425 F.3d 211 (2005)
CENTURY 21 REAL ESTATE CORPORATION; Coldwell Banker Real Estate Corporation; Era Franchise Systems, Inc.
v.
LENDINGTREE, INC., Appellant.

No. 03-4700.
United States Court of Appeals, Third Circuit.

Argued December 6, 2004.
October 11, 2005.
Bruce I. Goldstein, Saiber, Schlesinger, Satz & Goldstein, Newark, NJ, Jeffrey A. Conciatori [Argued], Quinn, Emanuel, Urquhart, Oliver & Hedges, New York, NY, for Appellant LendingTree, Inc.

Stephen W. Feingold [Argued], Pitney Hardin, New York, NY, for Appellees Century 21 Real Estate Corporation; Coldwell Banker Real Estate Corporation; Era Franchise Systems, Inc.

Before RENDELL, FISHER, Circuit Judges, and YOHN,[*] District Judge.

FISHER, Circuit Judge, concurring in part and dissenting in part.

OPINION OF THE COURT

RENDELL, Circuit Judge.

This case presents an opportunity for us to consider the contours of the traditional test for trademark infringement where the defendant asserts the defense of “nominative fair use.” More specifically, we must determine what role likelihood of confusion plays in a trademark infringement case where the defendant claims that its use was nominative and fair.

Appellees, Century 21, Coldwell Banker and ERA (“CCE”) complain that Appellant Lending Tree (“LT”), in the process of marketing its mortgage services, improperly referenced CCE’s trademarked services. LT contends that its use was nominative and fair, and permitted as a matter of law.

“Nominative” fair use is said to occur “when the alleged infringer uses the [trademark holder’s] product, even if the alleged infringer’s ultimate goal is to describe his own product. Nominative fair use also occurs if the only practical way to refer to something is to use the trademarked term.” KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 328 F.3d 1061, 1072 (9th Cir.2003) (rev’d. on other grounds) (quotations omitted). By contrast, “classic” fair use occurs where the defendant uses the plaintiff’s mark to describe the defendant’s own product. New Kids on the Block v. News America Pub., Inc., 971 F.2d 302, 308 (9th Cir.1992).

The use of the term “Volkswagen” by a car mechanic in an ad describing the types of cars he repairs has been held to constitute a nominative fair use. See id. at 307 (citing Volkswagenwerk Aktiengesellschaft v. Church, 411 F.2d 350 (1969)). Clearly, the mechanic is referring to another’s product, but does so in order to describe what he does. On the other hand, the use of the term “micro-colors,” a registered trademark of one make-up company, referring to the pigments of the product of a different and competing make-up company that it used in its own product, was classified as a classic fair use. See KP Permanent Make-Up, Inc., 328 F.3d at 1072. There, the reference to the mark of another was made in describing its own product and its attributes.

Traditionally, we have looked to whether a trademark is likely to cause confusion in order to determine whether a violation of the Lanham Act has occurred and, thus, whether the use should be enjoined and prohibited. However, it is unclear what role “likelihood of confusion” plays in the analysis when “fair use” is asserted as a defense. Recently, the United States Supreme Court provided guidance to the courts regarding the test for classic fair use in KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111, 125 S.Ct. 542, 160 L.Ed.2d 440 (2004). The issue before us is the extent to which its reasoning applies to the nominative fair use analysis as well.

I. Factual and Procedural Background

Appellees Century 21 and ERA have each been in business for over 30 years. Coldwell Banker has been in business for almost 100 years. Each of these real estate companies oversees a system of franchisees who operate by reference to the franchisor’s trademark (e.g., Century 21 Smith Realty). Franchise agreements permit those brokerage companies to provide realty services under trademarks held by Cendant Corporation. Every franchisee is granted a license to use its franchisor’s trademark only in connection with its “d/b/a” name. (Thus, a franchisee could only refer to itself as Century 21 Smith Realty and not just Century 21). There are over 8,200 franchisees in CCE’s collective systems in the U.S.

Appellant LendingTree describes itself as a diversified consumer-oriented Internet business that helps consumers identify and select qualified lenders, real estate brokers, auto insurers, and other financial service companies. It has a real estate referral service that consumers can access by visiting its website and inputting the location and characteristics of the house they are seeking to buy or sell. LT then selects and transmits information about up to four real estate companies participating in LT’s referral network that service that community. If consumers ultimately use an LT referred broker, they receive rewards, such as airline frequent flier miles and gift cards. LT also has an established mortgage referral program based on contractual relationships with participating financial institutions identified on its website.

At the time of this action, LT’s real estate referral network consisted of more than 650 real estate broker member companies in the U.S., which collectively operated more than 2,500 offices. More than 9,000 real estate agents in those offices were registered participants in LT’s network. At least 257 of the approximately 650 real estate broker member companies participating in the LT network operated a Century 21, Coldwell Banker, or ERA franchise.

LT’s real estate referral network was formed in 1998 by HomeSpace, a company from which LT acquired certain assets in August 2000. As early as 1999, HomeSpace publicly advertised its real estate referral network in printed materials as “including brokers representing the nation’s leading franchises, such as Coldwell Banker, Century 21 … and ERA.”

The alleged infringement here is based on the following uses of CCE’s marks:

(1) A Coldwell Banker “For Sale” sign with a woman, purporting to be a real estate agent, next to it, on which the blue and white Coldwell Banker logo was somewhat obscured by the word “SOLD.” LT’s phone number was at the bottom. This scene was depicted at the bottom of LT’s homepage on its website.

(2) A statement by LT on its “Find a Realtor” homepage stating that LT will “give you access to a national network of brokers representing the country’s leading real estate companies, including Coldwell Banker, ERA and Century 21.” These three names headed a bullet-pointed list of all such realtors to whom LT promised access. The marks on those pages were in block letter format.

(3) LT’s statement on its website’s Help Center that LT is “[r]epresented by large independent real estate companies and members of major franchises — Coldwell Banker, Century 21, Prudential, ERA, ReMAX, GMAC (formerly Better Homes & Gardens), and Realty Executives.”

(4) LT’s use of printed marketing materials that stated that “LT is affiliated with more than 700 certified brokers such as Coldwell Banker, Century 21, Prudential, ERA and RE/MAX.”

In January 2003, Kathryn Geib, in-house counsel for CCE’s parent company responsible for the enforcement of its trademarks, wrote to LT to demand that it stop using CCE’s “marks” on its website in any manner in the operation of its business. At that time, LT was using CCE’s logos on its website. After receipt of the letter, LT stopped using the logos (or any other of CCE’s marks) on that webpage, but continued to use CCE’s marks in block letter form on other webpages. In March 2003, Geib learned that LT was using a marketing coupon containing the “affiliated language” described above. Geib sent a letter asking LT to stop such use. In May 2003, CCE discovered that LT was using CCE’s marks, but in block letters, on its webpage. Geib again wrote, asking LT to stop such use.

Not satisfied with the response from LT, CCE commenced this action and applied for a preliminary injunction against LT’s use of its marks, claiming unfair competition and trademark infringement in violation of §§ 32 and 43(a) of the Lanham Act, 15 U.S.C. §§ 1114(1) and 1125(a). At issue in the case before the District Court was the use of CCE’s marks in block form in bullet points on LT’s webpage, marketing materials using the “affiliated with” language described above, and the Coldwell Banker “For Sale” sign.

After oral argument before the District Court, LT voluntarily implemented certain modifications to its website. It changed the background color of the “For Sale” sign from blue to red and removed the phone number at the bottom of the sign, moved the position of the bullet points with plaintiffs’ names on them from first to last in its “Find a Realtor” list, and added a disclaimer to its real estate homepage that “LT is not sponsored by or affiliated with the parent franchisor companies of any of the participating members of its network.” LT also added a note to a pop-up screen saying that “A particular franchisee in your geographic area may not be available to you through the LendingTree network if that franchisee is not a participating member of the network. In addition, it is possible that the network may not include any franchisee of a particular national company in a given area.” (Text from www.LendingTree.com at JA689.) LT notified the District Court of these changes and represented that all of the language allegedly giving rise to CCE’s motion had been permanently removed from LT’s website and would not be used in the future.

The District Court issued its ruling with respect to the language used by LT at the time the complaint was filed, as well as the modified language. The District Court determined that LT’s use of Appellees’ names was likely to cause consumer confusion, that the nominative use defense did not shield LT in this instance, and thus, granted CCE’s motion for a preliminary injunction as to the use of CCE’s marks and the “affiliation” language, but not the “For Sale” sign. The District Court’s analysis did not separate out the original language from the later modified language, as the District Court seemed to conclude that the revisions did not alleviate the problem. LT now appeals this issuance of the preliminary injunction.

On appeal, we must first decide whether the case was moot when the District Court issued its ruling, given the fact that LT had made several changes to the allegedly infringing language and the picture on its website. If the case was not moot, we must then determine the proper test to apply in the situation where a defendant is asserting a nominative fair use defense to claims of unfair competition and trademark infringement under the Lanham Act.

II. Jurisdiction

The District Court had jurisdiction to hear this case pursuant to 28 U.S.C. §§ 1331 and 1338. We have jurisdiction over this interlocutory appeal of a preliminary injunction under 28 U.S.C. § 1292(a)(1).

III. Mootness

Although a case may become moot “if the defendant can demonstrate that there is no reasonable expectation that the wrong will be repeated, this burden is a heavy one …” and has not been met here. United States v. W.T. Grant Co., 345 U.S. 629, 633, 73 S.Ct. 894, 97 L.Ed. 1303 (1953) (quotation omitted). The modified language that LT employed still referenced CCE’s marks and, thus, is potentially infringing and it cannot be said that LT’s voluntary discontinued use of certain language completely obviates the need for injunctive relief. See Lyons Partnership v. Morris Costumes, Inc., 243 F.3d 789, 800 (4th Cir.2001) (citation omitted). For instance, the use of CCE’s marks in block letter format could constitute trademark infringement in certain circumstances. Whether these circumstances are present in this case is still a question that needs to be resolved.

Furthermore, were we to hold that LT’s voluntary cessation of the alleged infringing activities rendered the case moot, this would potentially mean that LT would “simply be free to return to [its] old ways after the threat of a lawsuit had passed.” Iron Arrow Honor Soc. v. Heckler, 464 U.S. 67, 72, 104 S.Ct. 373, 78 L.Ed.2d 58 (1983) (quotation omitted). Therefore, the case is not moot and we will proceed to analyze the substantive claims of trademark infringement and unfair competition.

IV. Fair Use

It must be recognized at the outset that “fair use” presents a fact pattern different from that of a normal infringement suit. The typical situation in a trademark case involves the defendant’s having passed off another’s mark as its own or having used a similar name, confusing the public as to precisely whose goods are being sold. Likelihood of confusion is the sole issue. But the fair use defense, by reason of the circumstances giving rise to its applicability, alters the premise somewhat. The defendant is not purporting to be selling goods or services that the plaintiff has trademarked, but, rather, is using plaintiff’s mark in order to refer to defendant’s own goods or to the goods of the trademark owner in a way that might confuse the public as to the relationship between the two. See Cairns v. Franklin Mint Co., 292 F.3d 1139, 1151 (9th Cir.2002). Accordingly, the legal framework still involves a showing that A’s reference to B’s mark will likely confuse the public, but the analysis does not end there, for the use may nonetheless be permissible if it is “fair.”

In KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111, 125 S.Ct. 542, 545-46, 160 L.Ed.2d 440 (2004), the Supreme Court rejected the notion that, in the context of classic fair use, the party asserting the fair use defense to a claim of trademark infringement had any burden to negate the likelihood that the practice complained of will confuse consumers about the origin of the services or goods affected. Instead, plaintiff has the exclusive burden to demonstrate likelihood of confusion, and then defendant’s burden is only to show the affirmative defense of fair use. The Supreme Court stated, “[s]ince the burden of proving likelihood of confusion rests with the plaintiff, and the fair use defendant has no free-standing need to show confusion unlikely … it follows that some possibility of consumer confusion must be compatible with fair use….” Id. at 550, 125 S.Ct. 542. Thus, consumer confusion and fair use are not mutually exclusive. The latter will in essence rebut or excuse the former so that the use is permissible.

Before the Supreme Court spoke on the issue of classic fair use, the Court of Appeals for the Ninth Circuit had charted a path through a different fair use analysis — nominative fair use. In New Kids on the Block, 971 F.2d at 308, the Court of Appeals for the Ninth Circuit adopted its own test governing the nominative fair use analysis where the marks are used, as they are here and in the case of the mechanic’s ad described above that referenced Volkswagen, to refer to the plaintiff trademark owner’s product in order to help better describe the defendant’s product or service. Thereafter, it refined the test and clarified its application. See, e.g., Playboy Enterprises, Inc. v. Welles, 279 F.3d 796 (9th Cir.2002) (holding that the use of the terms “Playboy,” “Playmate,” and “Playmate of the Year 1981” on the website masthead and banner ads and in the metatags of the website of a former Playmate of the Year were nominative fair uses because they served to identify the defendant and did not imply current sponsorship or endorsement); Cairns v. Franklin Mint Co., 292 F.3d 1139 (9th Cir.2002) (holding that the sale of collectibles bearing the name and likeness of Princess Diana was a nominative fair use).

Few other courts have spoken on the precise issue of how nominative fair use is successfully invoked. Indeed, it seems that only the Second, Fifth, and Sixth Circuits have referenced the nominative fair use defense by name and even on these occasions have done so only to refer to what district courts had done with the issue or to decline to adopt the Ninth Circuit’s test as a whole. See Chambers v. Time Warner, Inc., 282 F.3d 147, 156 (2d Cir.2002) (noting that the district court had applied the standard for nominative fair use as articulated by the Ninth Circuit, but finding that the court had erred in its application); see also Pebble Beach Co. v. Tour 18 I Ltd., 155 F.3d 526, 546 n. 13 (5th Cir.1998) (adopting the Ninth Circuit’s nominative fair use test in part); Interactive Products Corp. v. a2z Mobile Office Solutions, Inc., 326 F.3d 687, 698 n. 6 (6th Cir.2003) (footnoting why a district court case involving the nominative fair use defense was distinguishable from the case before it); PACCAR Inc. v. TeleScan Technologies, L.L.C., 319 F.3d 243, 256 (6th Cir.2003) (declining to follow the Ninth Circuit’s analysis (formulated prior to the Supreme Court’s holding in KP Permanent Make-Up) and applying the likelihood of confusion test instead).

Furthermore, this issue is one of first impression in our Court. See Basic Fun, Inc. v. X-Concepts, 157 F.Supp.2d 449, 456 (E.D.Pa.2001) (stating that “the Third Circuit does not recognize the `nominative’ fair use defense, which is utilized as a defense solely in the Ninth Circuit”). The United States District Court for the District of New Jersey did venture into the area of nominative fair use defense in Liquid Glass Enterprises, Inc. v. Dr. Ing h.c.F. Porsche AG and Porsche Cars North America, Inc., 8 F.Supp.2d 398 (D.N.J.1998). That case involved the use of Porsche’s trademarks by Liquid Glass, a corporation that sold car care products. Liquid Glass used Porsche’s trademarks in its advertisements for car polish — in particular, an ad involving a provocatively dressed woman polishing a Porsche 911 with the trademark “PORSCHE” plainly visible. The court initially analyzed the case under the nominative fair use defense articulated by the Ninth Circuit Court of Appeals in New Kids On The Block, however, it conflated the test with the plaintiff’s burden of proving likelihood of confusion. See Liquid Glass, 8 F.Supp.2d at 403 (finding that Liquid Glass’ fair use defense was defeated, in part, because Porsche would “likely prevail in demonstrating that the use of Porsche’s trademarks … in Liquid Glass’s advertisements would likely confuse consumers as to Porsche’s connection with Liquid Glass”). The Court reviewed each of the traditional factors used to determine the likelihood of confusion, though this analysis is not even a part of the New Kids On The Block test. This court’s difficulty in applying the traditional test for likelihood of confusion together with the nominative fair use defense highlights the need to clarify the proper analysis in this area of the law. Given the paucity of the case law on this subject from every court other than the Court of Appeals for the Ninth Circuit, and that Court’s extensive study of this issue, we will consider the opinions of that Court.

In New Kids On The Block, the Court of Appeals for the Ninth Circuit considered whether two separate newspapers that had conducted polls asking readers to vote for their favorite New Kid On the Block, and charging money for each telephone vote cast, had employed the “New Kids” trademark in such a way as to violate the Lanham Act. The trademarked name was very valuable to the New Kids, as it had been used to sell over 500 different products, the profits of which partially inured to the benefit of the New Kids. New Kids alleged that the use of the name in conducting the polls implied that the New Kids were sponsoring the polls.

The Court began with a discussion of the history of trademark law, noting that:

[a] trademark is a limited property right in a particular word, phrase or symbol. And although the English language is a language rich in imagery, we need not belabor the point that some words, phrases or symbols better convey their intended meanings than others … Thus, trademark law recognizes a defense where the mark is used only to describe the goods or services of [a] party, or their geographic origin. 15 U.S.C. § 1115(b)(4). The `fair-use’ defense, in essence, forbids a trademark registrant to appropriate a descriptive term for his exclusive use and so prevent others from accurately describing a characteristic of their goods.
New Kids On The Block, 971 F.2d at 306 (quoting Soweco, Inc. v. Shell Oil Co., 617 F.2d 1178, 1185 (5th Cir.1980)).

The court reasoned that nominative fair use cases — “the class of cases where the use of the trademark does not attempt to capitalize on consumer confusion or to appropriate the cachet of one product for a different one” — are not the typical fair use cases. It described the Volkswagen use as an example of this atypical type of fair use:

A good example of this is Volkswagenwerk Aktiengesellschaft v. Church, 411 F.2d 350 (9th Cir.1969), where we held that Volkswagen could not prevent an automobile repair shop from using its mark. We recognized that in “advertising [the repair of Volkswagens, it] would be difficult, if not impossible, for [Church] to avoid altogether the use of the word `Volkswagen’ or its abbreviation `VW,’ which are the normal terms which, to the public at large, signify appellant’s cars.” Id. at 352. Church did not suggest to customers that he was part of the Volkswagen organization or that his repair shop was sponsored or authorized by VW; he merely used the words “Volkswagen” and “VW” to convey information about the types of cars he repaired. Therefore, his use of the Volkswagen trademark was not an infringing use.
New Kids On The Block, 971 F.2d at 307.

The Court then reasoned that such a situation should be viewed as “involving a non-trademark use of a mark …”:

Such nominative use of a mark — where the only word reasonably available to describe a particular thing is pressed into service — lies outside the strictures of trademark law: Because it does not implicate the source-identification function that is the purpose of trademark, it does not constitute unfair competition; such use is fair because it does not imply sponsorship or endorsement by the trademark holder.
Id. at 308.

The Court distinguished “nominative” fair use from “classic” fair use, noting that if defendant’s use of the trademark referred to something other than the plaintiff’s product, traditional fair use inquiry would continue to govern.

The court then articulated its own test for nominative fair use:

[W]here the defendant uses a trademark to describe the plaintiff’s product, rather than its own, we hold that a commercial user is entitled to a nominative fair use defense provided he meets the following three requirements: First, the product or service in question must be one not readily identifiable without use of the trademark; second, only so much of the mark or marks may be used as is reasonably necessary to identify the product or service; and third, the user must do nothing that would, in conjunction with the mark, suggest sponsorship or endorsement by the trademark holder.
Id.

In announcing this new test, New Kids On The Block rejected traditional trademark infringement analysis. It held that this test replaces the “likelihood of confusion” test for trademark cases where nominative fair use is asserted. See Cairns v. Franklin Mint Co., 292 F.3d 1139, 1151 (9th Cir.2002). In a more recent case, the Ninth Circuit Court of Appeals has elaborated on its reasoning for this shift. Application of the likelihood of confusion test, which focuses on the similarity of the mark used by the plaintiff and defendant, “would lead to the incorrect conclusion that virtually all nominative uses are confusing” because “[w]hen a defendant uses a trademark nominally, the trademark will be identical to the plaintiff’s mark, at least in terms of the words in question.” Playboy Enters., 279 F.3d at 801. Thus, the likelihood of confusion test as applied in nominative fair use cases would disadvantage the defendant by making confusion an all but foregone conclusion.

While we agree with the Ninth Circuit Court of Appeals that a distinct analysis is needed for nominative fair use cases, we do not accept the legal basis or advisability of supplanting the likelihood of confusion test entirely. First, we do not see nominative fair use as so different from classic fair use as to warrant such different treatment. The Ninth Circuit Court of Appeals believed that the two types of fair use could be distinguished on the basis that nominative fair use makes it clear to consumers that the plaintiff, not the defendant, is the source of the trademarked product or service, while classic fair use does not. Thus, the Ninth Circuit Court of Appeals believed that a different analysis was appropriate for nominative fair use and that it could abandon the need for proof of confusion in these circumstances. New Kids On The Block, 971 F.2d at 307-08.[1] Yet, it is clear to us that even a defendant’s nominative use has the potential of confusing consumers with respect to its products or services. Since the defendant ultimately uses the plaintiff’s mark in a nominative case in order to describe its own product or services, Cairns, 292 F.3d at 1151 & n. 8, even an accurate nominative use could potentially confuse consumers about the plaintiff’s endorsement or sponsorship of the defendant’s products or services. Thus, we disagree with the fundamental distinction the Ninth Circuit Court of Appeals drew between classic and nominative fair use.

In addition, the approach of the Court of Appeals for the Ninth Circuit would relieve the plaintiff of the burden of proving the key element of a trademark infringement case — likelihood of confusion — as a precondition to a defendant’s even having to assert and demonstrate its entitlement to a nominative fair use defense. The Supreme Court in KP Permanent Make-Up clearly established that it was plaintiff’s burden in a classic fair use case to prove likelihood of confusion. There, the Court noted the difference between fair use and other trademark infringement claims, opining, as we stated above, that likelihood of confusion and fair use can coexist. This does not mean that we should remove the need for finding confusion in the first instance. Instead, once the plaintiff proves likelihood of confusion, defendant only had to show that defendant’s use, even if confusing, was “fair.”[2]

This view finds support not only in the Supreme Court’s recent opinion, but also in the relevant statutory framework. The very language of the Lanham Act leads us to conclude that likelihood of confusion is an essential indicator of whether or not trademark infringement has occurred. Both §§ 32 and 43(a) of the Lanham Act, allegedly violated in this case, forbid use of words or marks in a way which is likely to cause confusion as to the origin, sponsorship, or approval of goods or services. Surely the plaintiff’s success in its claim must rely on a finding in this regard. Given this, we decline to read this requirement out of a case alleging trademark infringement.

We are thus left with the firm conviction that the burden of proving likelihood of confusion should remain with the plaintiff in a trademark infringement case — including one where the defendant claims nominative fair use. As detailed below, we will devise a modified likelihood of confusion test to be employed in nominative fair use cases that takes into account the concerns expressed by the Court of Appeals for the Ninth Circuit. Then, we will determine the extent to which we would adopt the test for nominative fair use that the Ninth Circuit Court of Appeals established in light of our disagreement with that court’s view that nominative fair use is fundamentally different from classic fair use.

V. The Proper Analytical Approach for Nominative Fair Use Cases

A. Overview

Today we adopt a two-step approach in nominative fair use cases. The plaintiff must first prove that confusion is likely due to the defendant’s use of plaintiff’s mark. As we discuss more fully below, because our traditional likelihood of confusion test does not apply neatly to nominative fair use cases, we suggest eliminating those factors used to establish confusion in other trademark infringement cases that do not “fit” in the nominative use context. Once plaintiff has met its burden of proving that confusion is likely, the burden then shifts to defendant to show that its nominative use of plaintiff’s mark is nonetheless fair. To demonstrate fairness, the defendant must satisfy a three-pronged nominative fair use test, derived to a great extent from the one articulated by the Court of Appeals for the Ninth Circuit. Under our fairness test, a defendant must show: (1) that the use of plaintiff’s mark is necessary to describe both the plaintiff’s product or service and the defendant’s product or service; (2) that the defendant uses only so much of the plaintiff’s mark as is necessary to describe plaintiff’s product; and (3) that the defendant’s conduct or language reflect the true and accurate relationship between plaintiff and defendant’s products or services.

As an initial matter, we recognize that our concurring colleague rejects the bifurcated approach that we now adopt. He argues instead that the factors we consider under the fairness test should be incorporated into the likelihood of confusion analysis. In his view, our bifurcated approach places a heavy burden on the defendant to negate confusion and is judicially unmanageable. However, our approach does nothing of the kind.

We conclude that the broad based likelihood of confusion test our concurring colleague proposes is misplaced for several reasons. First, it is largely out of sync with the existing jurisprudence on fair use. The concurrence’s test allows no real possibility of the co-existence of fair use with some likelihood of confusion, yet this is precisely what the Supreme Court’s holding in KP Permanent Make-Up specifically contemplates. See KP Permanent Make-Up, 125 S.Ct. at 550 (“[S]ome possibility of consumer confusion must be compatible with fair use….”). In addition, our concurring colleague rejects the notion that nominative fair use could be used as an affirmative defense, viewing it instead as a confusion substitute. Yet, the Supreme Court clearly views fair use (albeit classic fair use) as an affirmative defense. Id. at 548-49 (referring to the affirmative defense of fair use). The concurrence fails to explain why KP Permanent Make-Up should neither control nor inform our analysis here, choosing instead to ignore the Court’s dictates in that case as they apply to nominative fair use.[3] After that decision, it seems to us that neither classic or nominative fair use should rise and fall based on a finding of likelihood of confusion. Classic fair use and nominative fair use are different in certain respects, but it is unclear to us why we should ask radically different questions when analyzing a defendant’s ability to refer to a plaintiff’s mark in the two contexts. As we have already stated, in both nominative and classic fair use cases the defendant uses the plaintiff’s mark descriptively in a way that potentially confuses consumers about the relationship between the plaintiff and the defendant’s product or services. In the classic fair use context, the defendant uses the mark to describe its own product, and in the nominative context, the defendant references plaintiff’s product in order to describe its own. The key first inquiry in both situations should be whether there is a likelihood of confusion.[4] The only other court to consider the application of nominative fair use doctrine since KP Permanent Make-Up has embraced this logic. Ty, Inc. v. Publ’ns Int’l, Ltd., No. 99 C 5565, 2005 WL 464688, at *6-8 (N.D.Ill. Feb.25, 2005).

Second, while the concurrence worries that a nominative use defendant will be overly burdened under our bifurcated approach, we believe that his approach is actually more burdensome to such a defendant. If the factors for determining fairness were incorporated into the likelihood of confusion test, a plaintiff’s showing of confusion might well overwhelm a defendant’s showing of fair use. This would essentially force a defendant asserting nominative fair use to negate all likelihood of confusion to succeed, a proposition that the Supreme Court rejected in KP Permanent Make-Up. Under our approach, the defendant has no duty to negate confusion as such, but rather must merely show that its use of the plaintiff’s mark is fair, a burden which, by contrast, is not cumbersome. Thus, it is our view that the bifurcated approach is ultimately less burdensome to a nominative use defendant than the analysis the concurrence proposes.

Finally, we believe that the bifurcated approach that we adopt today is more workable than a unified confusion/fairness test. We leave the now familiar test for likelihood of confusion largely intact and in the form in which district courts are accustomed to applying it. Our test for nominative fair use considers distinct factors that are readily susceptible to judicial inquiry. By contrast, the concurrence would incorporate several new considerations into the already lengthy ten-part test for confusion and ask district courts to balance a plaintiff’s showing of confusion against a defendant’s showing of fair use. Because confusion and fairness are separate and distinct concepts that can co-exist, blending them together into one test is, to our mind, a much less manageable approach.

B. The Proper Test For Likelihood of Confusion

As we have noted, and as the Ninth Circuit in New Kids on the Block also stated, the “likelihood of confusion” test does not lend itself nicely to a nominative fair use fact pattern. The traditional likelihood of confusion test has been set forth in our case law, specifically in Scott Paper Co. v. Scott’s Liquid Gold, Inc., 589 F.2d 1225 (3d Cir.1978) and Interpace Corp. v. Lapp, Inc., 721 F.2d 460, 463 (3d Cir.1983). It is a multi-factor test that assesses the following:

(1) degree of similarity between the owner’s mark and the alleged infringing mark;
(2) strength of the owner’s mark;
(3) price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase;
(4) length of time the defendant has used the mark without evidence of actual confusion;
(5) intent of the defendant in adopting the mark;
(6) evidence of actual confusion;
(7) whether the goods, though not competing, are marketed through the same channels of trade and advertised through the same media;
(8) the extent to which the targets of the parties’ sales efforts are the same;
(9) the relationship of the goods in the minds of consumers because of the similarity of function; and
(10) other facts suggesting that the consuming public might expect the prior owner to manufacture a product in the defendant’s market or that he is likely to expand into that market.
589 F.2d at 1229.

We have come to call these factors the “Lapp” factors. Although they are often helpful in determining whether a certain use of a mark is likely to confuse consumers, we have recognized that “the Lanham Act does not require that they be followed precisely so long as the relevant comparisons suggested by the test are made.” A & H Sportswear II, 237 F.3d 198, 207 (3d Cir.2000). In the context of a nominative use of a mark, such as the one we are presented with here, certain Lapp factors are either unworkable or not suited or helpful as indicators of confusion in this context. That is because, by definition, nominative use involves the use of another’s trademark in order to describe the trademark owner’s own product. Further, certain of the Lapp factors applied mechanically would inevitably point towards likelihood of confusion where no likelihood of confusion may actually exist. Thus, we must tailor the test and measure only those factors that are meaningful and probative in the context of nominative fair use.

Specifically, the first two Lapp factors would indicate a likelihood of confusion in a case such as this one simply because the mark is being employed in a nominative manner. By way of example, looking at the similarity of the mark would automatically lead to the conclusion that the use is likely to confuse simply because the mark is not merely similar, it is identical. The first Lapp factor does not leave any room for the consideration of the context of the use — i.e., that the mark is being used to describe the plaintiff’s own product. Therefore, it is not appropriate for analysis in a nominative use case.

Looking at the strength of CCE’s marks in this case, and in most nominative use cases, would also weigh in favor of a finding that the use is likely to confuse. However, defendants in nominative use cases, like LT in this case, feel they need to use the actual mark to describe the plaintiff’s product because of its very strength and what it has come to represent. In reality, in many such cases, the use of the name may be the only way for defendant to easily and precisely refer to plaintiff’s product in a way that will be understood by consumers. Accordingly, the marks’ strength is not really probative of confusion here, whereas it would be if defendant were passing off its goods under a similar mark.

We find that all of the other Lapp factors, while perhaps not appropriate for analysis in this particular case, could be analyzed in future nominative use cases, depending on the factual situation. For example, in this case, looking at whether the goods are marketed through the same channels or advertised in the same media would be a misleading indicator in determining the true likelihood of confusion. It would be of little relevance in this case whether LT marketed its services through the same channels as CCE because LT is not attempting to use CCE’s marks to refer to LT’s own services, but rather is using the marks to refer to CCE’s services. Therefore, it would almost be expected that LT and most other defendants in a nominative use case would market themselves through a media in which the marks to which they are referring would be easily recognized and have meaning or relevance, namely channels similar to those used by plaintiff. However, there may be certain situations, such as that encountered in New Kids (where newspapers used the New Kids’ mark to inquire of readers, for a price, their feelings about the group), where the channels of marketing are so dissimilar that evidence as to this factor could mitigate against a finding of likelihood of confusion. Similarly, in looking at whether the targets of the parties’ marketing efforts are the same, one could expect a fair use defendant to be reaching out to a group of consumers who are likely to recognize the mark. The fact that LT markets its services to real estate consumers makes sense. These are the very people who are likely to recognize and appreciate CCE’s marks and be able to evaluate the benefit that CCE’s purported association with LT could bring to them in the search for a home. This is not likely to confuse in the same way as would be the case if the defendant is passing off plaintiff’s goods as its own or using a similar mark to that of the plaintiff, but this factor should not be completely eliminated from a district court’s arsenal in evaluating likelihood of confusion in a nominative use case.

In the context of the facts of the case before us, we will turn our focus to those Lapp factors that appear most relevant in assessing likely confusion. These include:

(1) the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase;
(2) the length of time the defendant has used the mark without evidence of actual confusion;
(3) the intent of the defendant in adopting the mark; and
(4) the evidence of actual confusion.
In focusing on these factors, the court will be better able to assess whether consumers are likely to be confused by the use not because of its nominative nature, but rather because of the manner in which the mark is being nominatively employed. Of course the determination of which factors are relevant and probative in a given factual setting should be made in the first instance by the District Court, but here we provide our view on this issue as guidance, lest this matter again should come before a panel of this Court on the issue of whether the District Court examined the appropriate factors.

Viewed in the context of the present case, it is apparent why these factors are appropriate: they analyze the likelihood that a consumer will be confused as to the relationship or affiliation between LT and CCE, the heart of the nominative fair use situation. Focusing on the care and attention expected of consumers when making a purchase or using a service allows the court to understand the true risk that consumers may be confused merely because of their own inattention. Moreover, focusing on evidence of actual confusion (factors 2 and 4) will allow the District Court to truly understand whether this is merely a theoretical or hypothetical fear of the plaintiff or whether there is real danger that consumers are likely to be confused.

Lastly, analyzing the intent of the defendant in using the mark will allow the court to understand the defendant’s reason for utilizing the mark in the manner that it did. That is, if the court finds that the defendant made use of the plaintiff’s mark with the very purpose of causing consumers to think the plaintiff endorses or sponsors plaintiff’s good or service, then the likelihood that consumers will be confused as to endorsement/affiliation is greater. Whereas the traditional “intent” prong looks at intent to adopt a similar mark, here the key inquiry is whether the mark is being used so as to convey a connection between the parties that may not exist.

We hold today that the burden of proving likelihood of confusion, even in a nominative use case, should remain with the plaintiff and that these four factors are the essence of the inquiry, although others may prove useful in certain contexts. Here, the District Court did place this burden on CCE, but the District Court also analyzed certain Lapp factors that we have concluded are inappropriate in the context of this nominative fair use case. Further, its application of certain of the four relevant factors was somewhat imprecise.

We will remand for the District Court to inquire as to whether confusion was likely in light of our clarification of which factors are applicable in this case and the following discussion of the way in which the four relevant factors should be applied.[5]

Factor 1: The Price of the Goods and Other Factors Indicative of the Care and Attention Expected of Consumers When Making a Purchase

The District Court noted that “we are talking about real estate purchases which in reference to most consumers is the most important purchase both at the time and, in most people’s instances, in their entire lives.” (Trans. of Oral Opinion at JA735.) The District Court was correct to acknowledge and consider the importance of a real estate purchase to the average consumer. One would expect that consumers going to LT website’s “Find a Realtor” section would be very careful in learning about the various realtors in LT’s network. The degree of care weighs against a finding of likely confusion.

Factor 2: The Length of Time the Defendant has used the Mark without Evidence of Actual Confusion

The District Court conflated this factor with the defendant’s laches defense and, therefore, failed to weigh in LT’s favor the fact that, according to the record, CCE’s marks had been used for at least five years with little evidence of actual consumer confusion.

Factor 3: The Intent of the Defendant in Adopting the Mark

The District Court appears to have inferred improper intent on the part of LT from its very use of CCE’s marks. This finding was erroneous without further evidentiary support. The relevant question in this context is not whether the defendant intended to use the plaintiff’s mark, which it always has in a fair use case, but whether it used the mark with the intent to confuse the public as to the relationship between the defendant and the plaintiff. Use of the mark alone is not sufficiently probative of such intent.[6] The plaintiff must show that the defendant intended the public to believe that the plaintiff endorsed or somehow supported its products or services.

There are myriad factors to which a plaintiff may point to demonstrate that a defendant intended to confuse the public as to its relationship with the plaintiff. For example, courts in traditional trademark infringement cases have considered a defendant’s persistence in adopting a mark despite being warned of potential confusion, KOS Pharms., Inc. v. Andrx Corp., 369 F.3d 700, 721 (3d Cir.2004), and evidence of a defendant’s bad faith, A & H Sportswear II, 237 F.3d at 226. On remand, the District Court should inquire as to whether any analogous facts or evidence of record are present here. If so, the Court may weigh this in favor of a finding of likely confusion.[7]

Factor 4: The Evidence of Actual Confusion.

In this case, CCE presented no evidence of actual confusion before the District Court. In fact, as the District Court noted, from the Court’s first meeting with the parties, they had agreed that they would offer no evidence of actual confusion to the Court. Therefore, this factor should clearly weigh in favor of LT.

We will remand the case to the District Court for proper application of these factors. In determining whether confusion is likely, as we noted in A & H Sportswear II, 237 F.3d 198, 215 (3d Cir.2000), “different factors may be accorded different weights depending on the particular factual setting.” If CCE is able to meet its burden of showing likelihood of confusion on remand, the burden of proof will then shift to LT to show that the use is a nominative fair use.

C. The Affirmative Defense of Nominative Fair Use

Under the nominative fair use test adopted by the Court of Appeals for the Ninth Circuit, a defendant must prove: (1) that the product or service in question is one not readily identifiable without use of the trademark; (2) that only so much of the mark or marks is used as is reasonably necessary to identify the product or service; and (3) that the user did nothing that would, in conjunction with the mark, suggest sponsorship or endorsement by the trademark holder. New Kids on the Block, 971 F.2d at 308. In the Ninth Circuit, if these elements are proven, the use is “fair” and defendant will prevail. Further, this nominative fair use test, as discussed above, replaces the likelihood of confusion test in the Court of Appeals for the Ninth Circuit. We must decide the extent to which we should adopt this test as our own, mindful that we will employ it as an affirmative defense to be proven by defendant after likelihood of confusion has been demonstrated by the plaintiff.

We are tempted to use the three-pronged Ninth Circuit test outright, as it has withstood the test of time, has been tinkered with in no less than seven opinions.[8] Moreover, the Ninth Circuit Court of Appeals is the only Court of Appeals to have specifically crafted its own test for nominative fair use. Indeed, the Supreme Court decision in KP Permanent Make-Up involved a reversal of the most recent Ninth Circuit opinion on the issue of “fair use.” As prolific as the jurisprudence of the Ninth Circuit Court of Appeals has been in this area, the Supreme Court’s rejection of its view of how the fair use defense works in a classic fair use case emboldens us to examine the elements of the Ninth Circuit Court of Appeals’ nominative fair use test a bit more closely.

In so doing, we conclude that the test as written suffers from a lack of clarity. This is evident in the contortions that the Ninth Circuit Court of Appeals itself has gone through in applying it, the confusion that the District Court here encountered in its application, and in our conviction that a modified inquiry would aid in reaching the right result. We will adjust the test to include a slightly different set of considerations:

1. Is the use of plaintiff’s mark necessary to describe (1) plaintiff’s product or service and (2) defendant’s product or service?

2. Is only so much of the plaintiff’s mark used as is necessary to describe plaintiff’s products or services?

3. Does the defendant’s conduct or language reflect the true and accurate relationship between plaintiff and defendant’s products or services?

The following discussion explains how these factors should be applied.

1. First Prong

The first element of the Ninth Circuit Court of Appeals’ test involved an inquiry only into the necessity of using plaintiff’s trademark to describe plaintiff’s product. Here, the District Court instead — probably mistakenly — examined the necessity of the use of plaintiff’s trademark in order to describe defendant’s product. See Cairns v. Franklin Mint Co., 292 F.3d 1139, 1153 (9th Cir.2002). The first prong of New Kids On The Block is at best confusing and at worst incomplete. While it should be asked whether plaintiff’s product needs to be described by reference to its mark, should it not also be examined whether defendant’s use of it, at all, is necessary to accurately describe what defendant does or sells, or whether its reference to plaintiff’s mark is actually gratuitous? The District Court’s inquiry into the latter aspect was not called for under the New Kids On The Block test, but it actually seems entirely appropriate.

The focus on the necessity of the mark in order for defendant to describe plaintiff’s product makes sense in the context of nominative fair use, where the plaintiff’s mark is being used because it identifies the plaintiff’s product. We further note that the court need not find that the use of the mark is indispensable in order to find this factor fulfilled. For, as we have stated before, “[t]he Lanham Act does not compel a competitor to resort to second-best communication.” G.D. Searle & Co., 715 F.2d 837 at 842. Furthermore, as the Ninth Circuit has observed:

it is often virtually impossible to refer to a particular product for purposes of comparison, criticism, point of reference or any other such purpose without using the mark. For example, reference to a large automobile manufacturer based in Michigan would not differentiate among the Big Three; reference to a large Japanese manufacturer of home electronics would narrow the field to a dozen or more companies. Much useful social and commercial discourse would be all but impossible if speakers were under threat of an infringement lawsuit every time they made reference to a person, company or product by using its trademark.
New Kids on the Block, 971 F.2d 302 at 307. Therefore, the court need only be satisfied that the identification by the defendant of plaintiff’s product or service would be rendered significantly more difficult without use of the mark.

Additionally, we believe that it is important for a court to understand how necessary the use of the mark is to the identification of defendant’s product. That is, the more dependent the ready identification of defendant’s product is on the description of plaintiff’s product through the employment of plaintiff’s mark, the more likely it is that the use is a fair one.[9]

In this case, the District Court looked at this test only in regard to defendant’s product. On remand the Court should additionally focus on whether requiring LT to describe plaintiff CCE’s services without using its mark is a forced reversion to second-best communications. Considerations such as the simplicity of description and the likelihood that consumers will understand a given reference to plaintiff’s services without use of the mark are appropriate to this analysis. See Cairns, 292 F.3d at 1153.

2. Second Prong

Here again, the New Kids On The Block test focuses on the “amount” of plaintiff’s mark that is used, asking whether only so much is used as is necessary to describe plaintiff’s product. That focus on limiting the quantum use of the attributes of plaintiff’s marks’ is appropriate, for it is the use of plaintiff’s marks that concerns us. Yet, the case law of the Ninth Circuit Court of Appeals has expanded on this prong to inquire, at this stage, about the defendant’s need to use of the mark, stating:

[W]hat is `reasonably necessary to identify the plaintiff’s product differs from case to case…. Where, as in the present case, the description of the defendant’s product depends on the description of the plaintiff’s product, more use of the plaintiff’s trademark is `reasonably necessary to identify the plaintiff’s product’ than in cases where the description of the defendant’s product does not depend on the description of the plaintiff’s product.

Cairns, 292 F.3d at 1154. Since we have positioned the assessment of the defendant’s need to use plaintiff’s mark as part of the first inquiry, we eliminate the confusion inherent in inquiring into “need” at the second step. Thus, under our approach, the second prong tests only whether the quantum of the plaintiff’s mark used by the defendant was appropriate.

In analyzing this factor, the District Court essentially predetermined the outcome of the second prong by its finding as to the first prong. The District Court found that because the use of CCE’s marks was not necessary to identify LT’s services as a whole under the first prong, the use could not possibly be only so much as was necessary. But the proper focus under this prong is on whether only so much of plaintiff’s mark as is reasonably necessary to identify plaintiff’s product or service has been used by defendant. Consideration should be given at this stage to the manner in which the mark is portrayed. For example, did the defendant use plaintiff’s distinctive lettering when using the plaintiff’s mark or did the defendant, as in this case, simply use block letters to spell out plaintiffs’ names?

In his concurring opinion, Judge Fisher posits that both Prongs One and Two of our modified nominative fair use test are, and should be, subsumed in the likelihood of confusion analysis. He suggests that the necessity of using the mark (Prong 1) and the reasonableness of that use (Prong 2) are just another way of examining the defendant’s intent in using the mark. This is a mischaracterization of both the intent factor under Lapp, as described above, and the first two prongs of the nominative use defense we espouse. Even if it is found that a defendant’s intent is improper and its use of a mark could cause confusion, the defendant could conceivably prove that even its potentially confusing use is fair because it is reasonably limited and necessary under the first two prongs of our nominative fair use test. These contours of fairness are not recognized by our concurring colleague, but they are fundamental to the concept of fair use according to the Supreme Court’s recent decision in KP Permanent Make-Up.

3. Third Prong

The New Kids test at this stage asks whether the user did anything that would, in conjunction with the mark, suggest sponsorship or endorsement by the trademark holder. However, we believe the appropriate question should be a bit broader: does the defendant’s conduct or language reflect the true and accurate relationship between plaintiff and defendant’s products or services? We believe that the Ninth Circuit Court of Appeals’ focus on whether the user’s conduct implies endorsement may not truly reflect whether the use is fair because sometimes the plaintiff’s relationship with defendant may be one of endorsement, but the nature of the endorsement as reflected by defendant’s employment of plaintiff’s marks may not be accurate. A defendant’s purposeful portrayal of plaintiff’s endorsement of its product through defendant’s conduct or language does not necessarily render the use unfair, as long as the depiction of the endorsement is accurate. In addition, our version suggests that we can consider the defendant’s failure to state or explain some aspect of the relationship, whereas the New Kids version focuses on affirmative acts, i.e., what the defendant did to suggest sponsorship.

In this case, the District Court concluded that an ordinary consumer would look at LT’s website and conclude that there was some affiliation between LT and the enumerated companies. Further, the Court inferred from the mere presence of the marks an intent on the part of LT to convey endorsement or affiliation. On remand, the District Court should specifically determine in analyzing this prong whether the portrayal of the relationship was accurate, and what more the defendant could have done to prevent an improper inference regarding the relationship. The mere presence or use of the mark does not suggest unfairness under this prong. Here, LT added a disclaimer, the significance of which the District Court downplayed, stating that “LendingTree is not sponsored by or affiliated with the parent franchisor companies of any of the participating members of its network.” (Text from www.LendingTree.com at JA687.) Far from unimportant, such a disclaimer must be considered in determining whether the alleged infringer accurately portrayed the relationship that existed between plaintiff and defendant. See Playboy Enters., 279 F.3d at 803 (finding that the fact that defendant’s website included a clear disclaimer of any connection to plaintiff satisfied this prong of the Ninth Circuit Court of Appeals’ test for nominative fair use). Therefore, the District Court on remand should look at the precise way in which what the defendant said, or did not say, other than the mere presence of the mark on the website, may have inaccurately implied endorsement or sponsorship by CCE. The District Court should consider whether the disclaimer was an affirmative action by LT that effectively negated an inaccurate implication of sponsorship or endorsement by CCE.

The Ninth Circuit’s test for nominative fair use does not explicitly include accuracy within the analysis, but the Supreme Court has recognized that “[a]ccuracy of course has to be a consideration in assessing fair use.” KP Permanent Make-Up, Inc., 125 S.Ct. at 551. In examining the conduct of the defendant to determine whether the defendant has done anything to affirmatively cause consumer confusion, it is only reasonable to consider the precision with which the defendant has described its relationship with plaintiff. In this case, the evaluation of accuracy would necessarily include consideration of LT’s characterization of the nature and extent of its relationship with CCE and its agents. This would include the District Court’s consideration of LT’s reference to its affiliation with CCE brokers in general, rather than referencing each more accurately by their “d/b/a/” title and whether this rendered the use inaccurate or was somehow misleading as to any endorsement or relationship.

V. Conclusion

In sum, we hold today that the Lapp test for likelihood of confusion still has an important place in a trademark infringement case in which the defendant asserts the nominative fair use defense. In this case, the test should focus on the four relevant factors: (1) the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase; (2) the length of time the defendant has used the mark without evidence of actual confusion; (3) the intent of the defendant in adopting the mark; and (4) the evidence of actual confusion.

Once plaintiff has met its burden of proving that confusion is likely, the burden then shifts to defendant to show that its nominative use of plaintiff’s marks is nonetheless fair. In this Circuit, we have today adopted a test for nominative fair use in which a court will pose three questions: (1) Is the use of the plaintiff’s mark necessary to describe both plaintiff’s product or service and defendant’s product or service? (2) Is only so much of the plaintiff’s mark used as is necessary to describe plaintiff’s products or services? (3) Does the defendant’s conduct or language reflect the true and accurate relationship between plaintiff and defendant’s products or services? If each of these questions can be answered in the affirmative, the use will be considered a fair one, regardless of whether likelihood of confusion exists.

We adopt a bifurcated approach that tests for confusion and fairness in separate inquiries in order to distribute the burden of proof appropriately between the parties at each stage of the analysis. The defendant has no burden to show fairness until the plaintiff first shows confusion. Furthermore, by properly treating nominative fair use as an affirmative defense, our approach allows for the possibility that a district court could find a certain level of confusion, but still ultimately determine the use to be fair. By contrast, a unified likelihood of confusion test would require a defendant to negate likelihood of confusion by undercutting the Lapp factors. Because the Supreme Court explicitly rejected such a proposition in KP Permanent Make-Up, we decline to adopt it.

We will therefore REVERSE the order of the District Court and REMAND for further proceedings consistent with this opinion.

FISHER, Circuit Judge.

As the majority correctly concludes, proper nominative use is permissible use, leaving the critical question of how to frame the nominative use analysis. I concur with the majority’s “firm conviction that the burden of proving likelihood of confusion should remain with the plaintiff in a trademark infringement case.” Maj. Op. at 221. I also concur in its conclusion that this Court’s Lapp factors,[10] where relevant, must be used in the analysis. I therefore join in the majority’s judgment remanding the case for further proceedings.

I depart from majority, however, in its adoption of a bifurcated analysis that looks first to a truncated Lapp-factor test, followed by an affirmative defense of “nominative fair use.” Despite professing a “firm conviction” that the burden of proving likely confusion remains on plaintiffs, the majority formulates an affirmative defense that shifts to defendant the burden of negating confusion. In so doing, the majority flouts binding caselaw holding that proper nominative use is use that is not likely to confuse, and that a plaintiff alone bears the burden of establishing likely confusion. Moreover, to the extent the majority places any burden on plaintiffs at all, it is so watered-down that plaintiffs might prove likely confusion on one Lapp factor alone. The majority’s bifurcated test is also judicially unmanageable because it requires courts to address identical factors on both sides of the equation.

Accordingly, I dissent. I first review nominative use cases and conclude that precedent, binding and otherwise, shows that the question of nominative use is a question of likely confusion and not an affirmative defense. Second, I address the majority’s bifurcated analysis, and conclude that it runs afoul of the Lanham Act and binding precedent because it places on defendant the burden of negating likely confusion. I also conclude that the test is judicially unmanageable because it requires courts to examine identical likelihood of confusion factors on both sides of the analysis. Finally, I propose that in nominative use cases, courts apply a modified likelihood of confusion inquiry based on Lapp factors two through ten, with the burden resting where it must, on plaintiffs. Because proper nominative use is use that is not likely to confuse, it is the majority’s “nominative fair use” test that is unfair, because it places the burden where it should never be, on the defendant.

I.

Despite the majority’s desire to “clarify the proper analysis,” Maj. Op. at 219, its approach significantly increases the uncertainty and — to use a term appropriate to the present context — confusion in this important area of the law. In light of this uncertainty, and in light of the Supreme Court’s holding in KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111, 125 S.Ct. 542, 160 L.Ed.2d 440 (2004), that a defendant bears no burden to negate confusion, it is vital to examine the nature of nominative use to determine whether it is an affirmative defense, or instead an established way to test for likelihood of confusion. Because it is the latter, the burden lies with plaintiffs and a bifurcated analysis is improper.

The majority, stating that the Ninth Circuit is the only Court of Appeals to craft a test for nominative use, mildly rewords that test and recasts it as an affirmative defense. As discussed in the subsections below, this approach is unprecedented, unwise, and impermissible. First, it is unprecedented. The Ninth Circuit’s decisions make it clear that the test is nothing more than a likelihood of confusion substitute, making it an inappropriate foundation for an affirmative defense. Second, it is unwise. Other Courts of Appeals have exercised great skepticism towards the Ninth Circuit’s test and none have adopted it outright, calling into serious question the wisdom of using it as a analytic model. Finally, the majority’s approach is impermissible. Binding precedent from the Supreme Court and this Court requires that nominative use be addressed through a likelihood of confusion analysis and that the burden remain on plaintiffs.

A.

The erroneous major premise underlying the majority’s conclusion is the assumption that nominative use must be analyzed through an affirmative defense. This assumption begins with the majority’s misuse of the test enunciated by the Ninth Circuit in New Kids on the Block v. News America Publishing, Inc., 971 F.2d 302 (9th Cir.1992). As stated by the majority: “We must decide the extent to which we should adopt [the Ninth Circuit’s] test as our own, mindful that we will employ it as an affirmative defense to be proven by defendant after likelihood of confusion has been demonstrated by the plaintiff.” Maj. Op. at 228 (emphasis added). However, the Ninth Circuit has made it crystal-clear that its test is a substitute for its normal multi-factor likelihood of confusion test, to be used in cases involving purported nominative use.[11] The Ninth Circuit has also made it clear that this test places on the defendant the burden of proving no likelihood of confusion. Because KP Permanent makes it clear that it is improper to place a burden of proving nonconfusion on the defendant, the majority grievously errs by treating the Ninth Circuit test as an affirmative defense.

As noted by New Kids, classic statutory fair use applies “where the defendant has used the plaintiff’s mark to describe the defendant’s own product.” 971 F.2d at 308 (emphasis in original). Thus, if BEST PRICE were used as a mark, third parties might still fairly use “best price” in its original descriptive sense to refer to their own goods or services, such as “We offer the best price on breakfast cereal.” In contrast, nominative use arises “where the defendant uses a trademark to describe the plaintiff’s product, rather than its own.” New Kids, 971 F.2d at 308. In New Kids, one of the defendants, a newspaper, had printed a poll saying “New Kids on the Block are pop’s hottest group. Which of the five is your fave? Or are they a turn off?” Id. at 304. Another defendant newspaper, under a picture of one of the New Kids, asked, “Which of the New Kids on the Block would you most like to move next door?” Id. The Court held:

a commercial user is entitled to a nominative fair use defense provided he meets the following three requirements: First, the product or service in question must be one not readily identifiable without use of the trademark; second, only so much of the mark or marks may be used as is reasonably necessary to identify the product or service; and third, the user must do nothing that would, in conjunction with the mark, suggest sponsorship or endorsement by the trademark holder.
Id. at 308 (note omitted). The first requirement was met because “[i]t is no more reasonably possible, however, to refer to the New Kids as an entity than it is to refer to the Chicago Bulls, Volkswagens or the Boston Marathon without using the trademark.” Id. The second requirement was met because defendants “do not use the New Kids’ distinctive logo or anything else that isn’t needed to make the announcements intelligible to readers.” Id. The third element was met as well because “nothing in the announcements suggests joint sponsorship or endorsement by the New Kids.” Id. In particular, one of the ads asked “whether the New Kids might be `a turn off.'” Id. at 309.

In subsequent cases, the Ninth Circuit has emphasized that the test is a substitute for the normal multifactor likelihood of confusion analysis — “nominative fair use analysis … replaces the likelihood of customer confusion analysis.” Cairns v. Franklin Mint Co., 292 F.3d 1139, 1151 (9th Cir.2002) (emphasis in original); see also KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 328 F.3d 1061, 1072 (9th Cir.2003) (“When analyzing nominative fair use, it is not necessary to address likelihood of confusion because the nominative fair use analysis replaces the likelihood of confusion analysis.” (emphasis added)), rev’d on other gds., 543 U.S. 111, 125 S.Ct. 542, 160 L.Ed.2d 440 (2004).

Fueling the majority’s mistaken use of the Ninth Circuit’s test as an affirmative defense is the fact that the Ninth Circuit confusingly refers to its own test at times as a defense, such as in New Kids where the Court called the test “a nominative fair use defense.” 971 F.2d at 308. But it is clear that the Ninth Circuit’s test is not an affirmative defense in the sense the majority envisions, as a defense to be asserted once a plaintiff has established a likelihood of confusion. This is because the Ninth Circuit’s so-called “defense” is itself a likelihood of confusion test. In Brother Records, Inc. v. Jardine, the Court acknowledged charges of “`confusion in the case law'” as to whether “nominative fair use” was really a distinct defense or instead “`merely one type of use which is not likely to cause confusion.'” 318 F.3d 900, 908-09 n. 5 (9th Cir.) (quoting 2 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition § 11:47 (4th ed.2001)), cert. denied, 540 U.S. 824, 124 S.Ct. 155, 157 L.Ed.2d 45 (2003). In response, the Court admitted that its test replaced the normal multi-factor likelihood-of-confusion analysis, with one significant difference: it “shifts to the defendant the burden of proving no likelihood of confusion.” Id. at 909 n. 5.

Thus, it is clear that the Ninth Circuit treats its “defense” as a substitute for testing likelihood of confusion. It is also clear that it places the burden of negating confusion on the defendant, which under KP Permanent, is now impermissible. In KP Permanent, the Court held that a defendant at no time possesses the burden of negating confusion, noting, inter alia, “the incoherence of placing a burden to show nonconfusion on a defendant,” 125 S.Ct. at 549, and that “it takes a long stretch to claim that a defense of fair use entails any burden to negate confusion.” Id. at 548. As the Supreme Court has made it clear that the burden of showing likely confusion belongs to plaintiffs alone, it is improper to twist the nominative use inquiry into an affirmative defense.

B.

The majority suggests that refusing to adopt its approach would be “largely out of sync with the existing jurisprudence on fair use.” Maj. Op. at 222. But considering the majority’s concessions that the Ninth Circuit “is the only Court of Appeals to have specifically crafted its own test for nominative fair use,” id. at 228, that “[f]ew other courts have spoken on the precise issue,” id. at 218, and that there is a “paucity of the case law on this subject from every [other] court” on nominative use, id. at 219, the majority cannot point to a set of existing jurisprudence from which to stray.

If anything, to the extent there is discernable jurisprudence regarding the Ninth Circuit’s approach, the trend is heavily against adopting its test. The Sixth Circuit, faced with the option of adopting the Ninth Circuit’s test, held “[t]his circuit has never followed the nominative fair use analysis, always having applied [our own multi-factor likelihood of confusion] test. We are not inclined to adopt the Ninth Circuit’s analysis here.” PACCAR Inc. v. TeleScan Technologies, L.L.C., 319 F.3d 243, 256 (6th Cir.2003). The Fifth Circuit applied the test only in part, declining to impose the first element of the test. See Pebble Beach Co. v. Tour 18 I Ltd., 155 F.3d 526, 546 n. 13 (5th Cir.1998) (refusing to adopt the first element because it would always be satisfied in disputes like the comparative advertising case before the Court, and declining to opine on whether the requirement should be imposed outside the comparative advertising context). Pebble Beach expressly recognized the Ninth Circuit’s “defense” for what it was, a likelihood of confusion test, stating that “[w]hile a claim that the use was to identify the markholder’s goods or services is analogous to the statutory fair-use defense, it is in actuality a claim that the use is noninfringing and thus creates no likelihood of confusion.” Id. at 545 (emphasis added).[12]

Other authorities have noted the deficiencies of the Ninth Circuit’s approach. A District Court in Maryland has criticized all three elements, noting that the first and third prongs are nothing more than “a restatement of two basic principles of trademark law,” first, that use of a mark “is not prohibited if the use is intended merely to refer to the holder of the mark,” and second, that such reference “is permissible provided it is not likely to cause confusion.” Nat’l Federation of Blind, Inc. v. Loompanics Enterp., Inc., 936 F.Supp. 1232, 1241 (D.Md.1996). Similarly, the second prong “appear[ed] to derive from a concern that confusion as to affiliation may result if the defendant’s use of the plaintiff’s mark exceeds its legitimate referential purpose.” Id. The test has been additionally and properly criticized as “cloud[ing] the issue of who carries the burden to prove likelihood of confusion.” Derek J. Westberg, Note, New Kids on the Block v. News America Publishing, Inc.: New Nominative Use Defense Increases the Likelihood of Confusion Surrounding the Fair Use Defense to Trademark Infringement, 24 Golden Gate U.L.Rev. 685, 709 (1994). The author notes that the third element requires “that the user must do nothing that would, in conjunction with the mark, suggest sponsorship or endorsement of the trademark holder,” but wryly notes that this inquiry “must be satisfied as part of the plaintiff’s prima facie case of infringement.” Id. at 705. I agree, and more importantly, so does the Supreme Court.[13]

C.

The irony of the majority’s well-intended but erroneous “nominative fair use” defense is that the majority felt the need to craft a defense in the first place. Binding caselaw from the Supreme Court and this Court have analyzed nominative use cases without resorting to a separate test and without placing the burden of negating confusion on defendants. I do not see how the majority’s approach can be reconciled with these cases.

In Prestonettes, Inc. v. Coty, the Supreme Court addressed a case involving what we would today refer to as nominative use. 264 U.S. 359, 44 S.Ct. 350, 68 L.Ed. 731 (1924). The plaintiff, owner of the marks COTY and L’ORIGAN, sold perfume and powder. The defendant bought genuine COTY and L’ORIGAN products and repackaged them. The plaintiff sought an injunction to prevent the defendant from any use of the plaintiff’s marks on the repackaged goods. The District Court had permitted the defendant to sell the plaintiff’s repackaged goods, with a disclaimer saying that defendant was not connected with plaintiff and that the perfume and powder were respectively “independently rebottled” and “independently compounded,” with every word in “letters of the same size, color, type, and general distinctiveness.” Id. at 367, 44 S.Ct. 350. Writing for the Court, Justice Holmes held that “[w]hen the mark is used in a way that does not deceive the public we see no such sanctity in the word as to prevent its being used to tell the truth. It is not taboo.” Id. at 368, 44 S.Ct. 350. Prestonettes is clearly a nominative use case, and the Supreme Court did not treat the use as one that had to be justified by the defendant.

The majority seems to dismiss Prestonettes on the basis that it is not a new case. See Maj. Op. at 223 n.4. I am not aware of any rule permitting us to disregard Supreme Court precedent simply because it is not as fresh as the Sunday paper. The majority fails to persuade that Prestonettes is anything other than a paradigm nominative use case. Indeed, even the Ninth Circuit, when formulating its nominative use test in New Kids, cited, quoted, and relied on Prestonettes when defining the nature of nominative use:

[W]e may generalize a class of cases where the use of the trademark does not attempt to capitalize on consumer confusion or to appropriate the cachet of one product for a different one. Such nominative use of a mark — where the only word reasonably available to describe a particular thing is pressed into service — lies outside the strictures of trademark law: Because it does not implicate the source-identification function that is the purpose of trademark, it does not constitute unfair competition; such use is fair because it does not imply sponsorship or endorsement by the trademark holder. “When the mark is used in a way that does not deceive the public we see no such sanctity in the word as to prevent its being used to tell the truth.” Prestonettes, Inc. v. Coty, 264 U.S. 359, 368, 44 S.Ct. 350, 351, 68 L.Ed. 731 (1924) (Holmes, J.).
971 F.2d at 307-08 (emphasis in original).

In any case, we need not turn to the Ninth Circuit, nor rely solely on older precedent, to find more recent binding nominative use cases. This Court faced a nominative use dispute in G.D. Searle & Co. v. Hudson Pharmaceutical Corp., 715 F.2d 837 (3d Cir.1983). In that case, the defendant used the plaintiff’s METAMUCIL mark to truthfully state that its product was “Equivalent to METAMUCIL.” Id. at 838. The District Court enjoined the defendant from making conspicuous use of the mark but otherwise permitted truthful use of the mark. On appeal, we affirmed that defendant could truthfully use the mark. Relying on Prestonettes, we concluded that “whether one is entitled to refer to a competitor’s trademark depends not on where the reference appears, but on whether the reference is truthful.” Id. at 843 (discussing and quoting Prestonettes, 264 U.S. at 368, 44 S.Ct. 350). We further held that such use was permissible “so long as it does not contain misrepresentations or create a reasonable likelihood that purchasers will be confused as to the source, identity, or sponsorship of the advertiser’s product.” Id. at 841 (quotes omitted, emphasis added).[14]

Three things are obvious in Searle. First, we had no doubt that Prestonettes was the controlling case for the disputed use, a use that modern parlance deems “nominative use.” See id. at 843 & n. 13 (quoting Prestonettes, 264 U.S. at 368, 44 S.Ct. 350). Second, we treated the question before us solely as whether the use was likely to confuse. “`[C]ollateral and truthful references to the trademark of another are permissible as long as the “unauthorized” reference does not cause confusion as to the source of the product advertised.'” Id. at 841 n. 9 (quoting Delaware and Hudson Canal Co. v. Clark, 80 U.S. (13 Wall) 311, 327, 20 L.Ed. 581 (1872)) (emphasis added). Third, notably missing from Searle was any suggestion that the defendant bore any burden of satisfying an affirmative defense. Instead, the question was whether the conduct would “create a reasonable likelihood that purchasers will be confused as to the source, identity, or sponsorship of the advertiser’s product.” Id. at 841 (emphasis added, internal quotes omitted).[15]

Like Prestonettes, Searle is a paradigm nominative use case. The mere fact that these decisions did not use the term “nominative use” when they were decided does not make them any less binding. Both cases make it clear that nominative use is nothing more than a likelihood of confusion inquiry, and neither places upon a defendant the burden of negating confusion or of putting forth an affirmative defense.[16]

II.

This section examines the inner workings of the majority’s bifurcated test. Subsection II.A shows that the majority’s approach permits plaintiffs to establish likely confusion on as little as on Lapp factor, and improperly shifts to defendant the burden of negating confusion. Subsection II.B shows that the majority’s approach is judicially unmanageable because it requires courts to consider identical likelihood of confusion factors on both sides of the analysis. Subsection II.C shows that the majority’s approach stems from a mistaken conflation of the statutory defense of descriptive fair use, which asks whether a defendant has used a mark in its primary descriptive sense, with nominative use, which simply asks if a defendant has created a likelihood of confusion.

A.

The majority’s approach whittles down this Court’s ten-factor Lapp test beyond recognition, permitting plaintiffs to show likely confusion on as little as one factor. The effect of this is to shift to defendant the burden of negating confusion. I will start by outlining the majority’s proposed order of analysis. Although the majority concedes that eight of the ten Lapp factors can be relevant in nominative use cases, it limits extended discussion to these four factors:

1. the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase;
2. the length of time the defendant has used the mark without evidence of actual confusion;
3. the intent of the defendant in adopting the mark;[17] and
4. the evidence of actual confusion.
Maj. Op. at 225-26.[18]

If likely confusion is shown by plaintiffs — which as discussed below, might be shown on one Lapp factor alone — then the majority would require defendant to satisfy its “nominative fair use” defense. As mildly reworded from the Ninth Circuit’s test, the majority would require defendant to show:

1. Is the use of plaintiff’s mark necessary to describe (1) plaintiff’s product or service and (2) defendant’s product or service?
2. Is only so much of the plaintiff’s mark used as is necessary to describe plaintiff’s products or services?
3. Does the defendant’s conduct or language reflect the true and accurate relationship between plaintiff and defendant’s products or services?
Maj. Op. at 228.

I will first consider the majority’s bare-bones Lapp test. Regarding the first factor — the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase — the District Court had little to say, noting that the issue was “a little difficult” and concluding that real estate purchases are the most important purchases at the time and perhaps in people’s lives. Although the District Court’s opinion is not clear, it appears that the Court either found that the factor was neutral or pointed somewhat towards defendant.

The second and fourth factors — looking to the absence or presence of actual confusion — do not assist plaintiffs in this case because they stipulated that they would not rely on evidence of actual confusion for these proceedings. This is unsurprising because actual confusion evidence is in many cases unavailable and is not required to find likely confusion.[19]

This leaves only one factor, intent. In this regard, the majority holds that “if the court finds that the defendant made use of the plaintiff’s mark with the very purpose of causing consumers to think the plaintiff endorses or sponsors plaintiff’s good or service, then the likelihood that consumers will be confused as to endorsement/affiliation is greater.” Maj. Op. at 226. Accordingly, holds the majority, “the key inquiry is whether the mark is being used so as to convey a connection between the parties that may not exist.” Ibid. I agree with the majority that the intent inquiry is important in nominative use cases. Indeed, if the plaintiffs in this appeal are to show likely confusion at all, they will have to prove it through this factor. And considering the parties’ serious dispute over the veracity and context of the defendant’s use of the plaintiffs’ marks, I have no doubt that on remand that the intent factor will be the locus of contention.

But the majority is unclear as to what factors the District Court might look in proving intent. It correctly notes that mere “[u]se of the mark alone is not sufficiently probative of such intent.” Maj. Op. at 227. This makes sense because the mere existence of nominative use, without more, lacks probative value as to whether the use was made with the intent to confuse. But the majority provides little guidance on what might serve as affirmative proof of intent in a nominative use case, noting unhelpfully that the District Court should consider whether defendant “used the mark with the intent to confuse the public as to the relationship between the defendant and the plaintiff.” Ibid. (emphasis in original). But that admonition merely restates the original inquiry, i.e., what evidence proves bad intent?

The majority’s terse discussion of the intent factor begs clarification of the kind of proof that would show intent to confuse in a nominative use case. Fortunately, such clarification is not difficult to find in the majority’s opinion, although its location is troubling. One might look to the purpose of a defendant’s use, which would be probative of whether the use was made in good faith or instead with the intent to confuse. One might look to the prominence of the use; if defendant used the mark in too-large or stylized form, that may show an intent to use more than is necessary, and again show an intent to confuse. One might also look to the truthfulness of the use: if the use was truthful, that would suggest a lack of intent to confuse; conversely, misleading or untruthful uses of plaintiffs’ marks may be probative of an intent to confuse.

If the above characterization sounds familiar, it should. Each and every one of the majority’s “nominative fair use” prongs is nothing more than an inquiry into the likelihood of confusion, specifically, whether the use is with the intent to confuse due to a presence or lack of good-faith purpose, prominence, and truthfulness. The majority’s first “nominative fair use” prong looks to purpose and asks: “Is the use of plaintiff’s mark necessary to describe (1) plaintiff’s product or service and (2) defendant’s product or service?” This prong asks why defendant is using the mark, i.e., its purpose in using it. By the majority’s formulation, if it is necessary for defendant to use the mark to describe both the plaintiffs’ and its own services, then such use may be fair.[20] But the reason for any “fairness” would be because the use was made with good intent, a paradigm likelihood of confusion inquiry, and one that falls within the normal Lapp analysis.

The second inquiry suffers the same infirmity. It asks, “Is only so much of the plaintiff’s mark used as is necessary to describe plaintiff’s products or services?” This inquiry looks to the prominence of the use, asking if the defendant “took too much,” i.e., whether it used bigger, bolder type or a logo when plain type might have sufficed. And why is this relevant? The answer is clear: the willful overuse of plaintiffs’ marks may evidence bad-faith intent and is more likely to confuse. Indeed, Searle itself affirmed an injunction that enjoined the defendant from setting the mark apart in larger or differently colored type, and required it to use “®” and a disclaimer. See 715 F.2d at 839, 843. As such, the majority’s prominence inquiry again is nothing more than part of the Lapp analysis.[21]

Like the first two prongs, the majority’s third prong is again nothing more than a likelihood of confusion inquiry. It asks, “Does the defendant’s conduct or language reflect the true and accurate relationship between plaintiff and defendant’s products or services?” By its own terms, the third prong looks to the truthfulness of the use of the mark and the defendant’s conduct in describing the relationship between the parties and their goods/services. But this truthfulness inquiry is identical to the statutory test of infringement under Lanham Act § 32(1)(a), which asks whether the use is “likely to cause confusion, or to cause mistake, or to deceive.” Lanham Act § 32(1)(a), 15 U.S.C. § 1114(1)(a) (emphasis added). Similarly, Section 43(a)(1)(A) asks whether the use is “likely to cause confusion, or to cause mistake, or to deceive as to the affiliation, connection, or association of such person with another person.” Id. § 43(a)(1)(A), 15 U.S.C. § 1125(a)(1)(A) (emphasis added). By its plain language, the Lanham Act looks to the truthfulness of the parties’ affiliation as crucial parts of the infringement analysis. In other words, truthfulness is a part of the likelihood of confusion analysis, and the majority’s formulation to the contrary is irreconcilable with the statute.

Moreover, the majority’s third fair use prong is substantively identical to its own “intent” Lapp factor. The majority’s third fair use prong asks whether “the defendant’s conduct or language reflect[s] the true and accurate relationship between plaintiff and defendant’s products or services.” Maj. Op. at 228 (emphasis added). This inquiry is identical to the majority’s intent test under Lapp, which states “the key inquiry is whether the mark is being used so as to convey a connection between the parties that may not exist.” Id. at 226 (emphasis added). Whether one calls uses the words “connection” or “relationship,” it is obvious that the majority’s third fair use prong and its Lapp intent inquiry are identical.[22] As the majority places identical likelihood of confusion factors on both sides of the equation, there can be no doubt that the majority expects defendant to prove nonconfusion.

In sum, the “nominative fair use” defense is nothing more than a venture into Lapp territory, namely, intent as shown through the defendant’s purpose, prominence, and truthfulness in connection with plaintiffs’ marks.[23] Although it is troubling that the majority would permit plaintiffs to establish likely confusion based on one Lapp factor alone, it is even more disturbing that the majority would require the defendant to dispel that finding through an affirmative defense looking to the same likelihood of confusion considerations.[24]

B.

Not only does the majority’s test run afoul of KP Permanent and the Lanham Act by placing upon defendant a burden of negating confusion, but it is also judicially unmanageable. On remand, the District Court will be required to consider the purpose, prominence, and truthfulness of the defendant’s conduct as part of the plaintiffs’ likelihood of confusion case. It will also be required to look to the same factors when entertaining the defendant’s affirmative defense. This makes no sense. The majority points to no meaningful distinctions between the kinds of “intent” that might be relevant to likelihood of confusion, and those which might be relevant to the affirmative defense. This is because they are identical. I respectfully submit that District Courts will not be able to make heads or tails of how to conduct the analysis.

The practical effect of the majority’s bifurcated analysis is that the nominative fair use defense will almost always be available only in cases where it is unneeded. Where a court finds the use to be made with the intent to confuse, then a finding of likelihood of confusion will often follow. According to the majority, in such a case, a defendant could then assert nominative fair use. However, such a defense would be hollow indeed, because the entire “nominative fair use” defense asks whether the use was made with the intent to confuse. Thus, a defendant found to have caused likely confusion due to illegitimate purpose, overly prominent use, or untruthfulness will be unable to assert a successful “nominative fair use” defense. Conversely, in cases where the alleged nominative use is made with good-faith purpose, is not overly prominent, and is truthful, then a plaintiff will likely fail in proving its case of likely confusion. In that case, the defense will be unneeded.[25]

The result will certainly be that the nominative fair use defense, as formulated by the majority, will remain unavailable in many of the cases where it is needed due to a bad-faith intent to confuse. The Supreme Court frowned heavily on this absurdity in KP Permanent where it held, “`[I]t defies logic to argue that a defense may not be asserted in the only situation where it even becomes relevant.'” 125 S.Ct. at 549 (quoting Shakespeare Co. v. Silstar Corp., 110 F.3d 234, 243 (4th Cir. 1997)). The Court further noted that “it would make no sense to give the defendant a defense of showing affirmatively that the plaintiff cannot succeed in proving some element (like confusion).” Id. Yet this is exactly what the majority has done.

C.

The majority’s bifurcated analysis appears to arise from a failure to recognize that KP Permanent addresses several issues, only the first of which is applicable to this case, namely, the Supreme Court’s holding that a defendant bears no burden of negating confusion. KP Permanent’s second holding concerns matters peculiar to the statutory descriptive fair use test, a test with no bearing in the present context.

The first issue in KP Permanent, and the one that controls the outcome of this case, regards which party bears the burden of demonstrating a likelihood of confusion. As the majority recognizes, KP Permanent clearly holds that this burden belongs to a plaintiff alone. Nothing in the Lanham Act requires a defendant to negate a likelihood of confusion.[26] As held by the Supreme Court, the Act requires that the burden remain on plaintiffs: in that case, the Court held the statute “places a burden of proving likelihood of confusion (that is, infringement) on the party charging infringement even when relying on an incontestable registration.” KP Permanent, 125 S.Ct. at 548. Thus, held the Court, “the burden of proving likelihood of confusion rests with the plaintiff.” Id. at 550.

The Court further held that it would be incoherent to place upon defendant the burden of showing nonconfusion:

[A] look at the typical course of litigation in an infringement action points up the incoherence of placing a burden to show nonconfusion on a defendant. If a plaintiff succeeds in making out a prima facie case of trademark infringement, including the element of likelihood of consumer confusion, the defendant may offer rebutting evidence to undercut the force of the plaintiff’s evidence on this (or any) element, or raise an affirmative defense to bar relief even if the prima facie case is sound, or do both. But it would make no sense to give the defendant a defense of showing affirmatively that the plaintiff cannot succeed in proving some element (like confusion); all the defendant needs to do is to leave the factfinder unpersuaded that the plaintiff has carried its own burden on that point. A defendant has no need of a court’s true belief when agnosticism will do.
Id. at 549. For the reasons noted above, I conclude that the majority’s bifurcated analysis and the phrasing of its “nominative fair use” defense violate this first and clear holding of KP Permanent, by placing the burden of showing nonconfusion on defendant.

A second issue in KP Permanent, in contrast, has no bearing to this case. That issue was what role, if any, a finding of likely confusion may play when a defendant asserts the statutory affirmative defense of descriptive fair use under Lanham Act § 33(b)(4). The statutory defense asks whether “the use of the name, term, or device charged to be an infringement is a use, otherwise than as a mark, . . . of a term or device which is descriptive of and used fairly and in good faith only to describe the goods or services of such party, or their geographic origin.” Lanham Act § 33(b)(4), 15 U.S.C. § 1115(b)(4). Unlike the present dispute, which asks whether defendant created a likelihood of confusion by using the plaintiffs’ marks to refer at least in part to plaintiffs, a descriptive fairuse defense would ask whether a defendant may use a trademark (e.g., “BEST PRICE”) in its primary descriptive sense (e.g., “we offer the best price on breakfast cereal”). In the latter case, a threshold inquiry and key distinguishing fact is that defendant would allege that it is not using the plaintiff’s marks to refer to plaintiff at all, but is instead using the mark in a descriptive way without reference to the plaintiff.

In KP Permanent, the alleged infringer argued that it used variations of the term “micro color” to refer to its goods in a descriptive way without reference to the trademark owner. The question before the Court was whether a finding of likely confusion would foreclose assertion of the statutory descriptive fair use defense.[27] The Court held that it would not, concluding that “our holding that fair use can occur along with some degree of confusion does not foreclose the relevance of the extent of any likely consumer confusion in assessing whether a defendant’s use is objectively fair.” 125 S.Ct. at 550.

Despite the majority’s suggestions to the contrary, the Court’s second holding is not earth-shattering — the Court merely held that even though a finding of likely confusion would not foreclose assertion of the statutory descriptive fair use defense, courts might nonetheless factor the extent of confusion into determining whether the purportedly descriptive use was not fair. In the context of descriptive fair use, the Court’s statement makes sense because the extent of confusion might be relevant to determining whether a the mark was “used fairly” or “otherwise than as a mark.” Lanham Act § 33(b)(4), 15 U.S.C. § 1115(b)(4). But descriptive fair use and nominative use are very different animals: in descriptive fair use, the key inquiry is whether a plaintiff’s mark is being used in a descriptive sense, but in nominative use cases, courts ask one question alone: is the defendant’s use likely to confuse?[28] Indeed, even while noting that some cases had treated the extent of confusion as a relevant consideration in descriptive fair use cases, the Supreme Court immediately and unequivocally cautioned that “[c]ertainly one cannot get out of [such cases] any defense burden to negate [confusion] entirely.” 125 S.Ct. at 549.

It makes no sense to force the square peg of KP Permanent’s second holding about descriptive fair use into the round hole of nominative use. Nominative use is nothing more than an inquiry into likely confusion. By reading KP Permanent to permit a bifurcated analysis where confusion is the central focus on each side, the majority reads KP Permanent to hold that both: 1) defendant bears no burden of negating confusion, and 2) defendant bears a burden of negating confusion. Nor might the majority justify its defense by characterizing it as an inquiry into good faith. As stated by the Supreme Court, “Nor would it make sense to provide an affirmative defense of no confusion plus good faith, when merely rebutting the plaintiff’s case on confusion would entitle the defendant to judgment, good faith or not.” Id. Indeed, good faith is merely the flip side of bad faith, and as discussed previously, the intent factor has always remained part of the plaintiff’s case under Lapp.

The majority’s attempt to rely on KP Permanent in support of its bifurcated analysis is further and finally belied by the fact that the Court took pains to note that it was not addressing nominative use. See id. at 546 n. 3 (“Nor do we address the Court of Appeals’s discussion of `nominative fair use.'”). The majority’s transplantation of the Supreme Court’s comments about descriptive fair use into the nominative use context is no more effective than replacing a healthy human heart with one from another species. In short, classic fair use is a defense. Nominative use is not.

III.

Merely criticizing the majority’s approach would be hollow commentary without providing in addition an outline on how nominative use analysis can proceed in a fashion consistent with the Lanham Act, KP Permanent, Prestonettes, Searle, and this Court’s long-standing Lapp test. Needless to say, because the nominative use analysis is an inquiry into the likelihood of confusion, the burden must remain with plaintiffs, although — as in any trademark infringement case — defendant remains free to put on arguments and evidence to rebut those put forth by plaintiffs. I start as I must with reference to our standard ten factors, as phrased in our decision in A & H Sportswear, Inc. v. Victoria’s Secret Stores, Inc.:

(1) the degree of similarity between the owner’s mark and the alleged infringing mark;
(2) the strength of the owner’s mark;
(3) the price of the goods and other factors indicative of the care and attention expected of consumers when making a purchase;
(4) the length of time the defendant has used the mark without evidence of actual confusion arising;
(5) the intent of the defendant in adopting the mark;
(6) the evidence of actual confusion;
(7) whether the goods, competing or not competing, are marketed through the same channels of trade and advertised through the same media;
(8) the extent to which the targets of the parties’ sales efforts are the same;
(9) the relationship of the goods in the minds of consumers, whether because of the near-identity of the products, the similarity of function, or other factors;
(10) other facts suggesting that the consuming public might expect the prior owner to manufacture both products, or expect the prior owner to manufacture a product in the defendant’s market, or expect that the prior owner is likely to expand into the defendant’s market.
237 F.3d 198, 215 (3d Cir.2000) (discussing Interpace Corp. v. Lapp, Inc., 721 F.2d 460 (3d Cir.1983)).[29] The majority is correct that the factors may be tailored to the circumstances: “[n]ot all factors will be relevant in all cases,” and “different factors may properly be accorded different weights depending on the particular factual setting.” Id.

I agree with the majority that Lapp factors one and two are unlikely to carry any weight in a nominative use case. In a case where a defendant is using a plaintiff’s mark to refer to the plaintiff, the use will always be identical.[30] Moreover, it makes little sense to give weight to factor two, the strength of the mark, as nominative uses will be made with both weak and strong marks.

The majority focuses in this case on only four of the ten Lapp factors (factors three through six), but agrees that in some cases, the remaining factors (seven through ten) might be relevant. I believe it is important to explore circumstances where Lapp factors seven through ten may be pertinent. In this case, much of the parties’ services, customers, advertising, and markets are similar and may be in some regards identical. Although the degree of identity or similarity is for the District Court to determine in the first instance, the majority correctly concludes that to the extent the parties share identical services, customers, advertising, and markets, such identity lacks value in determining the likelihood of confusion. See Maj. Op. at 225 (“it would almost be expected that [defendant] and most other defendants in a nominative use case would market themselves through a media in which the marks to which they are referring would be easily recognized and have meaning or relevance, namely channels similar to those used by plaintiff”). However, to the extent that the parties’ services, customers, advertising, and markets are different, factors seven through ten may counsel against a likelihood of confusion. Compare this case to one involving a newscaster. What if the hypothetical newscaster used plaintiffs’ logos as part of a newscast about plaintiffs? Regardless of First Amendment issues, the differences between the parties’ services would themselves counsel against a likelihood of confusion. Thus, as recognized by the majority, factors seven through ten are not per se irrelevant, and the door remains open to them in relevant cases.

In light of these considerations, I recommend that in cases involving colorable nominative use — i.e., where a defendant has used a plaintiff’s mark to refer at least in part to the plaintiff — the District Court consider Lapp factors three through ten, slightly reworded and renumbered one through eight:

1. The price of the goods or services and other factors indicative of the care and attention expected of consumers when making a purchase. As noted by the majority, this remains a relevant inquiry. The cheaper the goods or the less sophisticated the consumers, the more likely that a use may confuse. This factor will require careful attention by the District Court on remand. People buying and selling homes might be likely to take especial care in choosing an agent, but the use here may also give rise to concerns under the “initial interest confusion” doctrine. See Checkpoint Sys., Inc. v. Check Point Software Techs., Inc., 269 F.3d 270, 292 (3d Cir.2001) (holding initial interest confusion actionable under the Lanham Act); see also Lamparello v. Falwell, 420 F.3d 309, 316 (4th Cir.2005) (quoting Checkpoint and noting limitations on initial interest doctrine where use was for criticism or commentary rather than profit).

2. The length of time the defendant has used the mark without evidence of actual confusion arising. “If a defendant’s product has been sold for an appreciable period of time without evidence of actual confusion, one can infer that continued marketing will not lead to consumer confusion in the future.” Versa Prods., 50 F.3d at 205. “The longer the challenged product has been in use, the stronger this inference will be.” Id.

3. The intent of the defendant in using the mark. As noted by the majority, “the key inquiry is whether the mark is being used so as to convey a connection between the parties that may not exist.” Maj. Op. at 226. Although I disagree with the majority’s bifurcation and duplication of the intent inquiry, there can be no doubt that this inquiry is crucial and may include, as noted by the majority, “myriad” factors. See Maj. Op. at 227. Thus, the District Court should inquire as to the nature of the defendant’s conduct and the accuracy of its use of the plaintiffs’ marks (akin to the majority’s third “nominative fair use” prong). These inquiries extend further to looking to the purpose of the use (akin to the majority’s first “nominative fair use” prong),[31] and to whether defendant “took too much” with an eye towards creating consumer confusion (i.e., the second “nominative fair use” prong), such as a competitor using a plaintiff’s mark in logo form, larger type, or different colors, in a way designed to draw attention above and beyond proper nominative use. Indeed, this factor was expressly considered in Searle, where we enjoined the defendant from setting the mark apart in larger or differently colored type, and required it to use “®” and a disclaimer. 715 F.2d at 839, 843.

4. The evidence of actual confusion. The “more evidence of actual confusion that a plaintiff can muster, the stronger the likelihood of confusion in the future, but lack of evidence of actual confusion,” at least where the time period is short or the circumstances do not indicate that such evidence should have been available, “does not raise an inference that there is no likelihood of confusion.” Versa Prods., 50 F.3d at 205. Here, plaintiffs concede for purposes of this motion that there is no such evidence, which for present purposes weighs in defendant’s favor.

5. Whether the goods or services, competing or not competing, are marketed through the same channels of trade and advertised through the same media. Where the parties are competitors and the channels of trade or media are identical, this factor has no relevance because competitors by nature use identical channels and media. However, to the extent that the parties are not competitors, as in the case of the hypothetical broadcaster, or the parties use differing channels of trade or media, this factor lessens the likelihood of confusion because of differences in the target audiences and the places where consumers will see the marks used. In this case, defendant and plaintiffs are not identical competitors but appear to compete in some respects. They also appear to use websites geared towards similar audiences. The extent that the parties are or are not competitors, or do or do not have identical channels of trade or media, should be determined by the District Court upon remand. If identical, then this factor has no weight. If only slightly different, the factor has some but little bearing in reducing the likelihood of confusion. And the greater the differences, the more weight the District Court should place on this factor.

6. The extent to which the targets of the parties’ sales efforts are the same. The rationale behind factor six is similar to that for factor five. In a nominative use case, the mere fact that the parties target identical customers has no bearing. But to the extent the targeted customers differ, that lessens the likelihood of confusion. Upon remand, the District Court should determine the extent to which the targeted customers are identical or not.

7. The relationship of the goods or services in the minds of consumers, whether because of the near-identity of the products or services, the similarity of function, or other factors. The same rationale applies. The mere fact that the goods or services are identical has no probative value because competitors may oftentimes make otherwise nonconfusing nominative use. But to the extent the goods or services differ, that lessens the likelihood of confusion. Upon remand, the District Court should determine the extent to which the parties’ services are similar or not.

8. Other facts suggesting that the consuming public might expect the prior owner to provide both products or services, or expect the prior owner to provide a product or service in the defendant’s market, or expect that the prior owner is likely to expand into the defendant’s market. This factor asks whether consumers would expect plaintiffs to offer defendant’s services, or offer services in defendant’s market, or to expand into defendant’s market. The same rationale applies here as for factors five, six, and seven; namely, to the extent that the consuming public expects plaintiffs to offer defendant’s services, to offer services in the defendant’s market, or to expand into the defendant’s market, this factor has no bearing. But the less the consuming public would expect such conduct from plaintiffs, the more this factor cuts against a likelihood of confusion. Upon remand, the District Court should consider this factor as well.

IV.

The test outlined above best satisfies KP Permanent’s requirement that plaintiffs bear the burden of establishing a likelihood of confusion. It also satisfies the requirements of the Lanham Act, Prestonettes, and Searle, which provide that nominative use is use that is not likely to confuse. Unlike the majority’s approach, this test is judicially manageable because it does not bifurcate and duplicate the analysis.

The majority objects that it would be unfair to deprive a defendant of a nominative fair use defense. This complaint rings hollow: defining “fair use” in terms of nonconfusion does not give the majority license to place on defendant’s shoulders a burden that it should not bear. It is therefore the majority’s formulation that is “unfair.”

Moreover, the majority is simply wrong that defendants will have no way of undercutting a charge of infringement. As the Supreme Court noted in KP Permanent in response to this very objection, a defendant remains free to “offer rebutting evidence to undercut the force of the plaintiff’s evidence on” likelihood of confusion. KP Permanent, 125 S.Ct. at 549. As the Supreme Court held, “all the defendant needs to do is to leave the factfinder unpersuaded that the plaintiff has carried its own burden on that point. A defendant has no need of a court’s true belief when agnosticism will do.” Id.

V.

For the reasons discussed above, I cannot join in the majority’s formulation of a bifurcated test because it runs afoul of the Lanham Act, KP Permanent, Prestonettes, and Searle, each of which make it clear that the burden in nominative use cases remains with plaintiffs. Because the majority’s test impermissibly places the burden of negating confusion on defendant, and is in addition judicially unmanageable, I respectfully dissent. I nevertheless join in the judgment remanding the case for further proceedings.

[*] Hon. William H. Yohn, Jr., Judge of the United States District Court for the Eastern District of Pennsylvania, sitting by designation.

[1] Interestingly, the thinking of the Ninth Circuit Court of Appeals appears to have evolved in this regard. The court reasoned in New Kids On The Block that nominative use does not risk consumer confusion and is “outside the strictures of trademark law.” 971 F.2d at 308. Ten years later, however, it justified replacing the likelihood of confusion test by arguing that the test would be inaccurate in the nominative use context. Playboy Enters., 279 F.3d at 801. As explained below, we believe that neither rationale requires us to abandon the likelihood of confusion test in the nominative use context, and particularly not since the Supreme Court’s decision in KP Permanent Make-Up.

[2] As described above, the Ninth Circuit Court of Appeals stated its view in New Kids On The Block that the use of another’s mark to describe plaintiff’s own product is one step removed from the true scope of the trademark laws and the very concept of trademark infringement. But this thinking provides all the more reason that a plaintiff in that situation should be required to prove that confusion exists before putting the defendant to its proof. If such cases are truly outside of the scope of the trademark laws then a plaintiff would simply fail to show confusion and the court’s inquiry would end.

[3] The concurrence relies heavily on a mix of case law that was, in our view, largely undermined by the Supreme Court’s holding in KP Permanent Make-Up. Before the Supreme Court’s opinion in that case, a finding of likelihood of confusion sounded the death knell for a classic fair use defense in some circuits. See KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 328 F.3d 1061, 1072 (9th Cir.2003) (“[T]here can be no fair use if there is a likelihood of confusion….”); PACCAR, 319 F.3d at 256 (“A finding of likelihood of confusion forecloses a [classic] fair use defense.”). Likewise, the Courts of Appeals for the Fifth and Ninth Circuits stated, prior to KP Permanent Make-Up, that nominative fair use is not an infringement so long as there is no likelihood of confusion. Cairns, 292 F.3d at 1151; Pebble Beach, 155 F.3d at 546. These courts have not considered the viability of their approach since KP Permanent Make-Up was decided. See Ty, Inc. v. Publ’ns Int’l, Ltd., infra, at *7. We rely on the Ninth Circuit precedent insofar as it identifies the need for a separate test for nominative fair use, but we reject its view that this test should be a substitute for likelihood of confusion. A nominative use defendant need only prove fairness and is not required to negate confusion.

[4] Judge Fisher points to Prestonettes, Inc. v. Coty, 264 U.S. 359, 44 S.Ct. 350, 68 L.Ed. 731 (1924), and G.D. Searle & Co. v. Hudson Pharmaceutical Corp., 715 F.2d 837 (3d Cir.1983), as authority for limiting our inquiry to likelihood of confusion. However, even a cursory review of those opinions-dating back 80 and 20 years, respectively-reveals little in the way of guidance for an analysis of nominative fair use. Indeed, no reference to “fair use,” let alone “nominative fair use,” can be found in either opinion. Both cases seem, instead, to stand for the proposition that clear and truthful references to the products of others are permitted. Thus, the underlying theme of those opinions seems to support, rather than undercut, an analysis that specifically examines whether the use is clear and truthful, as does the affirmative defense we posit, and that the Supreme Court has embraced.

Moreover, our analysis does not disturb the holding of either Prestonettes or Searle. After our decision today, courts will continue to inquire into confusion, as they did in both of those cases. Since KP Permanent Make-Up has altered the landscape, however, we believe it is clear that likelihood of confusion need not, and should not, be solely determinative of nominative fair use. Fairness is a distinct concept from confusion, and it should be measured through a distinct inquiry.

[5] We review a district court’s conclusion as to likelihood of confusion for clear error. A & H Sportswear II, 237 F.3d 198, 237 (3d Cir. 2000).

[6] This principle emerges from our analogous holdings in traditional trademark infringement cases, in which we have noted that “a defendant’s mere intent to copy, without more, is not sufficiently probative of the defendant’s success in causing confusion to weigh such a finding in the plaintiff’s favor.” A & H Sportswear II, 237 F.3d at 225-26. Rather, the plaintiff should make some showing that the defendant adopted its mark with the intent to confuse or deceive the public. Id.; see also Checkpoint Sys. Inc. v. Check Point Software Techs., Inc., 269 F.3d 270, 286 (3d Cir.2001) (noting that “a party’s intentional use of another party’s mark to cause confusion” weighs in favor of finding likelihood of confusion) (emphasis added).

[7] The concurrence argues that our nominative fair use defense is merely this same intent inquiry dressed in different clothing. The tests for confusion and fairness are not duplicative, however, because a plaintiff can also establish likelihood of confusion through seven other Lapp factors besides intent. It is the circumstance in which a court does not find bad intent but does find confusion that a nominative fair use defense will be most useful.

[8] See, The New Kids on the Block v. Gannett Satellite Information Network, Inc., 971 F.2d 302 (9th Cir.1992); Kareem Abdul-Jabbar v. General Motors Corp., 85 F.3d 407 (9th Cir. 1996); Cairns v. Franklin Mint Co., 292 F.3d 1139 (9th Cir.2002); Playboy Enterprises, Inc. v. Welles, 279 F.3d 796 (9th Cir.2002); Brother Records, Inc. v. Jardine, 318 F.3d 900 (9th Cir.2003); Horphag Research Ltd. v. Pellegrini, 337 F.3d 1036 (9th Cir.2003); Playboy Enterprises, Inc. v. Netscape Communications Corp., 354 F.3d 1020 (9th Cir.2004).

[9] Although the test crafted by the Ninth Circuit Court of Appeals did not include this aspect under the first prong, as such, nonetheless, it does allude to it as part of the second prong, as we discuss below.

[10] Interpace Corp. v. Lapp, Inc., 721 F.2d 460 (3d Cir.1983).

[11] Like our Court and other Courts of Appeals, the Ninth Circuit uses a multi-factor test for likelihood of confusion that looks to, inter alia, the similarity of the marks, strength of the plaintiff’s mark, and other factors. See AMF Inc. v. Sleekcraft Boats, 599 F.2d 341 (9th Cir.1979).

[12] Even the majority admits “it seems that only the Second, Fifth, and Sixth Circuits have referenced the nominative fair use defense by name and even on these occasions have done so only to refer to what district courts had done with the issue or to decline to adopt the Ninth Circuit’s test as a whole.” Maj. Op. at 218.

[13] I recognize that the majority does not adopt the Ninth Circuit test verbatim. However, the majority nevertheless errs in treating its test as an affirmative defense because that defense, even as modified, analyzes nothing more than core issues of likelihood of confusion, issues that are within the purview of plaintiffs’ case.

[14] In addition, we affirmed the limited injunction, which enjoined defendant from setting the mark apart in larger or differently colored type, and required it to use “®” and use the disclaimer “a product of G.D. Searle, not a Hudson product.” Id. at 839, 843.

[15] Notably, in Pebble Beach, where the Fifth Circuit declined to adopt the Ninth Circuit’s test verbatim, the Court cited Searle as support for the proposition that “[t]his right to use a mark to identify the markholder’s products — a nominative use — however, is limited in that the use cannot be one that creates a likelihood of confusion as to source, sponsorship, affiliation, or approval.” 155 F.3d at 546 (citing, inter alia, Searle, 715 F.2d at 842) (emphasis added). Unlike the majority, the Fifth Circuit clearly understood Searle to be a nominative use case and that the issue was whether the use created a likelihood of confusion.

[16] The majority strangely complains that neither Prestonettes nor Searle contain any “reference to `fair use,’ let alone nominative fair use.'” Maj. Op. at 223 n.4. Of course they don’t, because both cases treat the analysis as one of likely confusion, not as an affirmative defense. And to state the obvious, neither case was decided before the Ninth Circuit coined the phrase “nominative use.” The majority’s approach has the impermissible effect of attempting to overrule Searle, which can only be done by the en banc Court.

[17] Because nominative use defendants use a plaintiff’s mark rather than adopt it, the more accurate phrasing would be “the intent of the defendant in using the mark.” I have followed this approach in Section III, infra.

[18] To be clear, the panel is unanimous in holding that in an appropriate case, all but the first two Lapp factors may be relevant to the nominative use analysis. See Maj. Op. at 225-26. Moreover, although the majority focuses on only four of the factors, id. at 225, 231-32, it at no point suggests that those factors constitute the sole set of factors for nominative use analysis; instead, it notes that the analysis will “include” those four factors. See Maj. Op. at 225. Though the majority expresses doubt regarding the weight of the last four Lapp factors to this particular dispute, ibid., at no point does it hold that the District Court is prohibited from considering them. As it states: “We find that all of the other Lapp factors, while perhaps not appropriate for analysis in this particular case, could be analyzed in future nominative use cases, depending on the factual situation.” Ibid. (emphasis added). Apparently concluding that the parties are actual or close competitors, the majority appears to suggest that the last four factors will be of little relevance in this case. Perhaps, but as the majority implicitly recognizes, the determination of the degree of dissimilarity and its weight is for the District Court to determine in the first instance.

Section III, infra, outlines how the eight relevant Lapp factors, including the final four, should be analyzed in a nominative use case, including this case. If the District Court finds that the parties’ services, customers, advertising, and markets are highly similar or identical — i.e., the more the parties are competitors — such findings have no bearing under the last four Lapp factors in establishing a likelihood of confusion. Competitors often make nominative use for purposes of “comparison, criticism, [or a] point of reference.” New Kids, 971 F.2d at 306; see also Searle, 715 F.2d at 841 (competitor selling laxatives described as “Equivalent to METAMUCIL”). In nominative use cases, focusing on the parties’ status as competitors would lead to false findings that all nominative uses are confusing; accordingly, the parties’ possible status as competitors has little to no bearing on establishing a likelihood of confusion, and other factors must be addressed to analyze likelihood of confusion.

But when the parties are not competitors, the last four Lapp factors may become relevant in cutting against a likelihood of confusion in the nominative use context. The more the parties’ services, customers, advertising, or markets diverge, the more they reduce the likelihood of confusion. This is because differently targeted consumers may see the marks in different contexts, through different advertising, and in relation to different goods or services. These differences may counsel against a likelihood of confusion.

[19] Although proof of actual confusion is not required, the more such evidence there is, the greater the likelihood of confusion. Versa Products Co., Inc. v. Bifold Co. (Mfg.) Ltd., 50 F.3d 189, 205 (3d Cir.1995). Conversely, a lack of actual confusion evidence does not raise an inference of nonconfusion “when the particular circumstances [do not] indicate such evidence should have been available.” Id. (alteration in original, quote marks deleted). But “[i]f a defendant’s product has been sold for an appreciable period of time without evidence of actual confusion, one can infer that continued marketing will not lead to consumer confusion in the future.” Id.

[20] I further object to the first prong of the “nominative fair use” defense because of its conjunctive requirement that a plaintiff’s mark be used to describe both plaintiff and defendant. Although “nominative” use requires that the mark be used at a minimum to refer to plaintiff, proper nominative use may or may not also use a plaintiff’s mark to refer to the defendant’s goods or services. Nominative use may occur “where the defendant uses a trademark to describe the plaintiff’s product, rather than its own.” New Kids, 971 F.2d at 308. Sometimes the use of the mark may be, as noted in New Kids, merely for “comparison, criticism, [or a] point of reference.” Id. at 306. In any case, I would note that even under the majority’s phrasing, “necessary” under the first prong does not mean “indispensable” — “the court need not find that the use of the mark is indispensable in order to find this factor fulfilled.” Maj. Op. at 229.

Another troubling problem that may arise from the first prong’s conjunctive phrasing is that the majority may have unintentionally foreordained the outcome of these proceedings through its passing statement that “[defendant] is not attempting to use [plaintiffs’] marks to refer to [defendant’s] own services, but rather is using the marks to refer to [plaintiffs’] services.” Maj. Op. at 225. The majority appears to suggest that defendant did not use the marks to refer to itself. I will emphasize that this determination is not for this Court to make and should be made by the District Court upon remand.

[21] The majority is also unclear regarding the phrasing of the second prong, which asks “[i]s only so much of the plaintiff’s mark used as is necessary to describe plaintiff’s products or services?” Maj. Op. at 228 (emphasis added). Later in the opinion, the majority states “the proper focus under this prong is on whether only so much of plaintiff’s mark as is reasonably necessary to identify plaintiff’s product or service has been used by defendant.” Id. at 230 (emphasis on “reasonably” added). There may be worlds of difference between what is “necessary” and what is “reasonably necessary.” Although I read the majority’s second statement as clarification of the meaning of “necessary,” the majority is not clear on this important point.

[22] The identity of the Lapp analysis and the affirmative defense is further underscored by the majority’s additional descriptions of the Lapp “intent” inquiry: whether “the defendant made use of the plaintiff’s mark with the very purpose of causing consumers to think the plaintiff endorses or sponsors plaintiff’s good or service,” Maj. Op. at 226 (emphasis added), and whether defendant “used the mark with the intent to confuse the public as to the relationship between the defendant and the plaintiff.” Id. at 227 (emphasis in original).

[23] Just like the majority, the Ninth Circuit’s first two “nominative fair use” prongs looks to purpose and prominence. The Ninth Circuit’s third prong embraces truthfulness and more, unabashedly requiring that “the user must do nothing that would, in conjunction with the mark, suggest sponsorship or endorsement by the trademark holder.” New Kids, 971 F.2d at 308. Try as it might to dress up the Ninth Circuit’s test in new clothes, the test was and remains an inquiry into likelihood of confusion, and the majority repeats the Ninth Circuit’s error of placing the burden of negating confusion onto the defendant.

[24] Ironically, the majority criticizes the District Court in Liquid Glass Enterprises, Inc. v. Dr. Ing. h.c.F. Porsche AG, 8 F.Supp.2d 398 (D.N.J.1998), suggesting that the Court “conflated the [New Kids] test with the plaintiff’s burden of proving likelihood of confusion.” Maj. Op. at 218. Yet the majority’s test stands guilty of the same accusation that it aims at Liquid Glass.

[25] Trying to find some use for its affirmative defense, the majority argues that “[i]t is the circumstance in which a court does not find bad intent but does find confusion that a nominative fair use defense will be most useful.” Maj. Op. at 227 n.7 (some emphasis added). Ironically, such a use would be almost the only circumstance where a nominative fair use defense would even theoretically be available. If likelihood of confusion exists due to bad intent, the defense will be unavailable because the intent is bad. If likelihood of confusion doesn’t exist, then the defense won’t needed. This leaves the circumstance noted by the majority: if likelihood of confusion is somehow shown despite a lack of bad intent, defendant can point to its good intent to prevail. But considering that a key inquiry in a nominative use case is why the defendant has used the plaintiff’s mark (purpose) and how it has been used (prominence and truthfulness), it is hard to imagine a case where necessary, non-prominent, and truthful use would be likely to confuse.

Such a finding under the facts of this case would appear to require highly unsophisticated consumers and an abundance of actual confusion evidence. But this hypothetical possibility — notably, circumstances not present in this case, possibly rendering the majority’s approach as nothing more than dicta — does not warrant the majority’s invention of an affirmative defense that duplicates likelihood of confusion intent inquiries and places upon defendant an affirmative burden of showing good faith. In any case, the majority’s unlikely scenario of likely confusion despite good faith may not warrant a complete bar on such use, and under appropriate circumstances, may merit at most a constrained remedy that continues to permit such truthful use, such as ordering clarifying language or a disclaimer, as happened in Searle. Cf. 715 F.2d at 839, 843.

[26] Lanham Act § 32(1)(a) states that “[a]ny person who shall, without the consent of the registrant — (a) use in commerce any reproduction, counterfeit, copy, or colorable imitation of a registered mark in connection with the sale, offering for sale, distribution, or advertising of any goods or services on or in connection with which such use is likely to cause confusion, or to cause mistake, or to deceive … shall be liable in a civil action by the registrant. . . .” Lanham Act § 32(1)(a), 15 U.S.C. § 1114(1)(a). Similarly, Section 43(a)(1)(A) asks whether the use is “likely to cause confusion, or to cause mistake, or to deceive as to the affiliation, connection, or association of such person with another person.” Id. § 43(a)(1)(A), 15 U.S.C. § 1125(a)(1)(A). Compare this language to the majority’s affirmative “nominative fair use” defense — which asks whether “defendant’s conduct or language reflect[s] the true and accurate relationship between plaintiff and defendant’s products or services.” Maj. Op. at 228 (emphasis added). It’s the same inquiry, and it improperly places the burden of negating confusion on defendant.

[27] Underlying the case was a long-standing circuit split over whether a finding of likely confusion foreclosed a statutory descriptive fair use defense. See KP Permanent, 125 S.Ct. at 547 (collecting cases).

[28] Descriptive fair use and nominative use “are quite different from each other and bear little resemblance except that both are types of non-infringing use.” 2 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition, § 11:45 (4th ed.2005). As stated by the Ninth Circuit in distinguishing nominative use from descriptive (or classic) fair use:

The distinction between classic and nominative fair use is important for two reasons:

(1) classic and nominative fair use are governed by different analyses; and (2) the classic fair use analysis only complements the likelihood of customer confusion analysis set forth in Sleekcraft, whereas the nominative fair use analysis replaces the Sleekcraft analysis.

Cairns, 292 F.3d at 1150. In other words, descriptive fair use looks beyond confusion considerations whereas nominative use looks only to a likelihood of confusion.

[29] This updated formulation post-dates Lapp and addresses both competitive and non-competitive uses.

[30] “When a defendant uses a trademark nominally, the trademark will be identical to the plaintiff’s mark, at least in terms of the words in question. Thus, application of the [normal multi-factor] test, which focuses on the similarity of the mark used by the plaintiff and the defendant, would lead to the incorrect conclusion that virtually all nominative uses are confusing.” Playboy Enterprises, Inc. v. Welles, 279 F.3d 796, 801 (9th Cir.2002).

[31] But as noted, I cannot ascribe to the majority’s conjunctive phrasing that would require that the use be necessary to describe both a plaintiff’s product or service and a defendant’s product or service. See supra note 11.