Carefirst of Maryland, Inc. v. First Care, P.C.

Affirmed by published opinion. Judge MOTZ wrote the opinion, in which Judge TRAXLER and Judge SHEDD joined. OPINION DIANA GRIBBON MOTZ, Circuit Judge: CareFirst of Maryland, Inc., a health maintenance organization associated with Blue Cross Blue Shield, brought this *266 trademark infringement and dilution action against First Care, P.C., a small group of family care physicians located in Southeastern Virginia. The district court granted summary judgment to First Care. We affirm. I. The CareFirst mark first appeared in 1977, when Metropolitan Baltimore Healthcare began marketing prepaid health care plans under the CareFirst name to approximately 3,000 of its Maryland members. In 1989 the then-owner of the CareFirst mark registered “CARE-FIRST” as a trademark, service mark, and collective membership mark with the United States Patent and Trademark Office. Blue Cross and Blue Shield of Maryland acquired the CareFirst mark in 1991. Approximately 250,000 members then held CareFirst plans, but by 1997 that number had declined to 50,000. In late 1997, Blue Cross and Blue Shield of Maryland agreed with other Blue Cross Blue Shield affiliates to operate jointly and collectively under the CareFirst mark. This umbrella organization has spent millions of dollars advertising its mark; in all of its advertisements, it denominates itself “Carefirst BlueCross BlueShield,” often accompanied by the distinctive Blue Cross Blue Shield logo. By the time CareFirst initiated the present action in March 2004, CareFirst had become the largest health maintenance organization in the mid-Atlantic states with 3.2 million members. At least eighty percent of these members reside in CareFirst’s direct service area, which consists of Maryland, Delaware, the District of Columbia, and Northern Virginia. First Care is a Virginia professional corporation of primary care physicians. In late 1994, it registered its corporate name with the state. Since 1995, First Care has operated using the “FIRST CARE” mark. First Care offers traditional family medical services in Portsmouth and Chesapeake, Virginia, which are close to but outside of CareFirst’s direct service area. Approximately 2,500 CareFirst members reside in First Care’s trade area. First Care currently consists of eleven physicians operating out of seven offices; at its height, it had twelve physicians and nine offices. First Care’s name and a description of its services appeared several times in a series of trademark search reports commissioned by CareFirst from January 1996 through November 2000. CareFirst, however, took no action against First Care until 2004, when First Care submitted a deposition in a separate trademark infringement suit that CareFirst was pursuing against another party. On February 13, 2004, after determining that at least ninety CareFirst members had received medical services from First Care, CareFirst sent First Care a cease- and-desist letter. When First Care refused to give up use of its mark, CareFirst brought this action, alleging that since 1995 First Care had infringed on and diluted CareFirst’s trademark, see 15 U.S.C. §§ 1114(1), 1125(a), 1125(c) (2000), and seeking $28 million in damages. In April 2004, First Care applied for state registration of its mark in Virginia. After extensive discovery, the parties filed cross-motions for summary judgment. The district court granted summary judgment to First Care. See Carefirst of Md., Inc. v. First Care, P.C., 350 F.Supp.2d 714, 726 (E.D.Va.2004). The court found that CareFirst had failed to show that a “likelihood of confusion existed between [the] FirstCare and CareFirst” marks and so rejected CareFirst’s infringement claim. Id. Additionally, the court concluded that CareFirst had failed to offer evidence that *267 its mark “was a famous and distinctive mark prior to 1995,” when First Care began operations, and for this reason rejected CareFirst’s dilution claim. Id. We review the district court’s grant of summary judgment de novo. Thus, although First Care argues otherwise, we need not defer to factual findings rendered by the district court. See Patsy’s Brand, Inc. v. I.O.B. Realty, Inc., 317 F.3d 209, 215-16 (2d Cir.2003); Seamons v. Snow, 206 F.3d 1021, 1026 (10th Cir.2000). II. To demonstrate trademark infringement under the Lanham Act, a plaintiff must prove, first, that it owns a valid and protectable mark, and, second, that the defendant’s use of a “reproduction, counterfeit, copy, or colorable imitation” of that mark creates a likelihood of confusion. 15 U.S.C. § 1114(l)(a); Petro Stopping Ctrs., L.P. v. James River Petroleum, Inc., 130 F.3d 88, 91 (4th Cir.1997). Indisputably, CareFirst has established the first element; because more than five years have passed since its registration of the CareFirst mark with the United States Patent and Trademark Office, the protect-ability of its mark has become incontestable. See 15 U.S.C. §§ 1065, 1115(b) (2000); Park ‘N Fly, Inc. v. Dollar Park & Fly, Inc., 469 U.S. 189, 191, 105 S.Ct. 658, 83 L.Ed.2d 582 (1985). Only the second element — likelihood of confusion — is disputed here. Likelihood of confusion exists if “the defendant’s actual practice is likely to produce confusion in the minds of consumers about the origin of the goods or services in question.” KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111, 125 S.Ct. 542, 547, 160 L.Ed.2d 440 (2004); see also Anheuser-Busch, Inc. v. L & L Wings, Inc., 962 F.2d 316, 318 (4th Cir.1992). In conducting the likelihood-of-confusion analysis, a court does not “indulge in a prolonged and minute comparison of the conflicting marks in the peace and quiet of judicial chambers, for this is not the context in which purchasers are faced with the marks.” 3 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 23:58 (4th ed.2005) [hereinafter McCarthy]. Rather, we look to how the two parties actually use their marks in the marketplace to determine whether the defendant’s use is likely to cause confusion. What-A-Burger of Va., Inc. v. Whataburger, Inc., 357 F.3d 441, 450 (4th Cir.2004). 1 To determine if a likelihood of confusion exists, we look to (1) the strength or distinctiveness of the plaintiffs mark as actually used in the marketplace; (2) the similarity of the two marks to consumers; (3)the similarity of the goods or services that the marks identify; (4) the similarity of the facilities used by the markholders; (5) the similarity of advertising used by the markholders; (6) the defendant’s intent; and (7) actual confusion. , Pizzeria *268 Uno Corp. v. Temple, 747 F.2d 1522, 1527 (4th Cir.1984). Not all of these factors are of equal importance, “nor are they always relevant in any given case.” Anheuser-Busch, 962 F.2d at 320. However, evidence of actual confusion is “often paramount” in the likelihood-of-confusion analysis, Lyons P’ship, L.P. v. Morris Costumes, Inc., 243 F.3d 789, 804 (4th Cir.2001). It certainly is of critical importance in the case at hand, as evidenced by the attention given to it by the parties. Accordingly, we turn first to analysis of that factor. A. CareFirst concedes that the only evidence it has proffered of actual confusion is a survey by Dr. Myron Helfgott, which CareFirst commissioned for this litigation. 2 The Helfgott survey interviewed 130 people by telephone, drawn from a list of CareFirst members in or near First Care’s place of business. In the unaided portion of the survey, the surveyors asked the respondents whether they had heard of First Care, whether they thought First Care was related to or affiliated with another health organization, and whether they thought First Care needed permission from another health organization to use its name. The surveyors then repeated these unaided questions, focusing on CareFirst on this second pass. In the aided portion of the survey, the surveyors asked the respondents whether they thought First Care and CareFirst were related to or affiliated with each other. The Helfgott survey does not supply evidence of actual confusion. At best, it shows merely a de minimis level of confusion. Only two of the 130 respondents had both heard of First Care and thought that it was related to or affiliated with Care-First. One additional respondent had not heard of First Care but, judging by its name alone, thought that it might be affiliated with CareFirst. Assuming that all three of these respondents were confused, the survey only shows a confusion rate of 2 percent, hardly a sufficient showing of actual confusion. CareFirst argues that two other groups of respondents should also be considered actually confused: 28 respondents who stated that First Care was affiliated with Blue Cross Blue Shield; and 16 separate respondents who answered “yes” when asked in the aided portion of the survey whether CareFirst and First Care were related. We do not believe that the responses of either of these groups show actual confusion between CareFirst and First Care. CareFirst offers no evidence besides mere speculation that when the 28 respondents indicated that First Care was affiliated with “Blue Cross Blue Shield,” they were actually referring to CareFirst. Indeed, the only evidence on this point suggests the opposite inference: at least two other respondents who indicated First Care’s affiliation with “Blue Cross Blue Shield” specified that they were thinking of Anthem Blue Cross Blue Shield, and both parties agree that First Care is part of Anthem’s network. Similarly, Care-First offers no evidence that the 16 aided respondents were confused in a way relevant to the likelihood-of-confusion analysis. CareFirst’s counsel admitted at oral argument that CareFirst covers the expenses of members who receive medical treatment at First Care clinics, and we know that CareFirst has in fact done so for at least *269 ninety of its members in Southeastern Virginia. The two entities therefore are related, and no evidence suggests that the aided respondents’ answers reflect anything other than this proper — and not confused — understanding of the relationship between the parties. Because we conclude that the Helfgott survey has little probative value, CareFirst lacks any evidence of actual confusion. Although proof of actual confusion is not necessary to show a likelihood of confusion, the absence of any evidence of actual confusion over a substantial period of time — here, approximately nine years— creates a strong inference that there is no likelihood of confusion. See Scotch Whis-ky Ass’n v. Majestic Distilling Co., Inc., 958 F.2d 594, 598 (4th Cir.1992). This important factor thus weighs heavily against CareFirst. B. We next consider the strength of the plaintiffs mark, a factor that Care-First emphasizes. The strength of a mark is the degree to which a consumer in the relevant population, upon encountering the mark, would associate the mark with a unique source. “The ‘strength’ of the trademark is evaluated in terms of its conceptual strength and commercial strength.” GoTo.com, Inc. v. Walt Disney Co., 202 F.3d 1199, 1207 (9th Cir.2000). 3 Measuring a mark’s conceptual or inherent strength focuses on the linguistic or graphical “peculiarity” of the mark, Perini Corp. v. Perini Constr., Inc., 915 F.2d 121, 124 (4th Cir.1990), considered in relation to the product, service, or collective organization to which the mark attaches. See U.S. Search, LLC v. U.S. Search.com Inc., 300 F.3d 517, 524 (4th Cir.2002). The commercial-strength inquiry, by contrast, looks at the marketplace and asks “if in fact a substantial number of present or prospective customers understand the designation when used in connection with a business to refer to a particular person or business enterprise.” See Perini, 915 F.2d at 125 (quoting Food Fair Stores, Inc. v. Lakeland Grocery Corp., 301 F.2d 156, 160-61 (4th Cir.1962)). CareFirst focuses on its registered mark “CareFirst,” standing alone, and argues that this mark is both conceptually and commercially strong. The record evidence renders this argument untenable. CareFirst contends that its registered mark “CareFirst” is conceptually strong simply because the text “CareFirst” is assertedly “suggestive,” as defined by Judge Friendly’s classic delineation in Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 10-11 (2d Cir.1976). The designation of “CareFirst” as *270 “suggestive” may or may not be correct, but this designation does not resolve the mark’s conceptual strength. This is so because many third parties in the health care field have previously used in their own marks the text of the CareFirst mark. “[T]he frequency of prior use of [a mark’s text] in other marks, particularly in the same field of merchandise or service,” illustrates the mark’s lack of conceptual strength. Pizzeria Uno, 747 F.2d at 1530-31; see also First Sav. Bank, F.S.B. v. First Bank Sys., Inc., 101 F.3d 645, 654 (10th Cir.1996); Amstar Corp. v. Domino’s Pizza, Inc., 615 F.2d 252, 259 (5th Cir.1980). “A strong trademark is one that is rarely used by parties other than the owner of the trademark, while a weak trademark is one that is often used by other parties.” Universal Money Ctrs., Inc. v. Am. Tel. & Tel. Co., 22 F.3d 1527, 1533 (10th Cir.1994) (quoting Exxon Corp. v. Tex. Motor Exch, 628 F.2d 500, 504 (5th Cir.1980)). Here, the record undeniably reveals substantial third-party use of the words “Care,” “CareFirst,” “First,” and “First Care” in the health care industry. Care-First’s own Thomson & Thomson trademark search reports state that many health-care-related businesses across the country use these marks; In addition, First Care has submitted dozens of web page print-outs from health-care-related businesses named CareFirst or First Care, as well as an investigator’s report confirming that many businesses with these names are currently active. Cf. U.S. Search, 300 F.3d at 525 (finding a mark “generic, or at most, descriptive” based on web page print-outs, Thomson & Thomson trademark search reports, media references, and other evidence of common usage). If the CareFirst mark were truly a distinctive term, it is unlikely that so many other businesses in the health care industry would independently think of using the same mark or similar variants of it. That so many health care businesses have denominated themselves CareFirst or First Care indicates that “CareFirst” is not a distinctive or unusual term in the industry, and hence not conceptually strong. This evidence of extensive third-party use also demonstrates that CareFirst’s mark lacks commercial strength in many parts of the country. In addition, we find unpersuasive the other evidence that Ca-reFirst has adduced to show the commercial strength of its registered mark standing alone. CareFirst points to the $50 million it has spent on advertising in the mid-Atlantic region over the last decade, its membership of several million individuals, and hundreds of press articles that mention its mark. Most of this evidence, however, does not involve CareFirst’s registered mark standing alone. The record reveals that, at least since 1998, company policy has mandated that CareFirst’s registered mark — “CareFirst”—always appear in public coupled with “Blue Cross Blue Shield.” This joint branding' — the subject of CareFirst’s multi-million-dollar publicity campaign for the last several years — has appeared in all advertisements as well as in provider directories and marketing materials. In addition, CareFirst’s registered mark, accompanied by the Blue Cross Blue Shield language, is usually presented in a corporate logo with a distinctive type-face and a graphic of a blue cross and shield. This sophisticated logo appears to be the standard configuration that CareFirst has used to communicate with the millions of new members it has attracted in the last several years. CareFirst’s evidence of strength thus consists almost entirely of materials that pair the CareFirst mark with the Blue Cross Blue Shield language, usually in a *271 sophisticated corporate logo. 4 This evidence does not show that CareFirst’s registered mark, standing alone, has conceptual or commercial strength. C. The public face of CareFirst’s mark is also relevant in evaluating another factor — the similarity of the parties’ marks. To determine whether two marks are similar, “we must examine the allegedly infringing use in the context in which it is seen by the ordinary consumer.” Anheuser-Busch, Inc., 962 F.2d at 319; see also Restatement (Third) of Unfair Competition § 21(a)(1) (1995). The two marks at issue in this case are mirror images of one another; thus, the bare text of the two is similar. But because the likelihood-of-confusion analysis looks to the actual use of competing marks, a comparison of the texts of the two marks alone is insufficient if the marks have different appearances in the marketplace. See Luigino’s, Inc. v. Stouffer Corp., 170 F.3d 827, 830 (8th Cir.1999); Elvis Presley Enters., Inc. v. Capece, 141 F.3d 188, 198 (5th Cir.1998). The two marks at issue here have very different appearances in the marketplace because the CareFirst mark is almost always paired with the Blue Cross Blue Shield language, while the First Care mark is always presented by itself, or at most with the suffix “P.C.” In contrast to the accouterments surrounding Care-First’s mark in its public appearances, First Care presents its mark plainly and without any graphics. If one of two similar marks is commonly paired with other material, that pairing will serve to lessen any confusion that might otherwise be caused by the textual similarity between the two marks. See, e.g., AutoZone, Inc. v. Tandy Corp., 373 F.3d 786, 797 (6th Cir.2004); Lang v. Retirement Living Pub. Co., Inc., 949 F.2d 576, 581-82 (2d Cir.1991); Amstar Corp., 615 F.2d at 261. This effect is most significant when, as here, the allegedly infringed mark, “CareFirst,” has little independent strength. Because CareFirst’s registered mark is weak, consumers encountering “CareFirst BlueCross BlueShield,” on the one hand, and “First Care,” on the other, are more likely to focus on the differences between the two, particularly when the most salient difference— the addition of “BlueCross BlueShield” — is itself a prominent mark. 5 See 3 McCar *272 thy, supra, § 23:48; cf. Duluth News-Tribune v. Mesabi Publ’g Co., 84 F.3d 1093, 1097 (8th Cir.1996) (conducting comparison of marks using “Duluth News-Tribune” rather than “News-Tribune” alone because “News-Tribune” was a weak mark). Thus, the substantial differences in the public presentations of these marks significantly reduce the likelihood of confusion, notwithstanding their textual similarity. See Petro, 130 F.3d at 94 (finding two uses of “Petro” not confusing due to the marks’ different colors, different accompanying words, and different graphics). D. As with the similarity of the marks, we measure the similarity of services with respect to each party’s actual performance in the marketplace. See Anheuser-Busch, 962 F.2d at 319. 6 Although this factor does not particularly assist First Care’s defense, neither does it aid CareFirst. The services offered by the parties here are no closer than the services we found dissimilar in Petro. The defendant in that case only sold fuel, while the plaintiff offered multiple other services in addition to selling fuel. Petro, 130 F.3d at 95. We held that despite this overlap, the parties offered dissimilar services. Id. Here, the actual services offered by the parties do not even overlap. First Care only offers direct medical services to individuals. Ca-reFirst does not; rather, it contracts with participating providers who agree to treat CareFirst members. But we recognize that in Petro we were reviewing a district court’s factual findings after trial and so applied a more deferential standard of review in reaching our conclusion about dissimilarity of services. Id. at 91-92. Thus, we do not believe that Petro mandates a holding that the services offered by CareFirst and First Care are so dissimilar as to militate against a finding of likelihood of confusion. Based on the record before us, however, we cannot hold that CareFirst significantly advances its likelihood-of-eonfusion claim by comparing its services — -health plans offered by a multi-million-member HMO — to those of First Care — direct medical care offered by a small group of physicians. E. None of the other Pizzeria Uno factors offer any support for CareFirst’s infringement claim. CareFirst contends that the parties have similar facilities because First Care’s offices are similar to the doctors’ offices of *273 CareFirst’s participating providers. Even assuming that this comparison is appropriate, this factor does not weigh in Care-First’s favor when there are “basic differences between plaintiffs and defendant’s modes of distributing their products” at these facilities. Amstar Corp., 615 F.2d at 262. First Care displays its mark on the outside of each of its doctors’ offices and provides direct medical services under its name. By contrast, CareFirst’s mark enjoys no similarly prominent placement with its participating providers, 7 and nothing in the record suggests that its participating providers directly market Care-First’s health care policies in their offices. Even if the offices of First Care’s doctors and CareFirst’s participating providers were otherwise identical, the significant differences between how consumers encounter the parties’ respective marks in these facilities reduce the weight of this factor. See Petro, 130 F.3d at 95. CareFirst also argues that the parties have similar advertising. In comparing advertising, we look at a variety of factors: the media used, the geographic areas in which advertising occurs, the appearance of the advertisements, and the content of the advertisements. Petro, 130 F.3d at 95; Pizzeria Uno, 747 F.2d at 1535. Here, at least three of these four factors show significant differences between the two parties. First, CareFirst concedes that First Care “advertises” primarily through word of mouth, and at any rate spends less than $2,000 per year on advertising. By contrast, CareFirst spends millions of dollars every year to advertise extensively in print, television, radio, the Internet, and elsewhere. Second, CareFirst advertises primarily in its direct service area, which does not include First Care’s location in Southeastern Virginia. First Care only advertises in its local area. That there is some spillover of CareFirst’s advertising into First Care’s area does not make their advertising similar. See Pizzeria Uno, 747 F.2d at 1535. Third, based on the record before us, CareFirst’s professionally designed advertisements contrast sharply with First Care’s advertising, which is limited to “external and internal signage, basic listings in local telephone directories, and running announcements in local newspapers.” CareFirst next argues that First Care engaged in bad faith by using the First Care mark. CareFirst acknowledges that First Care initially used the mark in good faith but contends that First Care evidenced bad faith by applying to register the mark in Virginia in April 2004, one month after CareFirst had filed this suit. The intent of a junior user is relevant only if the junior user intended to capitalize on the good will associated with the senior user’s mark. See Sara Lee Corp. v. Kayser-Roth Corp., 81 F.3d 455, 466 (4th Cir.1996); Pizzeria Uno, 747 F.2d at 1535. CareFirst proffers no evidence as to why First Care filed the state application. The fact of the state application, in and of itself, simply does not show an intent to capitalize on the good will associated with CareFirst’s mark. The state registration would do nothing to assist First Care in trading on CareFirst’s good will since the Lanham Act registration preempts any registration under state law. We cannot infer from this mere filing for state registration that First Care intended to capitalize on CareFirst’s good will when this action would have done nothing to achieve that purpose. *274 F. In sum, none of the Pizzeria Uno factors weigh in CareFirst’s favor, and most—including absence of evidence of actual confusion, lack of strength of Care-First’s mark, and lack of similarity between the two marks—weigh heavily against a finding of likelihood of confusion. For these reasons, we can only hold that the district court did not err in granting summary judgment to First Care on Care-First’s infringement claim. III. The district court also granted summary judgment to First Care on Care-First’s dilution claim. The Trademark Dilution Act provides that “[t]he owner of a famous mark” can enjoin “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.” 15 U.S.C. § 1125(c)(1). Dilution means “the lessening of the capacity of a famous mark to identify and distinguish goods or services.” Id. § 1127. The statutory scheme makes clear that a court may find dilution even if it would not find any likelihood of confusion. Id. The Supreme Court has held that the dilution statute “unambiguously requires a showing of actual dilution, rather than a likelihood of dilution.” Moseley v. V Secret Catalogue, Inc., 537 U.S. 418, 433, 123 S.Ct. 1115, 155 L.Ed.2d 1 (2003). CareFirst offers only two arguments in its attempt to establish that it could prove actual dilution. Neither is persuasive. First, CareFirst maintains that the similarity of CareFirst’s and First Care’s marks provides circumstantial evidence of actual dilution. The Supreme Court has suggested that “where the junior and senior marks are identical,” there would be “circumstantial evidence” of actual dilution. Moseley, 537 U.S. at 434, 123 S.Ct. 1115 (emphasis added). “[E]very federal court to decide the issue has ruled that a high degree of similarity, ranging from ‘nearly identical’ to ‘very similar,’ is required” to meet this standard. AutoZone, 373 F.3d at 806. Thus, “a mere similarity in the marks—even a close similarity—will not suffice to establish per se evidence of actual dilution.” Savin Corp. v. The Savin Grp., 391 F.3d 439, 453 (2d Cir.2004). In addition, “the issue of whether the marks are identical will be context- and/or media-specific and factually intensive in nature.” Id. Here, the text of the two marks is similar but not identical. Furthermore, as they appear in the marketplace to the average consumer, the two marks are not even “very similar”— company policy requires that CareFirst’s mark always appear as “CareFirst BlueCross BlueShield,” while the First Care mark appears unadorned. Therefore, the similarity of the marks provides little circumstantial evidence of actual dilution here. ? second argument— that the Helfgott survey provides evidence of actual dilution—is no more convincing. As noted above, the Helfgott survey was designed to measure likelihood of confusion, not dilution. See Avery Dennison Corp. v. Sumpton, 189 F.3d 868, 875 (9th Cir.1999) (“In the dilution context, likelihood of confusion is irrelevant.”); 4 McCarthy, supra, § 24:94.2 (describing surveys that might measure dilution). Thus, the survey at best measured whether consumers believe that First Care is associated with CareFirst, not whether First Care’s mark has “reduce[d] the capacity of the [CareFirst] mark to identify the [services] of its owner.” Moseley, 537 U.S. at 433, 123 S.Ct. 1115. The survey’s results provide no forecast of actual dilution. See Ringling Bros.-Barnum & Bai *275 ley Combined Shows, Inc. v. Utah Div. of Travel Dev., 170 F.3d 449, 462-63 (4th Cir.1999) (finding that a survey designed only to show “mental association” between competing marks was not evidence of dilution). Accordingly, the district court did not err in granting First Care summary judgment on CareFirst’s dilution claim. IV. For the foregoing reasons, the judgment of the district court is AFFIRMED. Footnotes [1] CareFirst argues that infringement of its collective membership mark only requires likelihood of confusion among its members. This suggestion is contrary to the rule that "[(likelihood of confusion of collective marks with other types of marks is determined according to the standard rules of trademark law.” 3 McCarthy, supra, § 19:101. The cases cited by CareFirst do not hold to the contrary. For example, In re Code Consultants, Inc., 60 U.S.P.Q.2d 1699 (T.T.A.B. 2001), merely states that the likelihood-of-confusion analysis for collective membership marks should not be limited to purchasers; it nowhere suggests that this analysis should be limited to members instead. Indeed, the decision notes that “ ‘relevant persons' would encompass all who might know of their services.” Id. at 1701 (quoting Elec. Design & Sales v. Elec. Data Sys., 954 F.2d 713, 716 (Fed.Cir.1992)). [2] We note at the outset that Dr. Helfgott never intended that his survey show actual confusion; indeed, he was careful to note that "this is a likelihood of confusion study, and not a study of actual confusion.” However, our criticism of the Helfgott survey applies regardless of how the survey is characterized. [3] CareFirst argues that its incontestable registration conclusively establishes the strength of its mark. That contention "confuses the issue of a trademark's validity with the separate inquiry into a mark’s strength for purposes of the likelihood of confusion determination.” Petro Stopping Ctrs., L.P. v. James River Petroleum, Inc., 130 F.3d 88, 92 (4th Cir.1997). "[I]ncontestability alone does not establish that the trademark is strong and therefore likely to cause the consumer confusion necessary for a finding of trademark infringement.” Lone Star Steakhouse & Saloon, Inc. v. Alpha of Va., Inc., 43 F.3d 922, 934 (4th Cir.1995) (citing cases). Moreover, although the inquiry into whether a mark has secondary meaning sufficient to be protectable under the Lanham Act is similar to the commercial-strength inquiry, see 2 McCarthy, supra, § 11:83, the two are analytically independent. Thus, like the plaintiff in Petro, see 130 F.3d at 92-94, although CareFirst need not show secondary meaning, it still must demonstrate commercial strength. See Therma-Scan, Inc. v. Thermoscan, Inc., 295 F.3d 623, 632 (6th Cir.2002) ("Even where a trademark is incontestable ... the significance of its presumed strength will depend upon its recognition among members of the public.”). [4] CareFirst seeks to counter this inevitable conclusion by noting that its mark appears unadorned and without the Blue Cross Blue Shield language in unsolicited press accounts and word-of-mouth referrals, and that it promoted its name for 24 years without the Blue Cross Blue Shield language. This argument is unpersuasive. Every mark will appear in public in a variety of formats, and companies frequently change the public presentation of their marks. In considering the appearance of a mark for purposes of the likelihood-of-confusion analysis, we must weigh more heavily the predominant manner in which the contemporary public perceives the mark. For the last decade, the CareFirst mark coupled with the Blue Cross Blue Shield language and logo has been the subject of $54.5 million in advertising and promotion in a wide variety of media and has been attached to approximately 12 million newsletters annually. The impact of this voluminous, expensive campaign decidedly outweighs unsolicited press accounts and verbal referrals. Furthermore, CareFirst changed the public presentation of its mark to include the Blue Cross Blue Shield text and logo at least seven years ago; its previous marketing of the Ca-reFirst mark standing alone reached a relatively small number of people compared to the millions exposed to its joint brand during the last few years of explosive growth. The record thus shows that the CareFirst mark, standing alone, no longer has a significant presence with the contemporary public. [5] Indeed, CareFirst's survey suggests that its own members strongly identify CareFirst with "BlueCross BlueShield” rather than with "CareFirst.” When asked to give the name of *272 their health care company, 44 percent of the CareFirst members surveyed said their insurance came from "BlueCross BlueShield”; 10 percent answered "CareFirst BlueCross BlueShield”; and only 10 percent answered "CareFirst” alone. These responses suggest that even among CareFirst's own members, the Blue Cross Blue Shield language is more distinctive than the CareFirst mark. [6] Because the marketplace provides the relevant forum for comparing services, there is no merit to CareFirst's claim that we should conduct this analysis using the services enumerated in CareFirst’s federal registrations. The case upon which CareFirst principally relies — Team Tires Plus, Ltd. v. Tires Plus, Inc., 394 F.3d 831 (10th Cir.2005)- — -is inappo-site. The district court in that case had "never analyzed the likelihood of confusion caused by the defendant’s use of the Tires Plus name,” id. at 833, in part because it had incorrectly interpreted the language of the plaintiff’s federal registration. Id. at 834-35. The Tenth Circuit reversed and ordered the district court to engage in the likelihood-of-confusion analysis that it had avoided. Id. at 835. At no point did the court suggest, as CareFirst does here, that the district court’s analysis of the likelihood-of-confusion factors should ignore the plaintiff's actual use of its mark. [7] CareFirst once operated a line of CareFirst-branded clinics, but it discontinued the clinics before First Care began operating.