Century 21 Real Estate Corp. v. Lending Tree, Inc.

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.

[. . .]

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. 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. 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 "represented 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." 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.

[. . .]

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. 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., 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, "since 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 ….” 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, 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.

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.

Furthermore, this issue is one of first impression in our Court. The United States District Court for the District of New Jersey did venture into the area of nominative fair use defense in Liquid Glass Enters. v. Dr. Ing. h.c.F. Porsche AG. 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. 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. 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.

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, 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." 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.

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.

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: Where 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.

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. 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 "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, 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. 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, 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."

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.

[. . .]

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.

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.” 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.

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." 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. 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, and evidence of a defendant's bad faith. 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.

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, "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. 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. 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. 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, "the Lanham Act does not compel a competitor to resort to second-best communication." 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. 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.

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.

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:

What 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. 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." 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. 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 "accuracy of course has to be a consideration in assessing fair use." 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.

[. . .]