GOULD, Circuit Judge:

Appellants Falls Media, LLC, Justin Heimberg, and David Gomberg (collectively “Falls Media”) appeal the district court’s summary judgment for Appellee Zobmondo Entertainment, LLC (“Zobmondo”) rejecting Falls Media’s action for trademark infringement, unfair competition, and related claims arising from Zobmondo’s use of Falls Media’s federally registered trademark, “WOULD YOU RATHER ... ?”. The district court held that “WOULD YOU RATHER ... ?” is not entitled to federal trademark protection because the mark is “merely descriptive” and lacks secondary meaning, and the district court ordered the mark can-celled from the federal trademark registry. We have jurisdiction under 28 U.S.C. § 1291. We conclude that there is a genuine issue of material fact whether “WOULD YOU RATHER ... ?” is merely descriptive. Hence we reverse the summary judgment and remand for further proceedings consistent with this opinion.

*1111 I

This appeal involves a persistent dispute between two competitors, Falls Media 1 and Zobmondo, over the right to sell products using the “WOULD YOU RATHER ... ?” mark. Both competitors use similar marks to identify board games and books that incorporate questions posing humorous, bizarre, or undesirable choices. 2 As of 2008, both competitors had established successful businesses earning millions of dollars in revenues from sales of ‘WOULD YOU RATHER ... ?” products.

On July 31, 1997, Falls Media filed an intent-to-use (“ITU”) application with the United States Patent and Trademark Office (“PTO”) stating its intent to use the mark in two product categories: books and games. 3 Falls Media published its first book using the mark, Would You Rather ... ? Over 200 Absolutely Absurd Dilemmas to Ponder, in October of 1997. It published a sequel to the book, Would You Rather 2, Electric Boogaloo, in 1999. These books were carried by retailers including Borders, Barnes & Noble, and Urban Outfitters. To promote the books, Falls Media also established a website, www.wouldyourather.com, which as of 2008 was receiving about 30,000 unique visitors per month. Although the books were not the subject of a paid advertising campaign at the time of their publication, they received unpaid print, radio, and television publicity, including mention in magazines such as Rolling Stone and comment on television shows such as Oprah and The Tonight Show with Jay Leno. Heimberg and Gomberg also made personal appearances to promote their books. As of 2002 — when Zobmondo began using “WOULD YOU RATHER ... ?” as a mark on its board games 4 — Falls Media had sold about 91,000 total copies of its two “WOULD YOU RATHER ... ?” books.

In September of 1997 — shortly after Falls Media filed its ITU application— Zobmondo’s founder, Randall Horn, also filed an ITU application seeking to register the mark “WOULD YOU RATHER.” Horn’s application was rejected by the PTO because of the likelihood of confusion with the mark that was the subject of Falls Media’s prior ITU filing. 5 Nonetheless, in *1112 1998 Zobmondo began producing games using a concept similar to the one adopted by Falls Media, including “Zobmondo!! That Crazy ‘Would You Rather’ Game,” and “Zobmondo!! The Outrageous Game of Bizarre Choices,” the latter of which was licensed to Hasbro between 2000 and 2002. When Hasbro’s license expired in 2002, Zobmondo self-released a new version of the game prominently featuring the mark “WOULD YOU RATHER ... ?” on the game’s packaging. Zobmondo has produced several versions of its game under the ‘WOULD YOU RATHER ... ?” mark since then.

In January of 2002, the PTO issued a Notice of Allowance of Falls Media’s ITU application. Falls Media had not yet released a game using the “WOULD YOU RATHER ... ?” mark. Falls Media sought and received five extensions of time from the PTO, finally releasing its first “WOULD YOU RATHER ... ?” board game in December 2004. On July 19, 2005, having received Falls Media’s statement of use, the PTO issued Falls Media a registration for the “WOULD YOU RATHER ... ?” mark.

Thereafter, and perhaps in response, Zobmondo filed suit against Falls Media in the Central District of California in 2006 alleging, among other things, statutory and common-law trade-dress infringement, copyright infringement, and unfair competition arising from Falls Media’s “WOULD YOU RATHER ... ?” board game. Falls Media then filed suit against Zobmondo in the Southern District of New York alleging trademark infringement, unfair competition, and other claims under federal and state law. Zobmondo in the New York action counterclaimed against Falls Media, seeking, inter alia, cancellation of Falls Media’s ‘WOULD YOU RATHER ... ?” trademark registration. The claims in the New York action were transferred to the Central District of California and consolidated with Zobmondo’s 2006 action against Falls Media. Thus the claims and counterclaims in both actions came to a district court within the Ninth Circuit.

Falls Media and Zobmondo filed cross-motions for summary judgment. The district court granted summary judgment to Zobmondo on Falls Media’s trademark infringement and unfair competition claims, Falls Media’s claim alleging a violation of California Business and Professions Code §§ 17200 et seq., and Zobmondo’s counterclaim seeking cancellation of Falls Media’s mark from the federal registry. 6 The district court concluded that, despite the presumption of validity given to Falls Media’s federal trademark registration, there was no genuine issue of material fact whether “WOULD YOU RATHER ... ?” was a protectable mark: This mark, as perceived by the district court, was “merely descriptive” of an aspect of Falls Media’s games, and no rational juror could find that Falls Media had acquired secondary meaning in the mark by the time Zobmondo began using it in 2002.

Falls Media timely appealed. On appeal, Falls Media contends that it raised genuine issues of material fact regarding whether ‘WOULD YOU RATHER ... ?” is a valid, protectable mark.

*1113 II

We review summary judgments de novo. Jada Toys, Inc. v. Mattel, Inc., 518 F.3d 628, 632 (9th Cir.2008). Summary judgment for the moving party is appropriate only if “there is no genuine issue as to any material fact and ... the movant is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(c)(2). “[Summary judgment is generally disfavored in the trademark arena.” KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 408 F.3d 596, 602 (9th Cir.2005) (quoting Entrepreneur Media, Inc. v. Smith, 279 F.3d 1135, 1140 (9th Cir.2002)).

To claim trademark infringement, a plaintiff must have a “valid, protectable trademark.” Brookfield Commc’ns, Inc. v. W. Coast Entm’t Corp., 174 F.3d 1036, 1046 (9th Cir.1999). The issue of trademark validity is considered “an intensely factual issue.” KP Permanent Make-Up, 408 F.3d at 605. The plaintiff bears the ultimate burden of proof in a trademark-infringement action that the trademark is valid and protectable. Yellow Cab Co. of Sacramento v. Yellow Cab of Elk Grove, Inc., 419 F.3d 925, 927-28 (9th Cir.2005).

To be valid and protectable, a mark must be “distinctive.” Distinctiveness measures “the primary significance of the mark to the purchasing public.” Quiksilver, Inc. v. Kymsta Corp., 466 F.3d 749, 760 (9th Cir.2006) (quotation marks omitted). In determining distinctiveness, “we are required to consider standards of meaning not our own, but prevalent among prospective purchasers of the article.” Bada Co. v. Montgomery Ward & Co., 426 F.2d 8, 11 (9th Cir.1970); see also 2 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 11:70 (4th ed.2010)[hereinafter McCarthy]. “The fact-finder is not the designated representative of the purchasing public, and the fact-finder’s own perception of the mark is not the object of the inquiry. Rather, the fact-finder’s function is to determine, based on the evidence before it, what the perception of the purchasing public is.” Lane Capital Mgmt., Inc. v. Lane Capital Mgmt., Inc., 192 F.3d 337, 344 (2d Cir.1999).

Marks are generally classified in one of five categories of increasing distinctiveness: (1) generic, (2) descriptive, (3) suggestive, (4) arbitrary, or (5) fanciful. Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 768, 112 S.Ct. 2753, 120 L.Ed.2d 615 (1992). Which category a mark belongs in is a question of fact. See Lahoti v. VeriCheck, Inc., 586 F.3d 1190, 1195-96 (9th Cir.2009). Suggestive, arbitrary, and fanciful marks are considered “inherently distinctive” and are automatically entitled to federal trademark protection because “their intrinsic nature serves to identify a particular source of a product.” Two Pesos, 505 U.S. at 768, 112 S.Ct. 2753. Generic marks are not eligible for trademark protection. Entrepreneur Media, 279 F.3d at 1141. Merely descriptive marks are somewhere in-between; although they are not inherently distinctive and are therefore not entitled to automatic trademark protection, a merely descriptive mark can become protectable if it has acquired distinctiveness “as used on or in connection with the applicant’s goods in commerce.” 15 U.S.C. § 1052(f). This acquired distinctiveness is referred to as “secondary meaning.” Two Pesos, 505 U.S. at 769,112 S.Ct. 2753.

Although the plaintiff in a trademark action bears the ultimate burden of proof that his or her mark is valid, federal registration provides “prima facie evidence” of the mark’s validity and entitles the plaintiff to a “strong presumption” that the mark is a protectable mark. 15 U.S.C. §§ 1057(b), 1115(a); KP Permanent Make-Up, 408 F.3d at 604. Where *1114 the PTO issues a registration without requiring proof of secondary meaning, the presumption is that the mark is inherently distinctive. 7 See Quiksilver, 466 F.3d at 760. If the plaintiff establishes that a mark has been properly registered, the burden shifts to the defendant to show by a preponderance of the evidence that the mark is not protectable. Tie Tech, Inc. v. Kinedyne Corp., 296 F.3d 778, 783 (9th Cir.2002); Vuitton et Fils S.A. v. J. Young Enters., Inc., 644 F.2d 769, 775-76 (9th Cir.1981).

In this appeal, we review the district court’s conclusion that Falls Media raised no genuine issue of material fact on two issues: whether “WOULD YOU RATHER ... ?” is suggestive and not merely descriptive, and whether Falls Media had acquired secondary meaning in the mark. We address these issues in turn.

Ill

Falls Media first contends that the district court erred by concluding that there was no genuine issue of material fact whether “WOULD YOU RATHER ... ?” is suggestive, concluding instead that the mark is merely descriptive. 8 A suggestive mark is one for which “a consumer must use imagination or any type of multistage reasoning to understand the mark’s significance ... the mark does not describe the product’s features, but suggests them.” Kendall-Jackson Winery, Ltd. v. E. & J. Gallo Winery, 150 F.3d 1042, 1047 n. 8 (9th Cir.1998). By contrast, a merely descriptive mark “describes the qualities or characteristics of a good or service.” Park ‘N Fly, Inc. v. Dollar Park and Fly, Inc., 469 U.S. 189, 194, 105 S.Ct. 658, 83 L.Ed.2d 582 (1985). It “define[s] qualities or characteristics of a product in a straightforward way that requires no exercise of the imagination to be understood.” Entrepreneur Media, 279 F.3d at 1141-42 (quoting Kendall-Jackson, 150 F.3d at 1047 n. 8); see also Union Nat’l Bank of Tex., Laredo, Tex. v. Union Nat’l Bank of Tex., Austin, Tex., 909 F.2d 839, 845 (5th Cir.1990) (“[I]n many cases, a descriptive term will be an adjective such as ‘speedy,’ ‘friendly,’ ‘green,’ ‘menthol,’ or ‘reliable.’ ”). “Whether a mark suggests or describes the goods or services of the trademark holder depends [on] what those goods or services are[, and w]e therefore adjudge a mark’s strength by reference to the goods or services that it identifies.... ” Entrepreneur Media, 279 F.3d at 1142 (quotation marks omitted). In distinguishing between suggestive and descriptive marks, we are aware that “[d]eciding whether a mark is distinctive or merely descriptive is far from an exact science and is a tricky business at best.” Lahoti, 586 F.3d at 1197 (quotation marks omitted); 2 McCarthy § 11:66 (“The descriptive-suggestive borderline is hardly a clear one”), § 11:71 (observing that the descriptive-suggestive dichotomy is not “some kind of concrete and objective classification system”).

*1115 Because Falls Media showed that “WOULD YOU RATHER ... ?” was registered without proof of secondary meaning, Falls Media was entitled to a presumption that the mark is inherently distinctive — i.e., suggestive — and the burden shifted to Zobmondo to show that the mark is “merely descriptive” by a preponderance of the evidence. See Tie Tech, 296 F.3d at 783; Quiksilver, 466 F.3d at 760. Federal registration in itself does not mean that Falls Media can necessarily survive summary judgment solely on the basis of its registration. “[Assuming the defendant can demonstrate through law, undisputed facts, or a combination thereof that the mark is invalid, the evidentiary bubble [created by the federal registration] bursts and the plaintiff cannot survive summary judgment. In the face of sufficient and undisputed facts demonstrating [invalidity], the registration loses its evidentiary significance.” Tie Tech, 296 F.3d at 783; see also Talking Rain Beverage Co. v. S. Beach Beverage Co., 349 F.3d 601, 603 (9th Cir.2003). Nonetheless, the presumption of validity is a strong one, and the burden on the defendant necessary to overcome that presumption at summary judgment is heavy. See, e.g., Americana Trading, Inc. v. Russ Berrie & Co., 966 F.2d 1284, 1287 (9th Cir.1992) (reversing a district court’s grant of summary judgment because the district court “gave insufficient weight to the presumptive effect of[the plaintiffs] federal registration.”). “There can be no serious dispute with the principle that a federal trademark registration of a particular mark supports the distinctiveness of that mark, because the PTO should not otherwise give it protection. Registration alone may be sufficient in an appropriate case to satisfy a determination of distinctiveness.” Lahoti, 586 F.3d at 1199. Stated another way, the federal officials who register a mark are perceived to have some expertise in assessing if it is entitled to registration, and absent a presentation of secondary meaning, their registration is entitled to a presumptive validity, a presumptive conclusion that the mark was distinctive else they would not have registered it. Nonetheless, while the registration adds something on the scales, we must come to grips with an assessment of the mark itself.

We have generally applied one or two “tests” to differentiate between suggestive and merely descriptive marks. Because the district court found the application of these tests, the imagination test and the competitors’ needs test, important to its summary judgment inquiry, we examine each in turn. We stress that these tests are only “criteria offering] guidance,” Self-Realization Fellowship Church v. Ananda Church of Self-Realization, 59 F.3d 902, 911 (9th Cir.1995), and that the burden to show that there is no genuine issue of material fact is on the moving party, Zobmondo, while the non-moving party, Falls Media, gets the benefit of reasonable inferences.

The first, and clearly the most-used, test is known as the “imagination” test, and asks whether “imagination or a mental leap is required in order to reach a conclusion as to the nature of the product being referenced.” Rudolph Int’l, Inc. v. Realys, Inc., 482 F.3d 1195, 1198 (9th Cir.2007) (quotation marks omitted). For example, the mark “ENTREPRENEUR” as applied to a magazine was descriptive, not suggestive, because “an entirely unimaginative, literal-minded person would understand the significance of the reference.” Entrepreneur Media, 279 F.3d at 1142. On the other hand, “ROACH MOTEL” was held suggestive because “an ordinary consumer having read or heard on television the words ‘roach motel’ would remember the conception ... a fanciful abode for roaches in an establishment normally frequented by human [travelers].” Am. Home Prods. Corp. v. Johnson Chem. Co., 589 F.2d 103, *1116 107 (2d Cir.1978). We have said that the imagination test is our “primary criterion” for evaluating distinctiveness. Self-Realization, 59 F.3d at 911.

The district court determined that the imagination test indicated that “WOULD YOU RATHER ... ?” is merely descriptive as a matter of law because it requires “no imaginative or interpretive leap to understand that this phrase is the main aspect of the game to which it is affixed.” Falls Media argues that the district court erred because multistage reasoning is needed to link the mark to the “essential nature of these products ... that the choices are ridiculous, bizarre, or themed and that they are limited to a two-option format.”

We reject Falls Media’s argument in part. The imagination test does not ask what information about the product could be derived from a mark, but rather whether “a mental leap is required” to understand the mark’s relationship to the product. Rudolph Int’l, 482 F.3d at 1198 (quotation omitted and emphasis added). Our prior precedent makes it clear that merely descriptive marks need not describe the “essential nature” of a product; it is enough that the mark describe some aspect of the product. See, e.g., Bada, 426 F.2d at 11 (holding the mark “Micro-Precision” merely descriptive when applied to wheel balancers and weights).

But we conclude that the district court erred in concluding that the imagination test indicates that “WOULD YOU RATHER ... ?” is merely descriptive as a matter of law. We cannot look the entire mark up in a dictionary; there is no literal meaning of the “WOULD YOU RATHER ... ?” phrase, given that the words precede an ellipse; one may infer that there is a question, but only imagination can tell us that the question will serve up a bizarre or humorous choice. See Surgicenters of Am., Inc. v. Med. Dental Surgeries, Co., 601 F.2d 1011, 1015 n. 11 (9th Cir.1979) (“While not determinative, dictionary definitions are relevant and often persuasive in determining how a term is understood by the consuming public .... ”); cf. Entrepreneur Media, 279 F.3d at 1142 (holding that “Entrepreneur” is merely descriptive as the name of a magazine because an “entirely unimaginative, literal-minded person” would understand its meaning). On the one hand, consumers who already understand the phrase “WOULD YOU RATHER ... ?” to refer specifically to a game of questions involving bizarre or humorous choices might not consider the mark very suggestive as the name of a board game, but to consumers who do not share such an understanding, “WOULD YOU RATHER ... ?” is simply the first three words of an open-ended question. For those consumers, the mark “WOULD YOU RATHER ... ?” may not “describe” anything, except that a question is asked, and may indeed require imagination and multistage reasoning to understand the mark’s relationship to the game to which it is affixed. See Rudolph Int’l, 482 F.3d at 1198; Self-Realization, 59 F.3d at 911 (observing that the mark “recovery” in the self-help context is descriptive, but the same mark is suggestive when used as the name of a business teaching counselors to teach self-help because “[the] extra step constitutes the difference between descriptiveness and suggestiveness”). Given the record before us, which lacks comprehensive consumer surveys, we cannot say with confidence precisely what consumers will understand the phrase ‘WOULD YOU RATHER ... ?” to mean, nor are we confident that our own understanding of the phrase is an adequate substitute. 9 *1117 “With respect to a registered mark ... the putative infringer’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.” Borinquen Biscuit Corp. v. M.V. Trading Corp., 443 F.3d 112, 119 (1st Cir.2006). When we give all reasonable inferences to Falls Media, and credit its evidence as true, we conclude that the imagination test is inconclusive by itself to determine if the challenged mark is descriptive or suggestive of a board game.

The second test, known as the “competitors’ needs” test, “focuses on the extent to which a mark is actually needed by competitors to identify their goods or services.” Rodeo Collection, Ltd. v. W. Seventh, 812 F.2d 1215, 1218 (9th Cir.1987). If competitors have a great need to use a mark, the mark is probably descriptive; on the other hand, if “the suggestion made by the mark is so remote and subtle that it is really not likely to be needed by competitive sellers to describe their goods or services[,] this tends to indicate that the mark is merely suggestive.” Id. (internal punctuation omitted). The competitors’ needs test is related to the imagination test, “because the more imagination that is required to associate a mark with a product or service, the less likely the words used will be needed by competitors to describe their products or services.” Id. (internal punctuation omitted).

The district court concluded that the competitors’ needs test was “difficult to apply in this case” and declined to consider it because these tests “are merely factors to consider” and other tests favored Zobmondo. Falls Media argues that this was error, and in this case we agree. Drawing all inferences in favor of Falls Media, the competitors’ needs test strongly favored Falls Media’s argument that “WOULD YOU RATHER ... ?” is suggestive. Falls Media proffered significant evidence suggesting that its competitors do not need to use “WOULD YOU RATHER ... ?” to fairly describe their products. Perhaps most important is the experience of Zobmondo itself. Zobmondo identified 135 possible alternative names for its game during development. Also, Zobmondo marketed and sold its game and a related book for a period of time without using the phrase “WOULD YOU RATHER ... ?” (instead using the name “The Outrageous Game of Bizarre Choices”), and another board game company used the name “Would You Prefer?” during the same time period. These titles are not linguistically inferior to “WOULD YOU RATHER ... ?”. Cf. Entrepreneur Media, 279 F.3d at 1143 (observing that others need the term “entrepreneur” because “[w]e are not aware of, nor has EMI suggested, any synonym for the word”). In the face of this evidence, credited as true on summary judgment, it’s difficult to say that Zobmondo necessarily needs to use “WOULD YOU RATHER ... ?” for its version of the board game of bizarre or humorous choices.

Zobmondo argues that ‘WOULD YOU RATHER ... ?” is needed to fairly describe its products because the meaning of the phrase “WOULD YOU RATHER ... ?” is entrenched in the minds of consumers in a way that renders other possible marks inherently inferior. This argument, however, depends on a disputed issue of fact regarding the meaning of the phrase to consumers, an issue that we have already suggested cannot adequately be decided at summary judgment *1118 on the basis of abstract theorizing alone. Giving all reasonable inferences from the evidence in favor of Falls Media, competitors do not need to use “WOULD YOU RATHER ... ?” to describe their products. They can say “Would you prefer” or ‘Would you most like” or use some other verbal formula to convey a choice of alternatives.

The district court also employed a third test, known as the “extent-of-use” test, which evaluates “the extent to which other sellers have used the mark on similar merchandise.” 2 McCarthy § 11:69; see also 555-1212.com, Inc. v. Commc’n House Int'l Inc., 157 F.Supp.2d 1084, 1089 (N.D.Cal.2001) (“[I]n determining whether a word has a descriptive or suggestive significance as applied to a commercial service, it is proper to take notice of the extent to which others in a similar commercial context use the word.”). Zobmondo introduced evidence of six published books with the words “would you rather” in the title, as well as printouts from Google.com search results using the phrase. Falls Media argues that the district court improperly credited this disputed evidence over Falls Media’s objections without considering whether the words “would you rather” were used as a trademark and without making all inferences in Falls Media’s favor. 10

We are not aware of any prior case in our circuit employing the extent-of-use test as a controlling measure of trademark validity, and we need not do so here. If, as we see it, a summary judgment is disfavored by both the imagination test and the needs test, then extent of use at most could only be one factor to be considered, and could not command a summary judgment in the face of disputed facts about how a mark might be perceived by consumers. Extensive use of a mark by third parties might indicate that the mark is merely descriptive of a given class of products. See, e.g., Sec. Ctr., Ltd. v. First Nat’l Sec. Ctrs., 750 F.2d 1295, 1300 (5th Cir.1985) (concluding that the mark “security center” is descriptive in part because of “the number of enterprises throughout the nation that use ‘security center’ in some form”). But having determined for purposes of summary judgment that the imagination test is inconclusive and that the competitors’ needs test favors suggestiveness, we are not persuaded that the result of the extent-of-use test, based on Zobmondo’s proffered evidence, renders “WOULD YOU RATHER ... ?” merely descriptive as a matter of law. Zobmondo’s evidence of third party use is relevant, and may be considered by the trier of fact. But it must be measured, not merely against the results of abstract theoretical tests, but against the presumption of validity afforded to Falls Media’s registered mark and against any other evidence of distinctiveness that Falls Media has proffered.

The district court relied on the following evidence in concluding that “WOULD YOU RATHER ... ?” is merely descriptive:

• A statement by Falls Media’s literary agent that the mark was “an utterly obvious title” for books “[b]ecause it was a clear explanation of the content of the book.”

• A statement by Falls Media’s sales agent that “WOULD YOU RATHER ... ?” is a better name than “Zobmondo” because “somebody looking at Zob *1119 mondo would not know that it contains would you rather questions.”

• A statement by James Pressman, who optioned Falls Media’s game, that “WOULD YOU RATHER ... ?” was a good name for the product because “it gave you a good idea of what the game was all about” because “the questions always started with would you rather.”

• An e-mail from Gomberg stating that “most people that make the purchase [of a WOULD YOU RATHER ... ?’ game] do it because they’re familiar with the concept — as opposed to the actual game play or content.”

• Zobmondo’s evidence of books, websites, and copyright registrations suggesting that the ‘WOULD YOU RATHER ... ?” concept predates Falls Media’s trademark application.

We agree with the district court that this evidence collectively has some persuasive weight, but we do not believe it is “so one-sided that there can be no doubt about how the question [regarding the distinctiveness of the mark] should be answered.” Packman v. Chi. Tribune Co., 267 F.3d 628, 637 (7th Cir.2001) (quoting Door Sys., Inc. v. Pro-Line Door Sys., Inc., 83 F.3d 169, 171 (7th Cir.1996)). That “WOULD YOU RATHER ... ?” is an “utterly obvious” title for a book does not necessarily mean the mark is an equally obvious title in the context of a board game. See Lahoti, 586 F.3d at 1201; 2 McCarthy § 11:71 (“It must be kept in mind that a term that is in one category for a particular product may be in a quite different category for another product.”). The statement that “WOULD YOU RATHER ... ?” is more descriptive than “Zobmondo” does not exclude the possibility that it is still suggestive, nor does Pressman’s comment that “WOULD YOU RATHER ... ?” gives a person a “good idea” of what the game is about; a mark can be suggestive and still convey information about a product. See, e.g., 2 McCarthy § 11:72 (listing marks that have been found “suggestive,” including “CITIBANK” for an urban bank, “CLASSIC COLA” for a soft drink, “DIAL-A-MATTRESS” for mattress sales, and “FLORIDA TAN” for suntan lotion). Gomberg’s e-mail that purchasers are generally familiar with the “WOULD YOU RATHER ... ?” concept says nothing about how board-game consumers — a larger group than ultimate purchasers of the “WOULD YOU RATHER ... ?” game — understand the mark. Viewing this evidence in the light most favorable to Falls Media, all of this evidence suggests only Falls Media’s belief regarding how consumers perceive the mark, and is not necessarily conclusive of how consumers in fact perceive the mark. And Zobmondo’s evidence of third-party use, without contextual information such as sales figures and distribution locations, falls short of establishing a long-standing consumer understanding of the ‘WOULD YOU RATHER ... ?” concept for purposes of summary judgment. 11 See Scarves by Vera, Inc. v. Todo Imports Ltd., 544 F.2d 1167, 1173 (2d Cir.1976) (observing that “[t]he significance of third-party trademarks depends wholly upon their usage” including whether the marks “were actually used by third parties, that they were well promoted or that they were recognized by consumers”).

Falls Media has not rested solely on its federal registration to survive summary judgment. See Tie Tech, 296 F.3d at 784 *1120 (distinguishing between cases in which a trademark holder did or did not adduce additional evidence to supplement its federally registered trademark to survive summary judgment). Falls Media proffered the testimony of a game-industry expert, Philip E. Orbanes, who said that “WOULD YOU RATHER ... ?” had never been used as the title of a board game before Zobmondo’s entry into the market. The district court concluded that this testimony was “entirely irrelevant” because the issue was whether the societal concept of a “WOULD YOU RATHER ... ?” dilemma predated Falls Media’s use of the mark. We disagree. The ultimate issue at summary judgment was whether the mark ‘WOULD YOU RATHER ... ?” should receive trademark protection as applied to Falls Media’s board and card games. The fact that the mark had not been used previously to describe a board game is relevant to that inquiry; it suggests that competitors do not find the mark useful in describing their products. See, e.g., Sec. Ctr., 750 F.2d at 1300 (“We look into actual and likely use of a mark in order to determine whether its protection, i.e., its exclusion from the language freely available for commercial use, interferes with competition among providers of the same product or service.”). Zobmondo asks us to conclude that the mark is merely descriptive on a board game because of Zobmondo’s showing that consumers’ understanding of the ‘WOULD YOU RATHER ... ?” game concept predates the parties’ use of the mark. But that idea does not render Orbanes’s testimony irrelevant. To the contrary, Orbanes’s testimony about lack of third-party use on games tends to negate the inference that Zobmondo seeks to establish: that the mark is widely understood by consumers to refer to a type of question game.

Other evidence in the record also supports Falls Media’s claim that “WOULD YOU RATHER ... ?” could be seen by consumers as suggestive. For example, that Zobmondo included descriptive information on its game boxes after it began using the “WOULD YOU RATHER ... ?” mark may suggest that Zobmondo itself thought that the mark did not fully describe its products. 12 And Zobmondo’s attempt to acquire trademark rights in the mark supports an inference that Zobmondo believed, at one time, that the mark was inherently distinctive. 13 This evidence is no less relevant to the distinctiveness inquiry than the evidence that persons associated with Falls Media thought the mark provided descriptive information. Both the evidence of Zobmondo’s trademark applications and the evidence that some persons in business relations with Falls Media thought the mark descriptive demonstrate that there is a factual contest at bottom in assessing the descriptiveness or suggestiveness of the “WOULD YOU RATHER ... ?” mark, rendering that issue unsuitable for a summary judgment determination.

*1121 Looking at the totality of evidence proffered by Zobmondo and Falls Media and taking into account the “strong presumption” accorded to Falls Media’s federal registration, we conclude that there is a genuine issue of material fact whether “WOULD YOU RATHER ... ?” is inherently distinctive to consumers. “Deference to the PTO’s classification decision is sensible because the PTO has special expertise that we lack on this fact-intensive issue.” Lahoti, 586 F.3d at 1199 (citing Lone Star Steakhouse & Saloon, Inc. v. Alpha of Va., Inc., 43 F.3d 922, 934 (4th Cir.1995)). This principle applies meaningfully at the summary judgment stage, where we are not permitted to weigh evidence or draw inferences against Falls Media. “[T]he fact that a mark is registered is strong evidence that the mark satisfies the statutory requirements for distinctiveness necessary for trademark protection.” Retail Servs., Inc. v. Freebies Publ’g, 364 F.3d 535, 542 (4th Cir.2004). Giving Falls Media the benefit of the presumptive effect of its federal registration, crediting the testimony of Falls Media, placing the burden of challenge to the mark’s distinctiveness on Zobmondo, and giving all reasonable inferences to Falls Media, we conclude that there is a genuine issue of material fact whether “WOULD YOU RATHER ... ?” is suggestive or merely descriptive as a mark for a board game. The issue of descriptiveness or suggestiveness in this case cannot correctly be resolved by summary judgment, and we remand for trial.

Because we reverse summary judgment in favor of Zobmondo on the ground that there is a genuine issue of material fact whether or not ‘WOULD YOU RATHER ... ?” is inherently distinctive, we need not reach Falls Media’s additional claim that there is a genuine issue of material fact regarding secondary meaning. The issue of secondary meaning must be reached only if the fact-finder determines that “WOULD YOU RATHER ... ?” is merely descriptive; in that scenario, the fact-finder will have to make the ultimate determination of secondary meaning in the context of the total record of evidence offered at trial.

This is a difficult case, and we express no opinion on how it may be resolved after findings of fact at trial. But this appeal is from a summary judgment granted in the face of complex issues of fact, in a trademark context where summary judgment is ordinarily disfavored, and despite the fact that the non-moving party has a federally registered trademark constituting “prima facie evidence” of the mark’s validity. 15 U.S.C. § 1057(b). Considering Falls Media’s federal registration, crediting all of Falls Media’s evidence and drawing all reasonable inferences in Falls Media’s favor, and placing the burden of proving mere descriptiveness on Zobmondo, we conclude that application of the imagination and the competitors’ needs tests are inconclusive on the disputed facts. We reverse and remand for trial.

REVERSED AND REMANDED.

. Appellants Justin Heimberg and David Gomberg formed Falls Media, LLC in 2004. For ease of reference, we use the term “Falls Media” to refer collectively to Appellants both before and after the formation of Falls Media, LLC.

. For example, a sample question posed by Falls Media asks, "Would you rather be able to expedite the arrival of an elevator by pressing the button multiple times or have the ability to sound incredibly natural and sincere on answering machines?”

. Ordinarily, trademark registrants cannot register a trademark until they have used the mark in commerce. WarnerVision Entm’t Inc. v. Empire of Carolina, Inc., 101 F.3d 259, 260 (2d Cir. 1996); see also 15 U.S.C. § 1051(a)(1). An ITU application, however, allows a person not yet using a mark to file an anticipatory application for registration on the basis of an intent to use the mark in the future. Once the PTO issues a "notice of allowance” of the ITU application pursuant to 15 U.S.C. § 1063(b), the applicant has six months, absent extension, to begin using the mark in commerce. WarnerVision, 101 F.3d at 260; see also 15 U.S.C. § 1051(d)(1)-(2). If the applicant files a statement of use within the proscribed timeframe, the mark is registered and the date the ITU application was filed becomes the applicant’s constructive-use date. See 15 U.S.C. § 1057(c). This gives the applicant priority-of-use over anyone who adopts the mark after the constructive-use date. See id.

. The district court concluded that the 2002 self-release was Zobmondo's first use of "WOULD YOU RATHER ... ?” as a trademark on its games. Construing the facts in the light most favorable to Falls Media, we agree that 2002 is the relevant use date for purposes of this appeal.

. Horn later opposed Falls Media’s trademark registration before the PTO, but his efforts were unsuccessful. He also approached *1112 Heimberg and Gomberg in 2002 about acquiring the “WOULD YOU RATHER ... ?” mark, but Heimberg and Gomberg declined to sell.

. The district court granted summary judgment to Falls Media on Zobmondo's trade-dress infringement, copyright infringement, and unfair competition claims, but Zobmondo does not appeal that portion of the decision. The district court also granted a separate summary judgment to Falls Media on Zobmondo’s fraudulent-procurement claim, which Zobmondo has also not appealed.

. A corollary of this principle is that the registrant is not entitled to a presumption of secondary meaning unless the PTO required proof of secondary meaning as part of the application for registration of the mark. See PaperCutter, Inc. v. Fay’s Drug Co., 900 F.2d 558, 564 (2d Cir.1990) ("[The] burden [to prove secondary meaning] does not shift upon a decision of the [PTO] to register the mark, absent evidence that the [PTO] registered the mark upon finding that it had acquired secondary meaning.”).

. The district court concluded that "WOULD YOU RATHER ... ?” is not generic, and Zobmondo does not press that argument on appeal. Falls Media conceded before the district court that "WOULD YOU RATHER ... ?” is not an arbitrary or fanciful mark. We assume without deciding that "WOULD YOU RATHER ... ?” is either a suggestive or a merely descriptive mark, which is how the parties have framed the issue.

. The underlying issue is the standard of meaning "prevalent among prospective purchasers of the article." Bada, 426 F.2d at 11. On that basis, some terms may not be susceptible to abstract "imagination test” analysis at *1117 summary judgment, and instead the application of the imagination test will be informed by expert testimony offered at trial suggesting how consumers will view this phrase on a board game.

. Falls Media also argues that the district court ignored its objections to the admissibility of Zobmondo’s extent-of-use evidence. After reviewing the record, we conclude that the district court considered and correctly overruled Falls Media's objections to this evidence.

. Zobmondo has established that collections of bizarre questions beginning with the words “would you rather” were posed in printed media before Falls Media began using the phrase as a mark. But this does not establish that there is no genuine issue of material fact that these types of questions were understood by consumers to be part of a "WOULD YOU RATHER ... ?” game of questions.

. Several Zobmondo game boxes follow up the mark “WOULD YOU RATHER ... ?” with the tag-line "the game of mind-boggling questions.” Another Zobmondo game box includes sample questions. The need to explain what a “WOULD YOU RATHER ... ?” question entails cuts against Zobmondo's argument that "WOULD YOU RATHER ... ?” is a well-established, culturally-pervasive concept.

. Zobmondo argues that its trademark applications were filed by Horn before he had obtained legal advice about the distinctiveness of marks. Maybe so, but this is a contested fact, and, whatever might be the conclusion on that after trial and findings of fact, for purposes of summary judgment we must give all reasonable inferences in favor of Falls Media. For these summary judgment purposes, Falls Media is entitled to the inference that Zobmondo’s trademark applications are to a degree an admission that at the time Zobmondo viewed the mark as suggestive.