Trademark Use Rides Again

Back in 2007, the Iowa Law Review published a dialogue between Graeme Dinwoodie and Mark Janis, on the one hand, and Stacey Dogan and Mark Lemley on the other.1 The topic was “trademark use,” and the question was whether such a doctrine really exists. Dinwoodie and Janis said no—that while only commercial use of a trademark can be considered infringing, there is no threshold requirement that the plaintiff prove that the defendant has used the mark in some particular “trademark” way.2 Dogan and Lemley said yes —that some “uses” of a mark simply don’t trigger liability, and a court needs to determine, as a threshold matter, whether the defendant has used the mark “as a brand.”3 I was not a disinterested bystander on this question. When I wrote, responding to both papers, I argued that trademark law does, and must, impose liability for only particular kinds of uses of a mark—uses “as a mark.”4 But because “trademark use” can only be determined from the perspective of consumers, I argued, the question of whether a particular use qualifies inevitably collapses into the likelihood of confusion analysis (and therefore isn’t a separate, threshold question.)5  

The impetus for this now well-cited exchange was a series of cases in which the issue was whether a party that delivers keyword advertisements infringes a mark owner’s rights by selling to competitors the right to have their ads triggered by keywords that correspond to the plaintiff’s trademark.6 In the years following the trademark use debate, courts largely sided with Dinwoodie and Janis (or, I think more accurately, with me)—rejecting the contention that trademark use is a separate, threshold requirement.7 That doesn’t mean that courts typically have found keyword advertising uses to be infringing. In the main, courts have held defendants’ sales of keywords for advertising non-infringing, though they usually have reached that result by finding that the defendants’ conduct was not likely to cause confusion.8

Still, for most trademark lawyers, the idea that there’s no such thing as “trademark use” always seemed wrong. In fact, trademark use is everywhere in trademark law—in cases dealing with acquisition of common law rights, in priority disputes, in infringement cases, and in cases involving a variety of defensive doctrines. And, as Alex Roberts details in her excellent article Failure to Function, even more pervasively in the registration context.9

The problem is that trademark use is only at issue when it’s lacking; when it’s present, it’s invisible. In this respect it should be no surprise that the Trademark Office has no real theory of trademark use, only a rule that prohibits registration when the claimed mark fails to function as a mark. Trademark use is also primarily a functional consideration; an indicator has been “used as a mark” when, by virtue of its use, consumers are likely to regard it as source-indicating. Because that is an empirical question for which the Trademark Office typically has no empirical evidence, failure to function has a sort of “know it when you see it” quality. Indeed, the Trademark Manual of Examining Procedure describes a purported mark’s failure to function primarily by reference to other functions performed the claimed matter—the theory being that we can identify when matter fails to function as a mark by determining whether the matter has some other function in relation to the goods or services. So, for example, a claimed mark might fail to function as a mark “because it is merely ornamentation,”10 or “informational matter,”11 or a “model or grade designation.”12

Against this background, Roberts advances three arguments: (1) use as a mark is undervalued as a theoretical and practical matter;13 (2) that is at least in part because we lack sufficiently clear rules for determining use as a mark;14 and (3) courts and the Trademark Office should merge considerations of distinctiveness and use as a mark such that the two are considered to be interdependent and inversely related (the less distinctive a term, the more evidence of use as a mark are needed, and vice versa).15

On the first point, Roberts notes that eligibility for registration depends not only on distinctiveness (whether the purported mark is the sort of thing that is capable of designating source), but on the applicant’s use of that mark in a source-designating way.16 That is to say that registrability is not only about whether the mark is ownable, but whether it is in fact owned by the party who claims it. The question here is about the qualitative characteristics of the use, not merely whether the claimant has made enough of it. No amount of use will suffice if it is not the right kind.

The difficulty is, of course, in determining what counts as the “right kind” of use. Here Roberts criticizes the Trademark Office and T.T.A.B. for failing to develop clear guidance, which she attributes to their overemphasis on distinctiveness.17

Without use as a mark, there can be no trademark, and consequently no trademark rights. Yet distinctiveness has received the lion’s share of attention, generating rules and tests applied in thousands of cases and USPTO decisions and discussed in hundreds of articles, books, and practice guides. A term’s inherent qualities are often treated as the sole predictor of whether consumers will understand it as a mark. At the same time, the USPTO and federal courts have struggled to articulate and a apply a clear standard for use as a mark and endeavored to separate it from distinctiveness.18

I agree with Roberts that use as a mark is underplayed in many discussions of trademark registration. But it is hard to know whether it is underappreciated by the Trademark Office. According to Roberts, the Office issued 26,450 failure to function refusals between 2003 and 2016.19 Since it is impossible to know how many refusals it should have issued, it is difficult to put that number in perspective. To be sure, Roberts discusses several instances in which the Trademark Office (perhaps erroneously) missed the opportunity to reject an application on failure to function grounds. Meanwhile, she notes, distinctiveness is evaluated in “thousands of cases and USPTO decisions.”20 But some errors are inevitable in a system that processes so many applications, and we have neither a sense of how frequently failure to function rejections were missed nor a baseline against which to evaluate that error rate.

Perhaps more fundamentally, there may be good reasons for the Trademark Office to focus primarily on distinctiveness. Examining attorneys base their judgments regarding use as a mark on a single specimen of use —they can’t, and don’t even attempt to, determine how the applicant generally uses the claimed mark. As a result, accepting that the claimed mark functions as a mark really just means accepting that the claimant used it as a mark in that one instance. Distinctiveness is in this way much more suited to the registration context, because distinctiveness deals with characteristics that depend less on the specific details of the claimant’s use.21 In that respect, use in the registration context isn’t really a proxy for how consumers will see uses of the mark in general; it’s just a way of making sure the mark can be and has been used in a way that would lead consumers to draw a source conclusion.

Roberts is certainly right that there are no clear rules for determining use as a mark. It seems unlikely, however, that the Trademark Office could do much better given modern law’s functional understanding of the concept. As Roberts acknowledges, use as a mark is ultimately determined from the perspective of consumers.22 Since it is not feasible in the registration context to test that understanding with survey evidence, the best examining attorneys can do is rely on rules of thumb.23 Roberts suggests that Office could look to empirical studies to learn at the wholesale level what sorts of clues consumers rely on to draw source conclusions. But as it turns out, the features that literature emphasizes correspond reasonably well to those the T.T.A.B. has highlighted. Specifically, the empirical research suggests that consumers focus on features like shape, color, prominence, font, and marketing context in drawing conclusions about whether particular words or other indicators serve as branding elements.24 Those are the same sorts of contextual clues on which the PTO focuses, as Roberts notes.25

The problem, then, doesn’t seem to be that the law hasn’t identified the right proxies. It’s more that proxies are only proxies, and functional evaluation resists predictable application. So it is with much of modern trademark law. The more holistic and robust use as a mark doctrine that trademark law once employed was tractable precisely because the law used to have a much stronger formalist dimension.26

Roberts’s argument for combining use as a mark and distinctiveness is an intriguing one. Given the empirical evidence demonstrating that contextual clues have a strong effect on consumer perception of source indication, there’s an obvious logic to making the Abercrombie categories (which are based on assumptions about consumer reaction that aren’t empirically supported) less determinative.27

There are a couple of reasons to be concerned, however. First is that, to the extent use as a mark is unclear, combining it with distinctiveness will only embed the lack of clarity in a more complicated analysis. Absent better tools for identifying use as a mark (formal tools that Roberts may resist), it seems unlikely that combining that concept with distinctiveness will lead to clearer or better results.

Second, neither the concept of distinctiveness nor the Abercrombie methodology are motivated solely by assumptions about consumer understanding of particular types of terms. Descriptive terms (and other terms assimilated to that category) were once not considered trademark subject matter (categorically), and that wasn’t because courts assumed those terms could never indicate source. Indeed, unfair competition actions were available to parties using descriptive terms when they could prove the terms did, in fact, indicate source, and that others were using the terms to divert customers. Descriptive terms were excluded from trademark proper because they didn’t automatically indicate source, and because others had potentially legitimate reasons to use those terms. The prospect of legitimate use by competitors remains a concern with respect to these terms, and as a result, there remain good reasons to discourage claiming of descriptive terms even if they could be used in a way that indicates source.  

A related concern has to do with the reason distinctiveness tends to dominate in the registration process in the first place—namely the fact that use as a mark is determined only by what is shown in the specimen, while that use can support registration of the mark in much broader form. Take, for example, an application to register “wonderful” in block letters for cookies. Imagine that application is supported by a specimen showing prominent use of that highly descriptive term in a stylized format with notable colors. When the Trademark Office emphasizes distinctiveness, it can easily reject the application on descriptiveness grounds. But if it were to combine use as a mark and distinctiveness, the Office might acknowledge the possibility that the context of the applicant’s use would lead consumers to treat the otherwise descriptive term as a trademark and therefore register the mark. Note, however, that the registration would issue for the mark in block letters —a format lacking all of the context that might make consumers regard it as a source in the context of the specimen. The benefits of registration then would accrue to the mark as registered, and it could be enforced against even very different uses.    

For that reason, the attractiveness of combining use as a mark and distinctiveness is largely a function of the feasibility of Roberts’s “radical” proposal that courts limit the scope of protection (or at least limit the import of the registration) to the uses identified in the specimens.28 Put differently, if we are to determine registrability primarily on the basis of the use shown in the specimen, then perhaps we ought to treat the specimen as the claim.29 Doing so would introduce greater coherence to the system as a whole—which currently is of two minds about the significance of registration.30

One recent case provides some hope in this respect, and some reason to think Roberts’s proposal might not be all that radical after all. In LTTB, LLC v. Redbubble, Inc.,31 the plaintiff owned incontestable trademark registrations for LETTUCE TURNIP THE BEET for, among other things, various clothing items.32 When LTTB initially applied to register that phrase, it submitted specimens showing use “in a large font across the center of the packing box, tote bag, and shirts.”33 The Trademark Office rejected the application on the ground the claimed mark failed to function since it was merely ornamental, a rejection LTTB overcame only when it submitted different specimens showing use of the mark on product labels or “hang tags.”34 But then LTTB sought to enforce its mark against a company that operated a marketplace on which third parties sold products (including t-shirts) bearing the phrase “Lettuce Turnip the Beet.”35 The court rejected LTTB’s claim, emphasizing that the Trademark Office only allowed registration of LTTB’s mark when it could show a different kind of use than the one to which it was objecting.36

According to the court, Redbubble’s defense “either implicat[ed] the rule that ‘decorative or ornamental’ features are not subject to trademark protection or the exclusion for ‘aesthetic functionality.’”37 The rule about decorative or ornamental features is, of course, only a rule of registration. Aesthetic functionality is the analogous validity concept in infringement actions, and the court’s uncertainty about the relationship between those two concepts reflects a broader uncertainty about the relationship between registration and infringement. What’s unusual about the court’s approach is that courts, including the Ninth Circuit, have been reluctant to invoke aesthetic functionality in cases in which the defendant claims, not that the mark is invalid, but that it simply can’t be enforced against the defendant’s use.38 But that reluctance was not in evidence here. The Trademark Office denied LTTB registration when LTTB could only show use of the phrase on the front of its products, concluding that sort of use is not “use as a mark.”39 LTTB could not overcome that refusal by showing a different type of use but then turn around and enforce its rights against a defendant making the very same kind of non-trademark use for which it was denied registration.

LTTB obtained its registration by submitting samples showing the claimed brand name being used on labels and hangtags. Were some competitor to use a brand name in similar fashion that created a likelihood of confusion, liability might follow. LTTB may not, however, recover for alleged trademark infringement based on any competitors’ use of the very kind of designs that the PTO found not to be eligible for trademark protection.40

The LTTB court framed its decision in aesthetic functionality terms, but this is trademark use by another name. And even if the court’s specific approach does not stand, its instinct to rely on the specimen to limit the scope of registered trademark rights is an important one. If Roberts’s focus on use as a mark can help push a broader rethinking of the relationship between registration and infringement, that will be a significant accomplishment.


  1. [1]. See generally Graeme B. Dinwoodie & Mark D. Janis, Confusion Over Use: Contextualism in Trademark Law, 92 Iowa L. Rev. 1597 (2007) [hereinafter Dinwoodie & Janis I]; Stacey L. Dogan & Mark A. Lemley, Grounding Trademark Law Through Trademark Use, 92 Iowa L. Rev. 1669 (2007); Graeme B. Dinwoodie & Mark D. Janis, Lessons from the Trademark Use Debate, 92 Iowa L. Rev. 1703 (2007) [hereinafter Dinwoodie & Janis II].

  2. [2]. Dinwoodie & Janis I, supra note 1, at 1602–03.

  3. [3]. Dogan & Lemley, supra note 1, at 1676.

  4. [4]. Mark P. McKenna, Trademark Use and the Problem of Source, 2009 U. Ill. L. Rev. 773, 773–78.

  5. [5]. Id. Likelihood of confusion is evaluated using a multifactor test that considers, among other things, the similarity of the parties’ marks, the similarity of the parties’ goods or services, the likelihood that the plaintiff will “bridge the gap” between the parties’ respective goods or services, and the defendant’s intent. See, e.g., Virgin Enters. Ltd. v. Nawab, 335 F.3d 141, 146–48 (2d Cir. 2003).

  6. [6]. Compare Buying for the Home, LLC v. Humble Abode, LLC, 459 F. Supp. 2d 310, 323 (D.N.J. 2006) (holding keyword use by advertisers constitutes trademark use in commerce), and Edina Realty, Inc. v., No. Civ. 04-4371JRTFLN, 2006 WL 737064, at *3 (D. Minn. Mar. 20, 2006) (holding the same), with Merck & Co. v. Mediplan Health Consulting, Inc., 425 F. Supp. 2d 402, 416 (S.D.N.Y. 2006) (holding keyword use is not trademark use in commerce). Compare Washingtonpost.Newsweek Interactive Co. v. Gator Corp., No. Civ. A. 02-909-A, 2002 WL 31356645, at *1 (E.D. Va. July 16, 2002) (holding use of keywords to trigger adware constitutes trademark use in commerce), with 1-800 Contacts, Inc. v., Inc., 414 F.3d 400, 412 (2d Cir. 2005) (holding use of keywords to trigger adware is not trademark use in commerce), and Wells Fargo & Co. v., Inc., 293 F. Supp. 2d 734, 764 (E.D. Mich. 2003) (holding the same), and U-Haul Int’l, Inc. v., Inc., 279 F. Supp. 2d 723, 729 (E.D. Va. 2003) (holding the same).

  7. [7]. See, e.g., Rescuecom Corp. v. Google Inc., 562 F.3d 123, 139–41 (2d Cir. 2009).

  8. [8]. 5 J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition §25A:7 (5th ed. 2019) (“Almost all District Courts have found that no likelihood of confusion was caused by the purchase of keywords alone.” (citing cases)).

  9. [9]. Alexandra J. Roberts, Trademark Failure to Function, 104 Iowa L. Rev. 1977, 1985–86 (2019).

  10. [10]. U.S. Patent & Trademark Office, Trademark Manual of Examining Procedure §1202.03 (2018).

  11. [11]. Id. at §1202.04.

  12. [12]. Id. at §1202.16(a).

  13. [13]. See Roberts, supra note 9, at 1989.

  14. [14]. Id. at 2016.

  15. [15]. Id. at 2039–40.

  16. [16]. Id. at 1994–95.

  17. [17]. Id. at 1986–87.

  18. [18]. Id. at 1982 (citations omitted).

  19. [19]. Id. at 1981 n.13.

  20. [20]. Id. at 1982.

  21. [21]. This is obviously an overstatement, in the sense that usage often informs the extent to which the mark describes, or is generic for, the goods for which it is claimed. But to the extent that is true, use as a mark is already necessarily part of the distinctiveness analysis, whether or not evaluated as a distinct consideration.

  22. [22]. Roberts, supra note 9, at 2018. Or at least that is what the law purports to do. For the most part, the law embeds assumptions about consumer behavior, and significant recent research has demonstrated that many of those assumptions are incorrect. See Jake Linford, Are Trademarks Ever Fanciful?, 105 Geo. L.J. 731, 757 (2017); Jake Linford, The False Dichotomy Between Suggestive and Descriptive Trademarks, 76 Ohio St. L.J. 1367, 1419 (2015); Mark P. McKenna, Testing Modern Trademark Law’s Theory of Harm, 95 Iowa L. Rev. 63, 94 (2009). But even if its assumptions are wrong and perhaps even mask other normative goals, the law lacks any vocabulary for alternative orientation, and Roberts embraces the idea that use as a mark is ultimately about consumer understanding.

  23. [23]. Roberts, supra note 9, at 2019–20.

  24. [24]. Id. at 2020–24. As she notes, one empirical “study found that context—the common indicators of trademark use... such as large, stylized font and prominent placement—had a substantial effect on consumer perception.” Id. at 2020 (citing Thomas R. Lee et al., An Empirical and Consumer Psychology Analysis of Trademark Distinctiveness, 41 Ariz. St. L.J. 1033, 1039–54 (2009)).

  25. [25]. Id. at 2003–04. See also McCarthy, supra note 8, at §3:4 (“Some of the common markers of whether a word, phrase or picture is being used as a trademark are: larger-sized print, all capital letters or initial capitals, distinctive or different print style, color, and prominent position on label or in advertising.”).

  26. [26]. See Mark P. McKenna, Property & Equity in Trademark, 24 Marq. Intell. Prop. L. Rev. (forthcoming 2020).

  27. [27]. The Abercrombie categories, first described in Abercrombie & Fitch Co. v. Hunting World, Inc., 537 F.2d 4, 9 (2d Cir. 1976), are used to determine the distinctiveness of marks by placing them in categories according to the relationship between the mark and the good.

  28. [28]. Roberts, supra note 9, at 2043.

  29. [29]. For more on claiming challenges in trademark law, see Jeanne C. Fromer & Mark P. McKenna, Claiming Design, 167 U. Pa. L. Rev. 123, 146–60 (2018).

  30. [30]. See Rebecca Tushnet, Registering Disagreement: Registration in Modern American Trademark Law, 130 Harv. L. Rev. 867, 878–81 (2017).

  31. [31]. LTTB, LLC v. Redbubble, Inc. 385 F. Supp. 3d 916, 916 (N.D. Cal. 2019).

  32. [32]. Id. at 917.

  33. [33]. Id. at 919.

  34. [34]. Id.

  35. [35]. Id. at 917.

  36. [36]. Id. at 921–22.

  37. [37]. Id. at 920.

  38. [38]. Id. at 921. (“Redbubble does not suggest that LTTB’s registered trademarks areper seinvalid, and nothing in this order should be construed as so holding. Redbubble instead appropriately frames the argument as precluding LTTB from showing a likelihood of confusion as to source, where the mere use of the pun on the face of various products cannot be source-identifying.”).

  39. [39]. Id. at 919.

  40. [40]. Id. at 921.


John P. Murphy Foundation Professor of Law, Notre Dame Law School.