademark cases because the resolution of the other three factors will depend in large part on whether the movant is likely to succeed in establishing infringement. See Keds Corp. v.

Renee Int’l Trading Corp., 888 F.2d 215, 220 (1st Cir.1989); see also I.P. Lund Trading ApS v. Kohler Co., 163 F.3d 27, 33 (1st Cir.1998) (noting that irreparable harm can be assumed if a trademark holder demonstrates that it is likely to succeed in establishing infringement). This emphasis on likelihood of success is fully consistent with the tenet that, as a matter of public policy, trademarks should be protected against infringing uses. See Volkswagenwerk Aktiengesellschaft v. Wheeler, 814 F.2d 812, 820 (1st Cir.1987).

A trial court’s conclusions on these factors and its determinations regarding their relative weighting engender deferential appellate review. See Ross-Simons, 102 F.3d at 16. Thus, a decision to grant a preliminary injunction will stand unless the court “mistook the law, clearly erred in its factual assessments, or otherwise abused its discretion in granting the interim relief.” McGuire v. Reilly, 260 F.3d 36, 42 (1st Cir.2001). On appeal, M.V. does not challenge the district court’s determinations regarding the last three components of the four-part preliminary injunction framework.

Consequently, we need consider here only the first of the four factors: likelihood of success. B. Likelihood of Success. Before a party can succeed in an infringement action, it must demonstrate both that its mark merits protection and that the allegedly infringing use is likely to result in consumer confusion. Boston Beer Co. v. Slesar Bros. Brewing Co., 9 F.3d 175, 180 (1st Cir.1993). For purposes of a preliminary injunction, however, the trademark holder need only show a likelihood of success on these two elements.

See Guilbert, 934 F.2d at 6 (explaining that “a court’s conclusions as to the merits of the issues presented on preliminary injunction are to be understood as statements of probable outcomes,” not as final determinations). M.V. claims that Borinquen did not adequately demonstrate below that it was likely to satisfy its burden with respect to either element. We examine each facet of this claim. 1. Eligibility for Trademark Protection. We start with M.V.’s contention that the district court erred, as a matter of law, in determining that Borinquen’s “RICA” mark was entitled to trademark protection.

This contention proceeds from the premise that the court should have required Borin-quen to adduce evidence that its mark had acquired secondary meaning. We reject this premise. It is common ground that, in order to be eligible for trademark protection, a mark must qualify as distinctive. See Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 769, 112 S.Ct. 2753, 120 L.Ed.2d 615 (1992). When considering whether a mark meets that standard, courts often employ a taxonomy that classifies marks along a continuum of increasing distinctiveness.

That continuum contains five categories: generic marks, descriptive marks, suggestive marks, arbitrary marks, and fanciful marks. See, e.g., id. at 768; I.P. Lund, 163 F.3d at 39. By definition, generic marks can never be ranked as distinctive. Park ‘N Fly, Inc. v. Dollar Park & Fly, Inc., 469 U.S. 189, 194, 105 S.Ct. 658, 83 L.Ed.2d 582 (1985). In contrast, suggestive, arbitrary, and fanciful marks are considered inherently distinctive. Two Pesos, 505 U.S. at 768, 112 S.Ct. 2753. The category of descriptive marks is more protean in nature.

Descriptiveness and distinctiveness are neither congruent concepts nor productive of mutually exclusive classifications. Descriptive marks are tentatively considered non-distinctive but can attain distinctive status upon an affirmative showing of secondary meaning. Id. at 769. This showing requires the trademark holder to establish that “in the minds of the public, the primary significance of [the mark] is to identify the source of the product rather than the product itself.” Inwood Labs., Inc. v. Ives Labs., Inc., 456 U.S. 844, 851 n. 11, 102 S.Ct. 2182, 72 L.Ed.2d 606 (1982).

A mark is, therefore, considered distinctive (and, thus, eligible for trademark protection) if it either is inherently distinctive or exhibits acquired distinctiveness gained through secondary meaning. Two Pesos, 505 U.S. at 769, 112 S.Ct. 2753. Both registered and unregistered trademarks may be eligible for protection against infringing uses. See 15 U.S.C. §§ 1114, 1125; see also Keebler Co. v. Rovira Biscuit Corp., 624 F.2d 366, 372 (1st Cir.1980). Withal, the proper mode of analysis regarding whether a mark satisfies the distinctiveness criterion varies based on whether or not the mark is registered.

When a party seeks protection for an unregistered trademark, it is incumbent on that party to demonstrate affirmatively that its mark is distinctive (either inherently or through acquired secondary meaning). Boston Beer, 9 F.3d at 180. The trial court must engage in a specific inquiry to determine whether the proponent has met that burden. See, e.g., I.P. Lund, 163 F.3d at 39-41. Here, however, Borin-quen’s mark is registered, so this requirement does not automatically attach. When a party seeks protection for a registered trademark, its burdens are lighter.

Registration is “prima facie evidence of the validity of the registered mark.” 15 U.S.C. § 1115(a). When, as in this case, the PTO registers a mark without first requiring the applicant to prove secondary meaning, the holder of the mark is entitled “to a presumption that its registered trademark is inherently distinctive, as opposed to merely descriptive.” Equine Techs., Inc. v. Equitechnology, Inc., 68 F.3d 542, 545 (1st Cir.1995); see Quabaug Rubber Co. v. Fabiano Shoe Co., 567 F.2d 154, 161 n. 12 (1st Cir.1977) (explaining that “registration is prima facie evidence that the ... mark is distinctive per se”).

The effect of that presumption varies depending on whether the registered mark has attained incontestable status. If the registered mark has become incontestable through the owner’s compliance with the applicable statutory formalities, the presumption is conclusive. 15 U.S.C. § 1115(b). If, however, the registered mark remains contestable, the putative in-fringer may defend a suit on the ground that the mark does not merit protection because it is not inherently distinctive (i.e., not suggestive, arbitrary, or fanciful) but, rather, is merely descriptive of the product. See id. § 1115(a).

It follows, then, that under the Lanham Act, the effect of registration for a contestable mark is “to shift the burden of proof from the plaintiff ... to the defendant, who must introduce sufficient evidence to rebut the presumption of the plaintiffs right to [exclusive] use” through proof that the mark is merely descriptive. Keebler Co., 624 F.2d at 373. Inasmuch as Borinquen’s mark is registered but contestable (that is, Borin-quen has not satisfied the requirements of 15 U.S.C. § 1065), see supra note 3, this is such a case. That sets the stage for our analysis.

When, as now, a mark is registered but contestable, a putative infringer must offer significantly probative evidence to show that the mark is merely descriptive. Then — and only then — does the burden of proof shift back, so that the party seeking protection must take an alternative route and assume the devoir of persuasion on the issue of whether its mark has acquired secondary meaning. See, e.g., Flexitized, Inc. v. Nat’l Flexitized Corp., 335 F.2d 774, 779-80 (2d Cir.1964).

Simply alleging descriptiveness is insufficient to compel the holder of a registered mark to go forward with proof of secondary meaning; the putative infringer must prove descriptiveness by a preponderance of the evidence before the mix of issues changes and the burden of persuasion on the new issue (secondary meaning) reverts to the trademark holder. See Keebler Co., 624 F.2d at 373. If the putative infringer does hot cross that threshold, the presumption holds and distinctiveness is presumed. See PIC Design Corp. v. Bearings Specialty Co., 436 F.2d 804, 807 (1st Cir. 1971).

In that event, the court can proceed to consider the remaining elements of the infringement claim without first demanding that the trademark holder offer proof of secondary meaning. See id. In the case at hand, these background principles get the grease from the goose. Before us, M.V. insists that the district court erroneously treated Borin-quen’s mark as incontestable because it did not require proof of secondary meaning. This argument conflates two issues and, in the bargain, mischaracterizes what transpired below.

The record, fairly read, supports a determination that M.V. failed to prove, by a preponderance of the evidence, that the “RICA” mark is merely descriptive. Therefore, the district court was warranted in not taking the next step and shifting the burden to Borinquen to show that its registered but contestable mark had acquired secondary meaning. See Keebler Co., 624 F.2d at 373. We explain briefly. The issue of distinctiveness ordinarily presents a question of fact. Equine Techs., 68 F.3d at 544. Although M.V. raised a descriptiveness defense, it presented minimal evidence and argument in support of that defense.

For example, while M.V. conducted a survey on the issue of consumer confusion, it adduced no comparable evidence in support of its contention that Borinquen’s mark is merely descriptive. By like token, M.V. devoted scant attention to the question of descriptiveness in its post-hearing brief. Given this dearth of both evidence and argumentation, we think that the decision to allow the distinctiveness presumption to control withstands scrutiny.

M.V.’s counter-arguments, though ably presented, are not sufficient, at the preliminary injunction stage, to offset the finding that Borinquen enjoyed a likelihood of sur-cess on the merits. First, M.V. states that “RICA” should be dubbed a descriptive mark because the word, translated literally, means “tasty” or “rich” and, therefore, .describes a desirable characteristic of a cookie or cracker.

With respect to a registered mark, however, the putative in-fringer’s burden is not simply to show that the mark describes a feature of the trademark holder’s product; rather, it must show that consumers regard the mark as merely descriptive of that product. See id. at 545. Here, the record contains no such evidence.

This shortcoming is all the more glaring because the first language of the prototypical Puerto Rican consumer is Spanish and, as Borinquen notes, Spanish grammatical rules caution against attributing a purely adjectival meaning to the term “RICA” when viewed in the context of Borinquen’s logo: “Galletas RICA Sunland.” A particular Spanish-language grammatical rule— the rule of concordance — requires strict relationships of gender and number between adjectives and nouns. See Gerald Erichsen, Spanish for Beginners: Noun-Adjective Agreement (2006), http://spanish.about.com/cs/forbeginners/a/adjec-tive_agree.htm.

Hence, Borinquen’s mark would need to use the plural (“RICAS”) if it were intended to serve as a grammatically correct descriptor of the noun “Galle-tas.” We think it follows that the average Spanish-speaking consumer would be unlikely to view the non-concordant mark as a mere descriptor for the underlying product. M.V.’s second counter-argument is equally unconvincing.

It maintains that Borinquen’s mark must be considered descriptive because, in 1964, the Puerto Rico Supreme Court determined that the term “Rico” (the masculine of “Rica”) was merely descriptive of the product “Café Rico” and, thus, not entitled to protection under local law. Cooperativa de Cafeteros v. Colón Colón, 91 P.R.R. 361 (1964). However, the distinctiveness inquiry is product-specific. See Boston Athletic Ass’n v.

Sullivan, 867 F.2d 22, 27 (1st Cir.1989) (explaining that when a party challenges the validity of a registered mark, it must show that the primary significance of the mark to the consuming public is to merely to describe a particular product). A term that merely describes one product may well be distinctive as to another.

The phrase “Café Rico,” as to which the Puerto Rico court found that the adjective merely described the noun, comports with the Spanish-language rule of concordance for nouns and adjectives whereas the phrase “Galletas RICA,” as to which the district court found that the adjective did not merely describe the noun, does not.

Given this fractured grammatical usage, it seems more probable that a Spanish-speaking consumer would find “Rico” merely descriptive of “Café” than that she would find “RICA” merely descriptive of “Galletas.” If more were needed — and we doubt that it is — M.V. adduced considerable evidence of the successful registration of other marks including the word “rica.” Although M.V. introduced this evidence in an apparent effort to show that Borinquen’s mark is weak, it tends to undercut M.V.’s contention that the term is, as a matter of law, merely descriptive of Borinquen’s product.

After all, the PTO requires an applicant to demonstrate that its mark is distinctive before it will approve a registration. See 15 U.S.C. § 1052(e). Courts frequently have accorded weight to these kinds of PTO determinations in evaluating whether a mark is descriptive or inherently distinctive. See, e.g., Lone Star Steakhouse & Saloon, Inc. v. Alpha of Va., Inc., 43 F.3d 922, 934 (4th Cir.1995). Given the PTO’s expertise in the field, we are in-dined to do so here. To that extent, then, the PTO’s acceptance of these other marks for registration supports the idea that “rica” can be an inherently distinctive term.

For these reasons, we conclude that the evidence adduced to date suggests that M.V. is unlikely to rebut the presumption of distinctiveness arising from Borinquen’s contestable mark. Consequently, the district court appropriately refrained from requiring Borinquen to prove secondary meaning. 2. Consumer Confusion. M.V.’s remaining assignment of error attacks the district court’s determination that its use of the “Nestlé Ricas” mark was likely to cause consumer confusion.

This court has enumerated eight factors to guide the inquiry into likelihood of confusion: (1) the similarity of the marks; (2) the similarity of the goods; (3) the relationship between the parties’ channels of trade; (4) the relationship between the parties’ advertising; (5) the classes of prospective purchasers; (6) evidence of actual confusion; (7) the defendant’s intent in adopting its mark; and (8) the strength of the plaintiffs mark. Astra Pharm. Prods., Inc. v. Beckman Instruments, Inc., 718 F.2d 1201, 1205 (1st Cir.1983); accord Pignons S.A. de Mecanique de Precision v.

Polaroid Corp., 657 F.2d 482, 487 (1st Cir.1981). A proper analysis takes cognizance of all eight factors but assigns no single factor dispositive weight. Keds, 888 F.2d at 222; Volkswagenwerk, 814 F.2d at 817. Because likelihood of confusion is a factbound inquiry, appellate review of a trial-level finding on that issue is for clear error. Keds, 888 F.2d at 222. The court below applied the eight-factor screen and concluded that, on the whole, the evidence preponderated in favor of a finding that M.V.’s use of its mark in Puer-to Rico was likely to result in consumer confusion.

This conclusion was predicated on a measured view of the evidence. On the one hand, the court noted that the parties’ goods were dissimilar, that no actual confusion had been shown, and that no evidence existed that M.V. intended to mislead consumers. D. Ct. Op. at 6-9. On the other hand, the court found that the remaining factors all tended to favor a likelihood of confusion. Id. at 5-9. The court then performed the necessary balancing and determined that the latter points outweighed the former. Id. at 9-10.

M.V. assails this determination, arguing that the court misapplied the eight-factor test in two principal ways: by underestimating the significance of Borinquen’s failure to present a consumer survey designed to show actual confusion and by deeming “RICA” a strong mark. After careful per-scrutation of the record, we reject both arguments. To begin, a trademark holder’s burden is to show likelihood of confusion, not actual confusion. While evidence of actual confusion is “often deemed the best evidence of possible future confusion,” Attrezzi, LLC v.

Maytag Corp., 436 F.3d 32, 40 (1st Cir.2006), proof of actual confusion is not essential to finding likelihood of confusion. See Volkswagenwerk, 814 F.2d at 818; see also Pignons, 657 F.2d at 490 (cautioning that “[ejvidence of actual confusion is not invariably necessary to prove likelihood of confusion”). Historically, we have attached substantial weight to a trademark holder’s failure to prove actual confusion only in instances in which'the relevant products have-coexisted on the market for a long period of time. See, e.g., Aktiebolaget Electrolux v.

Armatron Int’l, Inc., 999 F.2d 1, 4 (1st Cir.1993) (finding lack of evidence of actual confusion significant because products had coexisted in the same market for roughly six years). This is not such a case: M.V.’s product was introduced to the Puerto Ri-can market in April of 2003, and M.V.’s own evidence suggests that sales did not proliferate until the summer of 2004. This corresponds to the time frame in which. Borinquen discovered the presence of Nes-tlé Ricas and took action to protect its mark.

Since the preliminary injunction issued just over a year later, there was no protracted period of product coexistence. Nor is there any other compelling reason why, in this case, survey evidence should be required at the preliminary injunction stage. We hold, therefore, that while survey evidence would have been helpful, it was not indispensable to a finding of likelihood of confusion. See id.; see also Volks-ivagenwerk, 814 F.2d at 818 (upholding permanent injunction despite plaintiffs failure to produce evidence of actual confusion).

In much the same vein, the court did not clearly err in determining that “RICA” is a relatively strong mark. Various factors are relevant in ascertaining the strength of a trademark, including the length of time the mark has been used, the trademark holder’s renown in the industry, the potency of the mark in the product field (as measured by the number of similar registered marks), and the trademark holder’s efforts to promote and protect the mark. See Boston Athletic Ass’n, 867 F.2d at 32; see also Keds, 888 F.2d at 222.

In assessing the strength of Borinquen’s mark, the district court found that the mark had been registered for more than three decades; that Borinquen’s “RICA” was the only cookie, cracker, or biscuit trademarked under that name in the United States; and that Borinquen’s efforts in promoting and protecting its mark were in conformance with industry standards. D. •Ct. Op. at 9. These three findings, all of which are supported by substantial evidence in the record, furnish ample grounding for the court’s conclusion that “RICA” should be considered a strong mark. To say more on this issue would be supererogatory.

The short of it is that we discern no clear error in the findings of fact underlying the district court’s determination that Borinquen has a better-than-even chance of succeeding in establishing a likelihood of consumer confusion. III. CONCLUSION We need go no further. Because Borin-quen’s trademark is registered, it enjoys a presumption of inherent distinctiveness. Since M.V. failed adequately to rebut that presumption at the preliminary injunction stage, the district court did not err in determining, without requiring evidence of secondary meaning, that Borinquen’s mark was likely entitled to protection.

Nor did the court clearly err in finding that M.V.’s continued use of its competing mark probably would result in consumer confusion. Consequently, both the court’s determination that Borinquen had established a likelihood of success on the merits of its infringement claim and its related decision to award Borinquen preliminary injunctive relief are unimpugnable. Affirmed. . Borinquen also brought a dilution claim under 15 U.S.C. § 1125(c). That claim was not pressed at the preliminary injunction stage, so we make no further mention of it. .

Even when the PTO's decision to register a mark depends on the applicant’s proof of secondary meaning, the holder is still entitled to a presumption that its mark is distinctive and merits protection. See 15 U.S.C. § 1115(a). In that instance, however, the presumption is not that the mark is inherently distinctive, but, rather, that the mark is valid because it enjoys secondary meaning. See Arrow Fastener Co. v. Stanley Works, 59 F.3d 384, 393 (2d Cir.1995). Because Borinquen's mark was not registered on this basis, we do not address the effect of a secondary meaning presumption. .

A mark becomes incontestable when its owner files an affidavit with the PTO attesting that the following requirements have been met: (

i) there has been no final decision adverse to its ownership or enforcement rights for the preceding five-year period; (ii) there is no pending case or proceeding regarding the owner’s rights in the mark; and (iii) the owner is still using the mark. See 15 U.S.C. § 1065. . The absence of reliable survey evidence is not, in and of itself, dispositive when a putative infringer attempts to establish that a registered mark is not entitled to protection on descriptiveness grounds. The relevant inquiry focuses on the primary significance of the mark to the product's consumers. See Boston Athletic Ass’n v.

Sullivan, 867 F.2d 22, 27 (1st Cir. 1989). The preferred form of proof includes direct evidence (such as consumer surveys) suggesting that the average purchaser actually regards the mark as merely descriptive of the plaintiffs product. See Keebler Co., 624 F.2d at 375. Direct evidence is not, however, an absolute prerequisite. See id. (noting that, in some instances, indirect evidence may suffice). This concept dovetails neatly with the lessened proof requirements that apply at the preliminary injunction stage. .

In the interest of completeness, we note that the record does contain some evidence that bears upon the issue of secondary meaning. However, the view we take of this issue renders superfluous any evaluation of that evidence. . M.V.'s boast that the results of its own survey increased the importance of establishing actual confusion rings hollow. The district court supportably found M.V.’s survey to be of scant evidentiary value.

First, that survey was constructed to determine whether potential customers could spot differences in the products’ packaging, not to identify whether M.V.'s use of its mark created a likelihood of confusion. D. Ct. Op. at 7. Second, the small sample size and large margin of error combined to cast considerable doubt on its statistical integrity. Id.