The Opinion filed September 6, 2002, slip op. 13377, and appearing at 2002 WL 31001183 (Sep. 6, 2002), is amended as follows:

The caption is changed to reflect the addition of Case No. 00-55599, the companion case consolidated with No. 00-55293. This case number was inadvertently omitted from the opinion as filed.

BERZON, Circuit Judge.

Trek Bicycle Corporation ("Trek") has long used the trademark "TREK" on its bicycles and related products. In recent years, Thane International, Inc. ("Thane") began producing and selling a stationary exercise machine under the "OrbiTrek" mark. The question in this case is whether Thane has violated the federal trademark laws by its choice of name for its product.

Trek, a Wisconsin corporation, has manufactured bicycles under the TREK mark since 1977. Trek also uses the TREK label on a variety of bicycle accessories including clothing, hats, packs, wrist watches, bags, video games, toys, computer games, helmets, and gloves. In all, Trek identifies over 1,000 products with the TREK mark. Trek products are sold through more than 1,600 independent dealers in 2,000 locations across the country. Surveys published in Bicycling Magazine in 1995, 1997, and 1999, each indicate that, according to the magazine's subscribers, Trek is the country's most popular and most respected bicycle brand.

In 1981, Trek was granted a United States trademark for the use of TREK on bicycles and bicycle frames. Since then, Trek has been awarded several other trademarks based on variations of the word "trek" including: "TREKKING" in 1996 for bicycles; "TREK 100" in 1996 for providing support services during charitable bicycle rides; "TREK BMX" in 1998 for bicycles and bicycle frames; and "ELEC TREK" in 1999 for bicycles, bicycle frames and parts. Additionally, Trek registered TREK on May 12, 1997, for "exercise equipment, namely stationary exercise cycles."

Trek spends between $3 million and $5 million per year on advertising. Publications including Rolling Stone Magazine, Men's Journal, Outside Magazine, Spin, Playboy, Women's Sport & Fitness, Bicycling Magazine, Mountain Bike, Backpacker, and Velonews run Trek's ads. Trek estimates that its website attracts 4.5 million visitors annually. Trek's other advertising comes through a variety of promotional efforts, including sponsoring athletes to use its equipment. The most prominent athlete Trek sponsors is cancer-survivor Lance Armstrong, who has used TREK bicycles to capture several Tour de France titles. The world press has chronicled Armstrong's victories, generating considerable publicity for Trek. Pictures of Armstrong with a TREK bicycle have appeared in The New York Times and on Wheaties boxes. When Armstrong presented a TREK bicycle to President Clinton, coverage of that event also made its way into papers across the country.

In 1993, Trek entered the stationary exercise machine market by introducing six models of exercise bikes bearing the TREK mark. Thane maintains that this venture was a failure. Three years later, Trek sold its TREK Fitness line to Vision Fitness, a small company founded by two former Trek employees. The transaction included a license allowing Vision Fitness to use the TREK mark for exercise equipment until September 1997. Trek claims that some dealers continued to advertise and sell stationary exercise cycles bearing the TREK logo through March 31, 1999.

In January 1998, Trek began considering another foray into the stationary exercise machine market by preparing a "new product evaluation" for a new stationary trainer. This stationary trainer would not have pedals, a seat, or handlebars. Instead, the device would allow the user to convert her mobile bicycle into a stationary one for indoor use. At the time of the summary judgment motions, Trek planned to begin selling these trainers in the summer of 2000.

Based in La Quinta, California, Thane operates in the "direct response marketing" field. It airs lengthy "infomercials" on television broadcast and cable stations, encouraging viewers to use the telephone, direct mail, or the Internet to purchase products directly from Thane.

In 1997, Thane developed the OrbiTrek, characterized as a "dual directional elliptical glider stationary exercise machine for indoor use." The OrbiTrek has rectangular platform pedals large enough to support a person's entire foot. The pedals move in an elliptical motion "designed to simulate the body's natural stride." Movable handlebars extend straight up from the OrbiTrek's base and provide upper body exercise. The OrbiTrek has no seat, because it is meant to be used while standing.

Thane began airing its 28-minute infomercial for the OrbiTrek in December 1997. The infomercial stated that the OrbiTrek sells for the "unbelievable price of only $299.95," but if you "[c]all right now [] the exciting new OrbiTrek will be shipped to you immediately for the unheard of low price of only $199.95."

Thane explains that the Trek part of the OrbiTrek mark had an inspiration entirely independent of Trek bicycles and other equipment. Thane's executive vice-president, Denise DuBarry, was married to an actor who appeared in the original pilot for the television Star Trek show. DuBarry has long watched the Star Trek television series and attended Star Trek conventions. By using a word associated with Star Trek, Thane believes it depicts the OrbiTrek as a "space-age, high-tech, and futuristic product." Indeed, Thane claims none of its employees had heard of TREK bicycles prior to the current dispute. Thane did not perform a trademark search for TREK before adopting OrbiTrek as the name for its exercise machine.

Thane filed an application with the United States Patent and Trademark Office to register "ORBITREK" for goods described as "stationary exercise machines." Trek thereupon filed a Notice of Opposition with the Trademark Trial and Appeal Board. On February 11, 1999, Thane filed a complaint and demand for jury trial in federal district court, seeking a declaration that it had not violated trademark laws under the Lanham Act, state common law, or state statutory law. Trek responded with a counter-claim and demand for jury trial.1 Its opposition to Thane's trademark application was suspended pending the outcome of this case.

The parties then filed cross-motions for summary judgment on all claims. The district court granted Thane's motion for summary judgment and denied Trek's, holding that "any reasonable juror would conclude that there is no likelihood of confusion between Trek Bicycle Corporation's `Trek' mark and Thane's `OrbiTrek' mark." Thane next moved for attorney's fees, but the district court denied this motion. Trek now appeals the district court's denial of its summary judgment motion and the district court's grant of Thane's summary judgment motion. Thane appeals the district court's denial of its motion for attorney's fees.

The federal statute prohibiting trademark infringement requires a trademark holder to prove that the alleged infringer's use of a mark "is likely to cause confusion, or to cause mistake, or to deceive." 15 U.S.C. § 1114(1)(a) & (b).2 This requirement directly advances the dual purposes of infringement law: ensuring that owners of trademarks can benefit from the goodwill associated with their marks and that consumers can distinguish among competing producers. Entrepreneur Media, Inc. v. Smith, 279 F.3d 1135, 1140 (9th Cir.2002) (quoting Park `N Fly, Inc. v. Dollar Park and Fly, Inc., 469 U.S. 189, 198, 105 S. Ct. 658, 83 L. Ed. 2d 582 (1985)); Brookfield Communications, Inc. v. West Coast Entm't Corp., 174 F.3d 1036, 1053 (9th Cir.1999) (quoting Qualitex Co. v. Jacobson Products Co., Inc., 514 U.S. 159, 163, 115 S. Ct. 1300, 131 L. Ed. 2d 248 (1995)).3 The question whether an alleged trademark infringer's use of a mark creates a likelihood that the consuming public will "be confused as to who makes what" product is therefore the "core element" of trademark infringement law. Brookfield, 174 F.3d at 1053.

We ordinarily rely on an eight factor test, first articulated in AMF Inc. v. Sleekcraft Boats, 599 F.2d 341 (9th Cir. 1979), to help determine if a likelihood of confusion exists.4 The eight factors are: (1) the strength of Trek's mark; (2) the similarity of the Trek and Thane marks; (3) the proximity or relatedness of the goods or services; (4) Thane's intent in selecting its mark; (5) evidence of actual confusion; (6) the marketing channels used; (7) the likelihood of expansion into other markets; and (8) the degree of care purchasers are likely to exercise.

Unless properly used, this long list of factors has the potential to befuddle the inquiry. The list of factors is not a scorecard — whether a party "wins" a majority of the factors is not the point. Nor should "[t]he factors ... be rigidly weighed; we do not count beans." Dreamwerks Prod. Group v. SKG Studio, 142 F.3d 1127, 1129 (9th Cir.1998). "Some factors are much more helpful than others, and the relative importance of each individual factor will be case specific.... [I]t is often possible to reach a conclusion with respect to likelihood of confusion after considering only a subset of the factors." Brookfield Communications, 174 F.3d at 1054.

The nature of the likelihood of confusion inquiry largely shapes the role of district courts addressing motions for summary judgment and our role in reviewing grants and denials of summary judgments. Likelihood of confusion is a factual determination. Downing v. Abercrombie & Fitch, 265 F.3d 994, 1008 (9th Cir.2001). Therefore, a district court may grant summary judgment and we may uphold a grant of summary judgment only if "no genuine issue" exists regarding likelihood of confusion. Fed.R.Civ.P. 56(c). We have cautioned that district courts should grant summary judgment motions regarding the likelihood of confusion sparingly, as careful assessment of the pertinent factors that go into determining likelihood of confusion usually requires a full record. Clicks Billiards Inc. v. Sixshooters Inc., 251 F.3d 1252, 1265 (9th Cir.2001); Interstellar Starship Services, Ltd. v. Epix, Inc., 184 F.3d 1107, 1109 (9th Cir.1999).5

Here, both parties have moved for summary judgment. Although the analysis of each motion informs the other, we address each separately.

The grant of summary judgment for Thane falters upon consideration of a single Sleekcraft factor, evidence of actual confusion.

Trek provided evidence that consumers were actually confused by Thane's use of OrbiTrek. Based on that evidence, a reasonable jury could find a likelihood of confusion.

Evidence of actual confusion "constitutes `persuasive proof that future confusion is likely.'" Clicks Billiards, 251 F.3d at 1265 (quoting Fuddruckers, Inc. v. Doc's B.R. Others, Inc., 826 F.2d 837, 845 (9th Cir.1987)). This rule makes good sense. If enough people have been actually confused, then a likelihood that people are confused is established.

This is not to say that evidence of actual confusion will always compel a jury to find likelihood of confusion. In some cases, a jury may properly find actual confusion evidence de minimis and thus "unpersuasive as to the ultimate issue...." Entrepreneur Media, 279 F.3d at 1150. But if a party produces evidence from which a reasonable jury could surmise that an "appreciable number" of people are confused about the source of the product, then it is entitled to a trial on the likelihood of confusion — although it will not necessarily prevail at that trial. Id. at 1151 (emphasis in original).

Trek did not present direct evidence of actual confusion, such as the testimony of actual customers. It did, however, present extensive survey evidence of actual confusion. Survey evidence may establish actual confusion. Clicks Billiards, 251 F.3d at 1265; Wendt v. Host Int'l, Inc., 125 F.3d 806, 813 (9th Cir.1997); Levi Strauss & Co. v. Blue Bell, Inc., 778 F.2d 1352, 1360 (9th Cir.1985) (en banc).

After Trek's dispute with Thane began, Trek hired David A. Stewart, a professor at the Gordon Marshall School of Business at the University of Southern California, to oversee a survey into the likelihood of confusion between the two companies' products. The survey consisted of interviews with 400 people over the age of 18 who had purchased either a bicycle or exercise equipment in the last three years or planned to do so in the next year. The respondents were interviewed in shopping malls in eight metropolitan areas across the United States.

Three hundred of the respondents were shown pictures of advertisements for Trek products and OrbiTrek products and asked questions about the pictures.6 After analyzing the relevant data, Stewart concluded that "27.7 % of the respondents in the survey were confused with respect to the source or association of the OrbiTrek products based on the similarity of its name and/or logo to the Trek name or logo."

Thane has criticized the survey's methodology. But drawing all justifiable inferences from the survey in Trek's favor as we must on summary judgment, we must conclude that the survey provides evidence from which a reasonable jury could conclude that more than one quarter of those who encounter both Trek and OrbiTrek ads will be confused about the origin of the OrbiTrek exercise machine. Because actual confusion is persuasive proof of the likelihood of confusion, a reasonable jury could conclude that a likelihood of confusion exists. We therefore reverse the district court's grant of summary judgment to Thane.

With respect to Trek's motion, the district court determined that "no reasonable juror could find that likelihood of confusion exists." Although this conclusion exaggerates the weakness of the evidence in Trek's favor, it is not so far off the mark as to require summary judgment for Trek. We turn to two other relevant Sleekcraft factors — the similarity of the marks and the relatedness of the goods — to determine whether the district court properly denied Trek's motion. Cf. Brookfield Communications, 174 F.3d at 1054 ("factors ... such as similarity of the marks and whether the two companies are direct competitors... will always be important").

Both TREK and OrbiTrek include the syllable "Trek," and the syllable is separately capitalized in "OrbiTrek." A reasonable jury might nonetheless conclude that the marks are not similar. This is so even if the jury takes seriously the admonition that "[s]imilarities weigh more heavily than differences." Sleekcraft, 599 F.2d at 352.

OrbiTrek contains the two syllable prefix "Orbi," while TREK does not. So OrbiTrek has three times as many syllables as TREK and twice as many letters. In Entrepreneur Media we held that a reasonable fact finder could find "Entrepreneur" dissimilar from both "Entrepreneur Illustrated" and "EntrepreneurPR." In particular, we reasoned that "Entrepreneur Illustrated" is almost twice as long — to both the eye and ear — as "Entrepreneur," and "EntrepreneurPR" contains two more syllables than "Entrepreneur." 279 F.3d at 1145-46. Moreover, the TREK trademark appears with all four letters capitalized, distinguishing it visibly from OrbiTrek.

In addition, "the more closely related the goods are, the more likely consumers will be confused by similar marks." Entrepreneur Media at 1147. To determine whether the goods are related, we ask whether "the consuming public is likely somehow to associate" the OrbiTrek with Trek. Brookfield Communications, 174 F.3d at 1056.

Trek primarily sells bicycles and bicycle accessories. Although a mobile bicycle and the OrbiTrek machine both provide exercise, a jury could find this relationship insufficient to support a likelihood of confusion between these products.

Trek did sell six models of stationary exercise bikes from 1993 to 1996. Trek sold its fitness equipment line to Vision Fitness in 1996; that company continued to market stationary bikes bearing the TREK mark through at least 1997. The parties dispute whether any stationary bikes bearing the Trek logo were sold after Thane introduced the OrbiTrek in 1998. Trek did, however, submit evidence showing that it planned to sell a device beginning in 2000 that would allow a user to convert her bicycle into a stationary bicycle.

Taken all together, this evidence concerning the relatedness of Trek and Thane's products is sufficiently balanced that a reasonable juror could conclude either that Trek and Thane's products are related or that they are not. A jury, making all reasonable inferences in Thane's favor, could conclude that Trek's foray into the stationary bike market was a short-lived failure and thus unimportant to a likelihood of confusion determination, particularly if the jury concluded that Trek's venture ended entirely before the OrbiTrek was introduced. Also, even if a jury considered Trek an active or future producer of stationary bikes, it could reasonably conclude that stationary bikes and elliptical gliders are different enough from each other that consumers would not confuse the OrbiTrek with Trek's stationary exercise bikes. As the district court observed, "[t]he OrbiTrek does not have bicycle style pedals or a seat for someone to sit on as they would if they were riding a bicycle, stationary or otherwise." Based on this observation, a reasonable jury could conclude, as the district court did, that the products are not related.

Finally, although we conclude above that a reasonable juror could find actual confusion based on Trek's survey data, that data is not so compelling that a reasonable jury could not credit Thane's criticisms of the survey evidence and give the evidence of actual confusion little or no credence.

In sum, a reasonable jury could conclude that there is no credible evidence of direct confusion, that TREK and OrbiTrek are not similar, and that the products identified by the marks are not related. If it reached these conclusions, a jury could reasonably determine that there was no likelihood of confusion between the TREK and OrbiTrek marks. Therefore, Trek's motion for summary judgment must fail as did Thane's. We remand to the district court for trial on Trek's trademark infringement claim.

The owner of a famous mark shall be entitled, subject to the principles of equity and upon such terms as the court deems reasonable, to an injunction against another person's commercial use in commerce of a mark or trade name, if such use begins after the mark has become famous and causes dilution of the distinctive quality of the mark ...

§ 1125(c)(1). The statute goes on to set out factors pertinent in determining "whether a mark is distinctive and famous," § 1125(c)(1)(A) to (H), and to define "dilution" as "the lessening of the capacity of a famous mark to identify and distinguish goods or services, regardless of the presence or absence of — (1) competition between the owner of the famous mark and other parties; or (2) likelihood of confusion, mistake or deception." § 1127.

Trademark infringement principles protect both trademark holders and consumers from the consequences of confusion about the source of a product. In contrast, the animating concern of the dilution protection is that the user of the diluting mark appropriates or free rides on the investment made by the trademark holder. Playboy Enterprises, Inc. v. Welles, 279 F.3d 796, 805 (9th Cir.2002). Consistent with this distinction, likelihood of confusion is not an element of a trademark dilution cause of action under either federal law, as the above-quoted language makes explicit, or California law. See Cal. Bus. & Prof.Code § 14330(a).

The district court's grant of summary judgment to Thane on the dilution cause of action was based solely on its premise that there was no likelihood of confusion. That premise could not be determinative of the dilution claim even if true (and we conclude above that it is not true in the sense that summary judgment was appropriate on likelihood of confusion). So we must analyze the dilution claim afresh.

The Federal Trademark Dilution Act of 1995, Pub.L. No. 104-98, 109 Stat. 985 (1996), is relatively new, and presents formidable problems of interpretation. See generally Nabisco, Inc. v. PF Brands, 191 F.3d 208 (2nd Cir.1999). Further, the implications of a broad application of the federal antidilution statute are troubling, as "[d]ilution causes of action, much more so than infringement and unfair competition laws, tread very close to granting `rights in gross' in a trademark," Avery Dennison Corp. v. Sumpton, 189 F.3d 868, 875 (9th Cir.1999), thereby hampering competition and the marketing of new products.

Yet, Congress obviously perceived an inadequacy in the existing federal trademark laws when it passed the antidilution statute and intended to grant comparatively expansive rights to prevent the use of established marks in some circumscribed circumstances. We may not interpret the statute so narrowly as to compromise the evident intent. We therefore agree with the Second Circuit that, in considering this new statutory right, "courts ... do better to feel their way from case to case," Nabisco, 191 F.3d at 227, rather than ruling in sweeping brush strokes. It is with these cautions in mind that we proceed to consider the dilution issue in this case.

We begin from the recently-established requirement that for a dilution claim to succeed, the mark used by the alleged diluter must be identical, or nearly identical, to the protected mark. Playboy Enterprises, 279 F.3d at 805. Such a requirement comports with the statutory language, the four-part dilution test derived from that language outlined in Panavision Int'l L.P. v. Toeppen, 141 F.3d 1316, 1324 (9th Cir.1998) and Avery Dennison, 189 F.3d at 874, and with the statute's legislative history and purposes.

The statute establishes that the junior user, to be liable for dilution, must use "a mark or trade name ... after the mark has become famous." § 1125(c)(1) (emphasis added). As articulated in Panavision and Avery Dennison, the test for dilution similarly provides that to make out an antidilution cause of action, a plaintiff must show that "its mark is famous" and "the defendant is making commercial use of the mark in commerce." Avery Dennison, 189 F.3d at 874 (emphasis added); see Panavision, 141 F.3d at 1324; see also Mattel, Inc., v. MCA Records Inc., 296 F.3d 894, 903 (9th Cir.2002) (" `Dilution' refers to the `whittling away of the value of a trademark' when it's used to identify different products.") (emphasis added) (citation omitted). These locutions indicate that the defendant must use essentially the same mark, not just a similar one. Concomitantly, this circuit's description of the two most common forms of dilution — blurring and tarnishment — requires a defendant to use the plaintiff's actual mark, rather than a mark that is merely similar. See Panavision, 141 F.3d at 1326 n. 7 ("Blurring occurs when a defendant uses a plaintiff's trademark to identify the defendant's goods and services," and "[t]arnishment occurs when a famous mark is improperly associated with an inferior mark or offensive product or service.").

The legislative history, while not definitive on the issue, also suggests that the marks must be identical or close thereto. In explaining the difference between dilution and infringement, the Senate Report states that: "The concept of dilution focuses on the investment the owner has made in the mark and on the commercial values and aura of the mark itself, protecting both from those who would appropriate the mark for their own benefit." S.Rep. No. 100-515, at 7 (1988), reprinted in 1988 U.S.C.C.A.N. 5577, 5583 (emphases added). The Report then lists two hypothetical examples of dilution: "where a mark such as Kodak is used for pianos, or Buick is used for aspirin." Id. No example is given of the use of marks that are merely similar to the famous mark.

Further, a dilution claim alleges a form of appropriation. Playboy Enterprises, 279 F.3d at 805. Appropriation implies the adoption of the mark itself, not the use of a similar mark. As discussed previously, infringement is designed to protect against consumer confusion about the source of a product that may arise, inter alia, because a company uses a similar mark. Dilution, on the other hand, protects the distinctiveness of a particular mark whether or not the products compete or consumer confusion exists. § 1127. Because dilution and likelihood of confusion tests are directed at different actions, it does not make sense to import the relatively subjective similarity of the marks test from the likelihood of confusion context into the dilution context. See 4 J.McCarthy, Trademarks and Unfair Competition, § 24:90.2 (4th ed.2001).

In Playboy Enterprises, we elucidated the "identical or nearly identical" standard by adopting the Eighth Circuit's expression of the requirement: For marks to be nearly identical to one another, they "must be `similar enough that a significant segment of the target group of customers sees the two marks as essentially the same.'" 279 F.3d at 806 n. 41 (quoting Luigino's Inc. v. Stouffer Corp., 170 F.3d 827, 832 (8th Cir.1999)). Other circuits also have adopted more stringent similarity of marks tests in the dilution context than in the infringement context. For example, the Second Circuit recently stated that the anti-dilution act's purpose is to protect famous marks against dilution "when a junior user uses the same mark in a non-confusing way." TCPIP Holding Co., Inc. v. Haar Communications, Inc., 244 F.3d 88, 95 (2d Cir.2001). It has thus adopted the following test:

The marks must be of sufficient similarity so that, in the mind of the consumer, the junior mark will conjure an association with the senior.... ("We hold ... that the marks must be `very' or `substantially' similar and that, absent such similarity, there can be no viable claim of dilution.").

Nabisco, 191 F.3d at 218 (quoting Mead Data Central, Inc. v. Toyota Motor Sales, U.S.A., Inc., 875 F.2d 1026, 1029 (2d. Cir. 1989)); see also Federal Express Corp. v. Federal Espresso, Inc., 201 F.3d 168, 176 (2d Cir.2000) (quoting Nabisco, 191 F.3d at 218).7 The Fourth Circuit has held that there must be "a sufficient similarity between the junior and senior marks to evoke an instinctive mental association of the two by a relevant universe of consumers." Ringling Bros. — Barnum & Bailey Combined Shows, Inc. v. Utah Div. of Travel Development, 170 F.3d 449, 458 (4th Cir.1999) (internal quotations omitted). But see Eli Lilly & Co. v. Natural Answers, Inc., 233 F.3d 456, 469 (7th Cir. 2000) (using the same similarity test for a dilution claim used for a related infringement claim).

Applying the Playboy "identical or nearly identical" standard, we conclude that although TREK and OrbiTrek are not identical to one another, the possibility exists that a reasonable factfinder could find that OrbiTrek is using a mark nearly identical to Trek's mark. The OrbiTrek mark contains the entire TREK mark and highlights the TREK mark by capitalizing it in the middle of a single word. A reasonable trier of fact could determine that Thane "use[d]" the TREK mark by incorporating the same word into its own mark as a separate, visually identifiable element, and that a significant segment of the consuming public would likely focus on that element as an identifier essentially the same as the TREK mark. On the other hand, a reasonable factfinder could decide that examined as a whole, the OrbiTrek mark is sufficiently dissimilar from the TREK mark that the two could be viewed as not essentially the same. After all, "Trek" in "OrbiTrek" is conjoined with "Orbi" rather than appearing as a separate word and "Trek" is not all in capital letters. We conclude that the issue of identity cannot be decided on a motion for summary judgment.

Aside from establishing the identity or near identity of the marks, a party alleging dilution must satisfactorily prove that "(1) its mark is famous; (2) the defendant is making commercial use of the mark in commerce; (3) the defendant's use began after the plaintiff's mark became famous; and (4) the defendant's use presents a likelihood of dilution of the distinctive value of the mark." Avery Dennison, 189 F.3d at 874.8 As we conclude that a reasonable factfinder could not find TREK famous in any relevant market segment, we need not proceed beyond the first prong of this inquiry.

(1) General Principles: The federal anti-dilution statute limits protection to the owners "of a famous mark." § 1125(c)(1). To help "determin[e] whether a mark is distinctive and famous," the statute, in keeping with trademark law's apparent penchant for flexible eight-factor tests, lists eight non-exclusive factors "a court may consider."9 § 1125(c)(1). The only case in this circuit to explicate the "famousness" requirement, Avery Dennison, stressed that fame in the dilution context must be very narrow: "Dilution is a cause of action invented and reserved for a select class of marks — those marks with such powerful consumer associations that even non-competing uses can impinge their value." 189 F.3d at 875. As a result, "to meet the famousness element of protection under the dilution statutes a mark must be truly prominent and renowned." Id. (quoting I.P. Lund Trading ApS v. Kohler Co., 163 F.3d 27, 46 (1st Cir.1998) (internal quotations omitted)).

This limitation on dilution protection created by the narrow definition of famousness, like the identical or nearly identical requirement already discussed, is critical. Antidilution is "the most potent form of trademark protection" and has the potential of "over-protecting trademarks." Id. Further, the concept of dilution is more abstract than the concept of trademark infringement, which is anchored by the likelihood of confusion standard.10 Absent strict policing of the famousness requirement, neither participants in the commercial market-place nor courts are likely to apply dilution statutes in a predictable fashion. It is one thing to determine whether consumers are likely to transfer associations evoked by a truly famous mark to an unrelated product; we can say with some certainty that nearly anyone in the civilized world buying a product titled "Coca-Cola" will associate that product with the soft drink, its packaging, and its advertisements. It is quite another to determine whether a less well known mark will evoke such associations. With respect to less well known marks, inquiries into consumers' mindsets become fuzzier, as likely associations become more dependent on individual backgrounds and experiences.

(2) Fame in a Niche Market: Despite its repeated admonition that only truly prominent and renowned marks deserve protection under § 1125(c)(1), Avery Dennison held that marks famous in only a limited geographic area or a specialized market segment can be "famous" for the purposes of the federal anti-dilution statute. 189 F.3d at 877 ("fame in a localized trading area may meet" the famousness requirement as may fame in a specialized market segment when the "diluting uses are directed narrowly at the same market segment") (internal quotations and citations omitted). We are bound by Avery Dennison to accept and apply the niche fame concept, despite its apparent tension with Avery Dennison's overall message cautioning restraint in applying dilution protections.

Niche fame protection is, however, limited. The statute protects a mark only when a mark is famous within a niche market and the alleged diluter uses the mark within that niche. Id.

Applying that concept here, we conclude, first, that a reasonable fact-finder could find that Trek and Thane operate in the same "narrow market segment." Id. Avery Dennison indicated that a company providing integrated customer care to telephone and Internet customers operates in the same market segment as a company that provides engineering and installation services to the telecommunications industry. Id. at 878 (citing Teletech Customer Care Mgmt. Inc. v. Tele-Tech Co., 977 F. Supp. 1407, 1409-10 (C.D.Cal. 1997)). Similarly, a factfinder could conclude that stationary exercise bicycle manufacturers and stationary elliptical trainer manufacturers share the same market segment.11

A reasonable factfinder could not, however, conclude that mobile bicycles and elliptical orbit machines operate in the same narrow market segment for purposes of the niche fame concept, although both products can be used for exercise. To maintain coherence, the niche fame concept must focus on highly specialized market segments with an identifiable customer base. Where those conditions obtain, participants are likely to make associations between marks that the general public will not make. As the market segments in which the senior and junior products operate become less specialized and less unitary, the notion that participants in those diverse markets will necessarily recognize and form mental associations with an established mark becomes increasingly questionable.

In this case, the smallest market segment that bicycles and elliptical orbit machines could be said to share is the sporting goods market. This is a widely diverse market that encompasses everything from football helmets to ice skates. There is no reason why participants in this broad market will have any particular knowledge about products in submarkets in which they do not participate.

Although a factfinder could conclude that stationary exercise bicycle manufacturers and stationary elliptical trainer manufacturers share the same market segment, it could not reasonably find that TREK is a famous mark in that niche market segment, as opposed to in the market segment frequented by bicycle enthusiasts. One of the eight statutory factors for adjudging famousness is "the duration and extent of use of the mark in connection with the goods and services with which the mark is used." § 1125(c)(1)(B). Trek stationary bicycles were sold for a fairly short time. If not extinct at the time OrbiTrek began selling its elliptical exercise machines, they were at least a threatened species.

Any future plans Trek may have had to reenter the stationary exercise machine market is not pertinent to the famousness inquiry. The federal statute applies only if the junior use "begins after the mark has become famous." § 1125(c)(1). Therefore, any fame Trek may acquire for its mark in the future in the stationary exercise machine market could not preclude OrbiTrek from using its mark in that market.

Where famousness is the question, the extent and duration of the use of a mark within a particular market segment will often be dispositive. No matter the degree of distinctiveness, § 1125(c)(1)(A), the geographic reach of the mark in the pertinent market segment, § 1125(c)(1)(D), the extent to which third parties use similar marks, § 11225(c)(1)(G), or the registration status of the mark, § 1125(c)(1)(H), a mark that is not widely associated with a particular product within a particular niche market is almost surely not famous in that market. The focus of the antidilution statute is on preventing junior users from appropriating or distorting the goodwill and positive associations that a famous mark has developed over the years. Where there has been no successful, long-term development of goodwill with respect to particular markets, asserting fame within that specialized market is simply inconsistent with the purpose of the antidilution protection.12

Trek, it is true, primarily sells bicycles and bicycle accessories, and the evidence is more than sufficient to allow a trier of fact to find that TREK is a famous mark within the narrow market segment devoted to non-stationary bicycle production and sales. But there is no reason to expect that the typical purchaser of stationary exercise machines — particularly those who buy their exercise machines as a result of seeing television infomercials — buys bicycles or bicycle-related products, reads bicycle magazines or watches bicycle competitions on television, any more than any-body else does. That the mark "TREK" is famous with bicycle enthusiasts is therefore of little pertinence in gauging the fame of the mark in the market segment occupied by Thane, the junior user.

(3) Fame Outside a Niche Market: Finally, we consider whether TREK is famous outside any specialized market. While the antidilution statute can protect marks under the niche fame concept from dilution within their specialized markets, the main thrust of the statute is to provide select marks with broad anti-appropriation protection both within and beyond their specialized markets. Such protection is both more potent and more difficult to obtain than niche fame protection. It is more potent because, with important exceptions, § 1125(c)(4), the statute prohibits the use of its mark by any business, regardless of its industry or location. And it is more difficult to obtain because a finding of fame within only a specialized market is not sufficient for protection outside the niche fame context.

The statute does not indicate just how famous a mark must be in order to benefit from the anti-dilution statute's general protection. § 1125(c); TCPIP, 244 F.3d at 98. As the Second Circuit has observed, "the word `famous' is susceptible to widely different understandings," including some that would deem a mark famous even if only a small fraction of the population has heard of it. TCPIP, 244 F.3d at 98-99. If § 1125(c) provided protection to marks that were famous under these broader understandings, the statute would have a crippling effect on the marketing of products, as more and more marks would be off limits for use in any market.

Because some limitation on what qualifies as famous is necessary and because the statutory language provides no guidance in shaping this limitation, Congress likely passed § 1125(c) "counting on courts to understand the legislature's intentions and to interpret the word or phrase in a sensible manner to carry out those intentions." TCPIP, 244 F.3d at 99. To conduct this task we turn to a consideration of the statute's purposes, as expressed in its legislative history.

The legislative history does not contain an express statement regarding how famous a mark must be to merit protection, but it does provide lengthy explanations into the statute's purposes. These purposes provide significant insight into the famousness threshold Congress intended.

Congress passed the anti-dilution measure seeking to protect unauthorized users of famous marks from "attempt[ing] to trade upon the goodwill and established renown of such marks," regardless of whether such use causes a likelihood of confusion about the product's origin. H.R.Rep. No. 104-374, at 3 (1995) reprinted in 1995 U.S.C.C.A.N. 1029, 1030. The legislative history speaks of protecting those marks that have an "aura" and explains that the harm from dilution occurs "when the unauthorized use of a famous mark reduces the public's perception that the mark signifies something unique, singular, or particular." Id.; S.Rep. No. 100-515, at 7 (1988). For example, such harm occurs in the hypothetical cases of "DUPONT shoes, BUICK aspirin, and KODAK pianos," according to the legislative history. Id.

It is clear, then, that for the harm envisioned by Congress to occur, a mark must have achieved enough fame that someone operating a business in a completely different industry than the mark owner could reasonably believe that the mark possesses goodwill from which he could benefit. Congress' hypothetical piano manufacturer can free ride on the aura imbued in KODAK only because potential piano buyers have positive associations with that mark. These potential piano purchasers have warm feelings for KODAK not because the piano and photography industries have any particular overlap — they do not — but because KODAK has achieved fame throughout the population at large. Because the general public, and not just those with special knowledge about photography, perceive KODAK as unique, businesses in all industries may be tempted to trade on KODAK's name.

Thus, only marks that have achieved fame among the general consuming public, as opposed to a more particularized segment of the public, are susceptible to the kind of out-of-market free riding that Congress sought to prevent in passing the antidilution statute. Marks that have not achieved such fame are not susceptible to this free riding outside their own narrow market segment. If a piano manufacturer selects the mark of a photography business little known by the general public but well known to photographers, the piano firm will not benefit much from the goodwill associated with that mark, given the small number of professional photographers in the world. Most of the piano firm's potential customers will not have heard of the mark and will thus have no preconceived perception that the mark symbolizes something "unique, singular, or particular."

In short, for purposes of § 1125(c), a mark usually will achieve broad-based fame only if a large portion of the general consuming public recognizes that mark. Put another way, as a district court recently did, the mark must be a household name. A.B.C. Carpet Co., Inc. v. Naeini, 2002 WL 100604, at * 5 (E.D.N.Y.2002); see also McNeil Consumer Brands, Inc. v. U.S. Dentek Corp., 116 F. Supp. 2d 604, 608 (E.D.Penn.2000). This understanding is suggested by Avery, which limited the possibility of demonstrating fame to a "select class of marks," i.e., those that are "truly prominent and renowned," in order not to "upset the balance in favor of over-protecting trademarks, at the expense of potential non-infringing uses." 189 F.3d at 875. Further, the Avery court chided Avery Dennison for not providing evidence showing that "customers in general have any brand association" with Avery Dennison's marks. Id. at 879 (emphasis added).

In many cases, the list of famousness factors contained in the statute can be quite useful in assisting a fact-finder to determine whether a mark is famous. § 1125(c)(1)(A) to (H). But the party seeking protection must initially make at least a minimal showing of the requisite level of fame. In this case, Trek has presented no evidence that its TREK mark has achieved fame among the general consuming public, as opposed to simply among bicycle enthusiasts.

The closest Trek has come to demonstrating this type of fame has been to produce evidence showing that the champion bicycle racer Lance Armstrong has often been pictured with a Trek bicycle in prominent displays, such as the front page of large circulation newspapers and on Wheaties boxes.13 This incidental appearance of a Trek bicycle does not by itself constitute evidence that the bicycle brand is famous. Many products receive broad incidental media coverage. Such promotion does not lead to the conclusion that their trademarks have become a part of the collective national consciousness. On the other hand, surveys showing that a large percentage of the general public recognizes the brand, press accounts about the popularity of the brand, or pop-culture references involving the brand would provide evidence of fame.14 TCPIP, 244 F.3d at 99; see also Mattel, 296 F.3d at 903 (noting that the commercial success of the Barbie Girl song establishes the fame of the Barbie mark.)

Because TREK has produced no evidence from which a reasonable fact-finder could find that TREK is famous among members of the general consuming public, we conclude that, for reasons different from those given by the district court, Trek's dilution cause of action cannot succeed. Summary judgment was properly granted to OrbiTrek on that cause of action.

As we reverse the judgment for Thane and remand this case to the district court for further proceedings, Thane has not yet met the prevailing party threshold necessary for recovering attorney fees under 15 U.S.C. § 1117(a) or any other authority cited.

Because a reasonable jury could decide the likelihood of confusion issue in favor of either party, we reverse the district court's grant of summary judgment for Thane on those claims that turn on likelihood of confusion. We affirm the district court's grant of summary judgment for Thane on the dilution claims, because Trek has not presented evidence from which a factfinder could conclude that TREK is famous in a pertinent market.

REVERSED in part and REMANDED for proceedings in accord with this opinion.