google it: verb status doesn't mean Google is generic

Elliot v. Google Inc., No. CV–12–1072, 2014 WL 4447764 (D. Ariz. Sept. 10, 2014)

Based on the evidence submitted here, Google seems to have thought long and hard about genericity, and prepared consumer surveys accordingly.  FWIW, my opinion has long been that generic and brand understandings can coexist in the same person; many people who ask for “a kleenex” are also aware that there is a particular brand of tissue called Kleenex.  And it’s much more so when the name is used as a different part of speech.  “To google” is an activity; Google is an entity.

The Google marks at issue covered “computer hardware; computer software for creating indexes of information, indexes of web sites and indexes of other information resources” (the 502 mark) and a mark with a longer description for pretty much the same thing except this time starting with “Computer services” (the 075 mark).  Elliot registered 763 domain names that combined “google” with a brand, person, or term (e.g., googlebarackobama.net), and Google promptly filed a UDRP proceeding.  The UDRP naturally ordered the transfer of the domain names.  Elliot then sued for cancellation of the marks.  Google’s counterclaims were the Lanham Act/state ones you’d expect. This opinion deals with summary judgment on genericity.

The court’s summary of Elliot’s argument:

Plaintiffs contend the GOOGLE mark has become generic because a majority of the public understands the word google, when used as a verb, to mean the indiscriminate act of searching on the internet without regard to the search engine used. Underlying Plaintiffs’ argument is the proposition that verbs, as a matter of law, are incapable of distinguishing one service from another, and can only refer to a category of services.

Google responded that there was no evidence that could justify a finding that a significant portion, let alone a majority, of the consuming public failed to identify Google as an indicator of source, regardless of how the term was employed grammatically.

The relevant question was the primary significance of the mark to the consuming public, in connection with the goods or services at issue. A trademark doesn’t cease to function when it’s used primarily as a verb; “a trademark performs its statutory function so long as it distinguishes a product or service from those of others and indicates the product’s or service’s source.”  If someone verbifies a brand name, if they mean that the activity will be performed with the trademark owner’s product or service, then they’re still using the mark as source-denoting: “e.g., ‘I will PHOTOSHOP the image’ could mean the act of manipulating an image by using the trademarked Photoshop graphics editing software developed and sold by Adobe Systems.”  (Incidentally, this suggests that an example of possible genericide long used in the Ginsburg et al. trademark casebook isn’t as existentially threatening as all that; see also the Xerox ad below futilely begging people not to use Xerox as a verb.)

Is LapLink giving Xerox and FedEx anything to worry about?

Take two trademarks and call me in the morning.

Verbification, however, isn’t inherently brand-protective: “[A] mark can also be used as a verb in an indiscriminate sense so as to refer to a category of activity in general, e.g., ‘I will PHOTOSHOP the image’ could be understood to mean image manipulation by using graphics editing software other than Adobe Photoshop.”  That usage doesn’t perform a trademark function.  “[I]nstead, it functions as a synecdoche describing both a particular species of activity (e.g. using Adobe’s PHOTOSHOP brand software) and the genus of services to which the species belongs (e.g. using image manipulation software in general).” 

Comment: I would disagree. As long as the user is aware that it’s a synecdoche, the distinction between mark and non-mark service remains, just as people are well aware that ranch workers are more than disembodied “hands.”  “Aspirin,” by contrast, is not a synecdoche; it’s the name of a product.  And the court immediately walks that back: “It cannot be understated that a mark is not rendered generic merely because the mark serves a synecdochian ‘dual function’ of identifying a particular species of service while at the same time indicating the genus of services to which the species belongs.”  Rather, the generic sense must be the term’s “principal” significance before genericity exists.  Moreover, per McCarthy, casual conversational uses aren’t evidence of generic use; primary significance is determined by use and understanding in the context of purchasing decisions. (Somewhat awkward for Google’s free service, then; someone who says “google it!” is in fact suggesting the equivalent of a purchase.)

Trademark law rewards effort in building a brand.  “It is thus contrary to both the letter and spirit of trademark law to strip a mark of legal protection solely because the mark—cultivated by diligent marketing, enforcement, and quality control—has become so strong and widespread that the public adopts the mark to describe that act of using the class of products or services to which the mark belongs.” The court cited Laura Heymann: “top-of-mind use of a trademark in its verb form, far from indicating the mark’s generic status, may well indicate the enduring fame of the brand.” Laura A. Heymann, The Grammar of Trademarks, 14 Lewis & Clark L.Rev. 1313 (2010).  This is especially true for arbitrary or fanciful marks. 

“If the primary significance of such a mark to a majority of the consuming public is to differentiate one service from the services of others, then the mark is not generic. This is true regardless of whether the public also uses the mark as an indiscriminate verb.”  As a result, the proper test is whether the primary significance of “google” to a majority of the public who performs searches on the internet is a reference to the Google search engine as opposed to a descriptive term for search engines in general.

Google’s expert, linguist Geoffrey Nunberg, opined that for some “highly distinctive and famous marks,” “the name of a particular product is used to convey the genus without actually denoting it.” His expert report supported Google’s position that figurative uses of marks can include verbifying them “to denote the characteristic action associated with the product or service they represent.” His examples were TiVo, Fed–Ex, Skype, and Google.  As he noted, “when somebody says, ‘I need the book tomorrow-can you Fed Ex it to me?’ we ordinarily assume that a shipment by UPS will be acceptable as well, without assuming that the verb to Fed–Ex simply means to ship by priority courier.”

Google’s survey expert, Gerald Ford, conducted a Teflon survey, first asking participants in a phone survey whether “Hewlett Packard” and “computer” were brands names or common names. All 420 respondents successfully identified “Hewlett Packard” as a brand name and “computer” as a common name.  They were then asked to classify six terms: STP; Coke; Jello; refrigerator; margarine; and aspirin.  They were told that “don’t know” or “no opinion” were acceptable; “both” answers were also recorded.  They were then asked to classify five more terms: browser; website; Amazon; Yahoo; and Google, specifically with respect to searching on the internet.  The 19 respondents who said they didn’t search on the internet were subsequently excluded.

Ultimately, 93.77% of internet searchers identified GOOGLE as a brand name and 5.25% identified GOOGLE as a common name.  The results were similar for YAHOO!, and both beat COKE (89.53% brand name, 6.73% common name).  (I understand the rhetorical reason for taking these results out to two decimal places, but I doubt it’s really justified.)  The only mark that beat Google was AMAZON, 96.51 % and 2.99%, respectively.  Ford calculated that, at a 95% confidence level, his results had an estimated error +/-2.37%.

Elliot objected that the study was irrelevant because it didn’t account for verb usage.  While the 9^th Circuit criticized Teflon surveys in its Anti-Monopoly decision, Congress then passed the Trademark Clarification Act of 1984 to overturn that decision and “clarify that a mark may have a ‘dual purpose’ of identifying goods and services and indicating the source of the goods and services.”  The survey was relevant to answer the question of the primary significance of the term “google.”

Elliot’s counsel used “Google Consumer Surveys” to create surveys that asked respondents to select one of three answers to the prompt: “I most often use the word google to mean.” The 1,033 responses for the first survey were: “to search something on the internet” (52.2%); “the name of a specific search engine” (28.7%); and “the internet (in general) (19.1%).” The 1,007 responses for the second survey were “to search something on the internet” (72%); “the name of a company” (11.5%); and “the internet (in general)” (16.6%).  Plaintiffs argued that these results were evidence of a predominant use of “google” as an indiscriminate verb meaning to search on the internet.

The court found that there was no evidence these surveys were conducted according to generally accepted principles, or analyzed with reliable statistical methods.  Counsel didn’t have adequate training to design a survey or to interpret survey results. Nunberg, who the court found qualified to opine about designing survey questions about the meanings of words, testified that he thought the two main surveys were “worthless” because asking “what does X mean to you” is “the vaguest possible question you can ask” and because the possible responses did not allow respondents to answer that the word “google” meant “to use the Google search engine.” Because neither the surveys nor the opinions derived therefrom met the threshold standard of reliability under the FRE, they were inadmissible.

Plaintiffs’ consumer survey expert, James Berger, conducted a “substantially modified” Thermos survey.  While the Teflon survey is more common, he used the Thermos version because it allowed him to test verb use.  (In a footnote, the court opined that there were doubtless better surveys for genericity than either, given that genericity is not binary as the primary sigificance test often frames it.  But of course advocates are conservative and only want to approve surveys that have a good chance of being admissible, a chance maximized by the form having actually been admitted in other cases.)  In the survey, 251 respondents were screened and then asked: “If you were going to ask a friend to search for something on the Internet, what word or phrase would you use to tell him/her what you want him/her to do?” Slightly over half of the validated respondents’ answers (129) contained the word google.

However, Berger testified that his survey did nothing to test whether consumers understand that “google” as a mark refers to one company (framed that way it’s tautological, though), and that it was not important to ask respondents about their understanding of the word google.  Indeed, he stated that his survey didn’t test the primary significance of “google” or whether the term was generic for search engine hardware and software.  Thermos surveys ordinarily ask several questions, but his survey asked only one substantive question.  The court found this opinion partially admissible (not the part where he opined about counsel’s surveys).  While there was no authority accepting a one-question Thermos survey, it wasn’t inadmissible junk science.

So, the heart of the matter: the primary significance of “google” to the consuming public.  First, Google got the strong presumption of validity conferred by its (incontestable) federal registrations, which includes the specific presumption of non-genericity.  Thus, Google met its initial burden of showing that plaintiffs hadn’t raised a genuine issue of material fact.  Plaintiffs were required to show specific facts from which a jury could find that GOOGLE was generic.

Relevant evidence includes dictionary usage; mark-holder usage; competitor usage; media usage; and consumer surveys.  Plaintiffs couldn’t find a dictionary whose definition of the word “google” neglected to mention the trademark significance of the term.  They attributed that to Google’s “intimidation.”  For example, Google asked wordspy.com to modify its definition of google (“To search for information on the Web, particularly by using the Google search engine”) to “take into account the trademark status of Google.” Likewise, Merriam-Webster stated that “we were trying to be as respectful as we possibly could be about Google’s trademark.” Viewing the dictionary evidence in the light most favorable to plaintiffs, it showed that “google” had meaning as an “indiscriminate” verb.

Mark-holder usage: Plaintiffs highlighted that Google’s co-founder Larry Page stated on July 8, 1998, “Have fun and keep googling,” and that entering the search query “define: google” into the Google search engine resulted in a verb definition of: “Use an internet search engine, particularly google.com.”  They also noted other domain names using “google” that they didn’t buy.  But Google also used GOOGLE as a mark in national ad campaigns; had policies establishing “strict standards” for third party uses; published rules and guidelines for use of the mark; and spent a lot of money policing use.  (It’s not a lot of money relative to how valuable the brand is (over $113 billion, estimated), but you don’t have to spend some arbitrary fraction of brand value on enforcement.)  Nonenforcement may be evidence of genericity, but on these facts it would be unreasonable to conclude that Google didn’t enforce its rights.

Competitors’ use: there was no evidence that competitors used the term in a non-trademark fashion.  Murphy Door Bed Co., Inc. v. Interior Sleep Sys., Inc., 874 F.2d 95 (2d Cir.1989), says that competitors’ non-use is not independently sufficient to prove non-genericness because threats of enforcement of the mark might deter use. But it’s still “peripheral[]” evidence that the mark is not generic. Competitors’ ability to accurately describe their products or services without using the term in question suggests the term is not generic. As a corollary, the existence of a short and simple descriptive term for the genus to which the trademarked species belongs also evidences the mark in question as not generic. “Internet search engine” is that term here.

Media use: plaintiffs argued that the media often used “google” as an indiscriminate verb.  Google noted that some of plaintiffs’ evidence also recognized the trademark significance of “google,” and plaintiffs didn’t find a major media outlet referring to a competing search engine as a “google.” This evidence showed that “google” is sometimes used as verb to mean search on the internet.

Consumer survey evidence confirmed that the word google is indeed used as a verb. Berger’s survey could be used to infer that a majority of the consuming public understands the verb to refer to the indiscriminate act of searching on the internet. But that wasn’t dispositive about primary significance; the Berger survey didn’t test the primary significance of “google.”

While summary judgment is disfavored in trademark cases (obligatory cite to 9^th Circuit precedent, arguably honored more in the breach in both the original and later senses of that phrase), it’s still appropriate when there’s no genuine issue of material fact. That was the sistuation here.  “Congress has spoken with particular clarity and force on the issue of whether a registered trademark is subject to cancellation as generic because it has more than one significance: ‘A registered mark shall not be deemed to be the generic name of goods or services solely because such mark is also used as a name of or to identify a unique product or service.’” Therefore, as a matter of law, a mark is not generic only because it simultaneously signifies more than just the trademarked product.  (That’s … actually not what the statutory language says at all, and it wasn’t what overruling anti-Monopoly was about.  Google’s search engine isn’t a unique product, which is why the verb use matters.)   

The court identified four possible meanings for “google”: (1) a trademark designating the Google search engine; (2) a verb referring to the act of searching on the internet using the Google search engine; (3) a verb referring to the act of searching on the internet using any search engine; and (4) a common descriptive term for search engines in general.  Cancellation would be appropriate “only if the fourth meaning is the primary significance of the word google to a majority of the consuming public.” 

But plaintiffs’ evidence only showed that (2) and (3) existed as meanings, and, drawing all reasonable inferences in their favor, that (3) was more significant than (2).  But the leap to “(3) is the most frequently used meaning” wasn’t justified, since plaintiffs had no evidence that verb use was more frequent than non-verb use.  And even if that leap were allowed, there was still no evidence to suggest that (4) was the primary significance of “google.”  It was undisputed that “well over 90% of the consuming public understands the word google with respect to searching on the internet as designating not to a common name, but a particular brand.” That fact established that (1) was more significant than (4) to a “vast majority” of the consuming public.  Even if a majority of the public understands (3) as a perfectly cromulent meaning, (1) and (3) weren’t mutually exclusive.  Summary judgment for Google on genericity.