FOR PUBLICATION UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT MULTI TIME MACHINE, INC., Plaintiff-Appellant, No. 13-55575 D.C. No. 2:11-cv-09076DDP-MAN v. AMAZON.COM, INC.; AMAZON SERVICES, LLC, Defendants-Appellees. OPINION Appeal from the United States District Court for the Central District of California Dean D. Pregerson, District Judge, Presiding Argued and Submitted April 9, 2015—Pasadena, California Filed July 6, 2015 Before: Barry G. Silverman and Carlos T. Bea, Circuit Judges and Gordon J. Quist,* Senior District Judge. Opinion by Judge Bea; Dissent by Judge Silverman * The Honorable Gordon J. Quist, Senior District Judge for the U.S. District Court for the Western District of Michigan, sitting by designation. 2 MULTI TIME MACHINE V. AMAZON.COM SUMMARY** Trademark The panel reversed the district court’s summary judgment in a trademark infringement action under the Lanham Act against online retailer Amazon.com. Multi Time Machine, Inc., manufacturer of MTM Special Ops watches, alleged that Amazon’s website infringed its trademark because of the manner in which the website responded to a shopper’s search request for the watches. The panel held that a jury could find that Amazon had created a likelihood of confusion under an “initial interest confusion” theory by responding to a search request with a page showing “MTM Special Ops” three times above a search result displaying similar watches manufactured by MTM’s competitors. Dissenting, Judge Silverman wrote that because Amazon’s search result clearly labeled the name and manufacturer of each product offered for sale and even included photographs of the items, no reasonably prudent shopper accustomed to shopping online would likely be confused as to the source of the products. ** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. MULTI TIME MACHINE V. AMAZON.COM 3 COUNSEL Eric Levinrad (argued) and Ryan Stonerock, Wolf, Rifkin, Shapiro, Schulman, & Rabkin, LLP, Los Angeles, California; Jeffrey Cohen, Millen, White, Zelano & Branigan, P.C., Arlington, Virginia, for Plaintiff-Appellant. Marc C. Levy (argued) and Kathryn Feiereisel, Faegre Baker Daniels LLP, Denver, Colorado, for Defendants-Appellees. OPINION BEA, Circuit Judge: We are called upon to determine whether the operation of a retailer’s website infringes a trademark because of the manner in which it responds to a shopper’s search request for the trademarked goods. What the website’s response states, together with what its response does not state, determines whether its response is likely to cause confusion. If confusion results from the website’s response, there may be trademark infringement. MTM Special Ops watches are high-end, military style watches manufactured by Multi-Time Machines, Inc. (“MTM”). Online retailer Amazon.com (“Amazon”) does not carry MTM watches. If her brother mentioned MTM Special Ops watches, a frequent Amazon shopper might try to purchase one for him through Amazon. If she were to enter “MTM Special Ops” as her search request on the Amazon website, Amazon would respond with its page showing MTM Special Ops (1) in the search field (2) “MTM Specials Ops” again—in quotation marks—immediately 4 MULTI TIME MACHINE V. AMAZON.COM below the search field and (3) yet again in the phrase “Related Searches: MTM special ops watch,” all before stating “Showing 10 Results.”1 What the website’s response will not state is that Amazon does not carry MTM products. Rather, below the search field, and below the second and third mentions of “MTM Special Ops” noted above, the site will display aesthetically similar, multi-function watches manufactured by MTM’s competitors. The shopper will see that Luminox and Chase-Durer watches are offered for sale, in response to her MTM query. MTM asserts the shopper might be confused into thinking a relationship exists between Luminox and MTM. As a result of this initial confusion, MTM asserts she might look into 1 Our recitation of the facts, and our decision, are based on the evidence submitted below. However, we may take judicial notice of facts which are publicly available and “not subject to reasonable dispute in that [they are] . . . capable of accurate and ready determination by resort to sources whose accuracy cannot reasonably be questioned.” See Daniels-Hall v. Nat’l Ed. Ass’n, 629 F.3d 992, 998–999 (9th Cir. 2010) (citing Fed. R. Ev. 201). Amazon’s website is such a source. As of June 26, 2015, Amazon hosts a static webpage which states that “At Amazon.com, we not only have a large collection of mtm special ops watch products [which, of course, is flatly untrue], but also a comprehensive set of reviews from our customers. Below we’ve selected a subset of mtm special ops watch products and the corresponding reviews to help you do better research, and choose the product that best suits your needs.” Amazon, http://www.amazon.com/gp/feature.html?ie=UTF8&docId=1001909381. As of the same date, when an Amazon shopper searches “mtm special ops,” under the search query playback he will see “9 results for ‘mtm special ops.’” Amazon, http://www.amazon.com/s/ref=nb_sb_noss_ 2?url=search-alias%3Daps&field-keywords=mtm+special+ops. Though unnecessary to our result, we think that a jury might find that these pages provide insight as to defendants’ intent to confuse (p. 18–19, infra) and thus give rise to an even greater likelihood of confusion than the earlier format used by Amazon. MULTI TIME MACHINE V. AMAZON.COM 5 buying a Luminox watch, rather than junk the quest altogether and seek to buy an MTM watch elsewhere. MTM asserts that Amazon’s use of MTM’s trademarked name is likely to confuse buyers, who may ultimately buy a competitor’s goods. The district court found Amazon’s use of MTM’s trademark created no likelihood of confusion as a matter of law. But we think a jury could find that Amazon has created a likelihood of confusion. We therefore reverse the district court’s grant of summary judgment in favor of Amazon. Facts and Procedural History Founded in 1992, MTM manufactures and markets watches under various brand names, including MTM, MTM Special Ops, and MTM Military Ops. MTM holds a registered trademark in “MTM SPECIAL OPS”2 for timepieces. MTM sells its watches directly to customers, through various distributors and retailers, and on military bases. MTM markets its watches to men 22–55 years of age who are drawn to rugged, military-style outdoor products. Thinking to cultivate and maintain an image as a high-end, exclusive brand, MTM decided not to sell its watches to Amazon for resale. Nor does MTM authorize its distributors to sell MTM watches on Amazon. MTM’s agreements with its distributors require them to seek MTM’s permission to sell MTM’s products anywhere but at their own retail sites. Amazon claims to offer “Earth’s Biggest Selection of products,” products which include watches manufactured by 2 We refer to the various capitalizations of the trademark as “MTM Special Ops” herein. 6 MULTI TIME MACHINE V. AMAZON.COM various competitors of MTM. Amazon users who search for “MTM Special Ops” on Amazon’s site are routed to a screen which shows the phrase “MTM Special Ops” in the query field (the “search query playback”); again immediately below as “MTM Special Ops” with quotation marks, directly below the search line; and immediately again after with the words “Related Searches.” After the three iterations of MTM’s trademark the screen lists search results, including watches manufactured by MTM’s competitors and listed by name. Customers cannot purchase the watches from the search results page, but must navigate to the “product detail” page by clicking on a particular search result. Once the customer has clicked on a particular result, he will see the particular product’s brand name and the product title, which also shows the brand name (e.g., Luminox). On the top of the product detail page, the customer’s initial inquiry, “MTM Special Ops,” will still appear in the search field. Nothing on either of the pages states that Amazon does not carry MTM products. Not so the websites of Amazon’s competitors Buy.com and Overstock.com. They clearly announce that no search results match the “MTM Special Ops” query and those websites do not route the visitor to a page with both MTM’s trademark “MTM Specials Ops” repeatedly at the top and competitors’ watches below. Their pages show the search query playback but then forthrightly state that no results for the “MTM Special Ops” search query were found, and then list competitors’ products. MTM’s competitors’ products appear in the Amazon search query response in part because Amazon’s search algorithm responds to its customers’ behavior using a Behavior Based Search technology (“BBS”), which uses data about what customers view and purchase after searching certain terms. Amazon does not program the terms; the MULTI TIME MACHINE V. AMAZON.COM 7 function responds solely to customer behavior. If enough customers search for a certain keyword, “X,” and then look at or purchase another product “Y,” even if X and Y are not obviously related, future customers who search for X may receive search results including Y. But the BBS function is not solely responsible for the search results. The results list also includes matches based on a search of terms on Amazon’s pages—for instance, streaming video of a show called Special Ops Mission may be called up. Whether a particular result appears because of BBS or a traditional search of matching terms is not evident from the matches, and the relevant products (which are based on search terms) and recommended products (based on BBS) are mingled together. MTM sued Amazon, alleging that Amazon had infringed MTM’s trademarks in violation of the Lanham Act. MTM sought injunctive relief barring use of the trademark and damages. On Amazon’s motion, the district court granted summary judgment to Amazon. MTM timely appealed. Standard of Review We review de novo the district court’s grant of summary judgment, and must consider the evidence in the light most favorable to the nonmoving party. Surfvivor Media, Inc. v. Survivor Prods., 406 F.3d 625, 630 (9th Cir. 2005). This court may affirm the grant of summary judgment on any ground that finds support in the record. Karl Storz Endoscopy Am., Inc. v. Surgical Technologies, Inc., 285 F.3d 848, 855 (9th Cir. 2002). 8 MULTI TIME MACHINE V. AMAZON.COM Discussion Under the Lanham Act, a defendant infringes a trademark when the defendant uses the mark in commerce in a manner likely to cause confusion as to a good’s source. Trademark infringement also occurs when the trademark’s use in commerce is likely to cause confusion as to the affiliation, association, or approval of the trademark holder with the trademark user.3 A defendant who infringes another’s trademark is liable for damages and subject to injunction. 15 U.S.C. §§ 1114(1)(a), 1125(a)(1). Put another way, a defendant who creates likelihood of confusion by using another’s mark has infringed the mark. Playboy Enterprises, Inc. v. Netscape Commc’ns Corp., 354 F.3d 1020, 1024 (9th Cir. 2004).4 3 “[A]ny person who shall, without consent of the registrant, “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” is liable for damages, and the registrant may be entitled to injunctive relief. 15 U.S.C. § 1114(1)(a). Any person who “uses in commerce any word, term, name, symbol, or device . . . which is likely to cause confusion . . . as to the affiliation, connection, or association of such person with another person, or as to the origin, sponsorship, or approval of his or her goods, services of commercial activities” is also subject to injunction and liable for damages to one likely to be damaged. 15 U.S.C. § 1125(a)(1). 4 In Playboy, defendant search engine Netscape required adult-oriented websites who purchased advertising from Netscape to link their advertisements to search keywords “playboy” and “playmate,” trademarks of Playboy Enterprises (“PEI”). 354 F.3d at 1023. When Netscape users searched the terms, they would be presented with banner ads for non-PEI operated websites that said “click here” and routed users to advertisers’ websites. Id. PEI sued Netscape for Lanham Act violations, and the district court granted summary judgment in favor of defendants. Id. This MULTI TIME MACHINE V. AMAZON.COM 9 A defendant can create likelihood of confusion, and thereby infringe the trademark, through a type of confusion referred to as “initial interest confusion.” Initial interest confusion occurs not where a customer is confused about the source of a product at the time of purchase, but earlier in the shopping process, if “customer confusion . . . creates initial interest in a competitor’s product.” Id. at 1025. Even if that confusion is dispelled before an actual sale occurs, initial interest confusion still constitutes trademark infringement because it “impermissibly capitalizes on the goodwill associated with a mark and is therefore actionable trademark infringement.” Id. A. Likelihood of Confusion This court considers eight non-exhaustive factors, known as the Sleekcraft factors, to determine whether a trademark use gives rise to a likelihood of confusion: (1) strength of the mark(s); (2) proximity or relatedness of the goods; (3) similarity of the marks; (4) evidence of actual confusion; (5) marketing channels; (6) degree of consumer care; (7) the defendants’ intent; and (8) likelihood of expansion. Network Automation, Inc., v. Advanced Sys. Concepts, Inc., 638 F.3d 1137, 1145 (9th Cir. 2011) (citing AMF Inc. v. Sleekcraft Boats, 599 F.2d 341, 348–49 (9th Cir. 1979)). As the Network Automation court explained, in the context of internet commerce, though likelihood of confusion might be shown where “consumers saw banner advertisements that were ‘confusingly labeled or not labeled at all’. . . .clear labeling ‘might eliminate the likelihood of confusion.’” court reversed. We found that initial interest confusion supported an infringement theory even if users realized they were not at a PEI site before making a purchase. Id. at 1025, 1026–29. 10 MULTI TIME MACHINE V. AMAZON.COM Network Automation, 638 F.3d at 1153–54 (citing Playboy, 354 F.3d at 1023, 1030 n.43).5 Network Automation and Playboy addressed the unauthorized use of a trademark to sell advertising keywords 5 In Network Automation, plaintiff Network Automation and defendant Advanced Systems Concepts both sold job scheduling and management software. 638 F.3d at 1142. Network Automation advertised its product by purchasing certain keywords—including registered trademarks belonging to Advanced Systems—which, when typed into various search engines, produced a results page that included Network Automation’s website “www.NetworkAutomation.com” as a labeled, sponsored link. Id. Advanced Systems alleged violation of the Lanham Act and moved for a preliminary injunction. Id. at 1143. The district court granted a preliminary injunction to Advanced Systems, and Network Automation appealed. Id. On appeal, this court reversed and vacated the preliminary injunction. This court considered the eight Sleekcraft factors and held that the district court had not weighed the factors flexibly and that “[b]ecause the linchpin of trademark infringement is consumer confusion, the district court abused its discretion in entering the injunction.” Id. at 1154. The court held that because “the sine qua non of trademark infringement is consumer confusion, when we examine initial interest confusion, the owner of the mark must demonstrate likely confusion, not mere diversion.” Id. at 1149. Amazon contends that because “mere diversion” does not constitute initial interest confusion, the doctrine is inapplicable to the internet. However, whether customers are merely diverted is a question of fact. This court properly considered whether the facts favored Advanced Systems in Network Automation because a preliminary injunction requires “the moving party [there, the plaintiff alleging infringement] demonstrate a fair chance of success on the merits or questions serious enough to require litigation.” Arc of Cal. v. Douglas, 757 F.3d 975, 993 (9th Cir. 2014). Therefore, the Network Automation court properly considered the weight of the evidence to decide whether Advanced Systems had a fair chance of success on the merits. Here, we are not tasked to determine whether MTM is likely to succeed, nor to consider the weight of the evidence. As this is an appeal from a summary judgment, we must decide whether there is a genuine triable issue of material fact. MULTI TIME MACHINE V. AMAZON.COM 11 to search engines, not the use of a trademark to sell competitors’ products. However, we think that in the sale context, just as the “labeling and appearance of the advertisements as they appear on the result page includes more than the text of the advertisement, and must be considered as a whole,” here the labeling of search results which feature competitors’ products is important. Network Automation, 638 F.3d at 1154. Because of its importance, we first address labeling—which we find gives rise to a genuine issue of fact—and then turn to the traditional Sleekcraft factors. 1. Labeling MTM submitted an expert report that stated that the search results on Amazon are “ambiguous, misleading, and confusing.” Dist. Ct. Order, 926 F.Supp.2d 1130, 1141 (N.D. Cal. Feb. 20, 2015). The district court found that the expert analysis showed only that customers could be confused about why they receive certain search results, but that there was no evidence that Amazon users were likely to be confused as to the source who manufactured the competing goods We disagree. A jury could infer that users who are confused by the search results are confused as to why MTM products are not listed. Unlike its competitors Buy.com and Overstock.com, Amazon does not forestall any confusion by informing customers who are searching “MTM Special Ops” that Amazon does not carry any such products. A jury could infer that users who are confused by the search result will wonder whether a competitor has acquired MTM or is otherwise affiliated with or approved by MTM. See Brookfield Commc’ns, Inc. v. West Coast Entm’t Corp., 174 F.3d 1036, 1057 (9th Cir. 1999). This is especially true 12 MULTI TIME MACHINE V. AMAZON.COM as to a brand like MTM, as many luxury brands with distinct marks are produced by manufacturers of lower-priced, betterknown brands—just as Honda manufactures Acura automobiles but sells Acura automobiles under a distinct mark that is marketed to wealthier purchasers, and Timex manufactures watches for luxury fashion houses Versace and Salvatore Ferragamo. Like MTM, Luminox manufactures luxury watches, and a customer might think that MTM and Luminox are manufactured by the same parent company. The possibility of initial interest confusion here is likely much higher than if, for instance, a customer using an online grocery website typed “Coke” and only Pepsi products were returned as results. No shopper would think that Pepsi was simply a higher end version of Coke, or that Pepsi had acquired Coke’s secret recipe and started selling it under the Pepsi mark.6 In any event, even as to expensive goods—for instance, pianos sold under a mark very similar to the famous Steinway and Sons brand’s mark—the issue is not that a buyer might buy a piano manufactured by someone other than Steinway thinking that it was a Steinway. The issue is that the defendant’s use of the mark would cause initial interest 6 The dissent also mentions Coke and Pepsi in conjunction with the labeling inquiry, and John Belushi’s Saturday Night Live “cheezborger” refrain—“No Coke. Pepsi.” However, Belushi’s line is analogous to the message on Overstock’s and Buy’s websites, which state the equivalent of “No Coke” rather than simply inundating the shopper with images of Pepsi. The dissent acknowledges that a retailer who offers competitors’ products for sale, without mentioning that he does not carry a brand requested by a customer, is “sort of like what happens when you order a Coke, and are clearly told that they only have Pepsi.” Dissent at 23. But it is only sort of like the Belushi scenario, because unlike Belushi’s “No Coke,” Amazon does not say “No MTM.” MULTI TIME MACHINE V. AMAZON.COM 13 confusion by attracting potential customers’ attention to buy the infringing goods because of the trademark holder’s hardwon reputation. Brookfield, 174 F.3d at 1063 (citing Grotrian, Helfferich, Schulz, Th. Steinweg Nachf. v. Steinway & Sons, 523 F.2d 1331, 1341–42 (2d. Cir. 1975)). A jury could infer that the labeling of the search results, and Amazon’s failure to notify customers that it does not have results that match MTM’s mark, give rise to initial interest confusion. If so, a jury might find that Amazon customers searching for MTM products are subject to more than mere diversion, since MTM is not required to show that customers are likely to be confused at the point of sale. Playboy, 354 F.3d at 1025. We agree with the district court’s conclusion that the product details for competitors’ itemized products were clearly labeled, but we find that the clarity of the search results page at issue is open to dispute. We must not substitute our determination of what constitutes clear labeling, nor its importance, for that of a jury. 2. Sleekcraft Factors The Sleekcraft “eight-factor test for likelihood of confusion is pliant. Some factors are much more important than others, and the relative importance of each individual factor will be case-specific.” Brookfield, 174 F.3d at 1054. On a motion for summary judgment, courts may consider whether any of the Sleekcraft factors give rise to a genuine issue of fact. Fortune Dynamic, Inc. v. Victoria’s Secret Stores Brand Mgmt., Inc., 618 F.3d 1025, 1031 (9th Cir. 2010). A court may be “far from certain that consumers were likely to be confused [and still be] confident that the question 14 MULTI TIME MACHINE V. AMAZON.COM is close enough that it should be answered as a matter of fact by a jury, not as a matter of law by a court.” Id. If there is a genuine issue of fact as to any of the factors, there is more likely to be a genuine issue of fact as to whether there is likelihood of confusion. To avoid summary judgment a plaintiff need not show that every factor weighs in his favor, only to make a strong showing as to some of them (though we note that, because the factors are not exhaustive, a plaintiff could presumably survive summary judgment by adducing evidence of likelihood of confusion that did not fall within one of the factors). Surfvivor, 406 F.3d at 630. Mindful that our analysis must be tailored to this case, we discuss below the five Sleekcraft factors we deem relevant to resolution of the question of summary judgment: the strength of the mark, relatedness/proximity of the goods, evidence of actual confusion, defendant’s intent, and the degree of care exercised by purchasers. Three of the Sleekcraft factors are irrelevant: similarity of marks, marketing channels, and likelihood of expansion. As to similarity of marks, Amazon is using MTM’s mark, not another mark in its display of “search results.” As to marketing channels, both MTM and Amazon sell watches on the internet, which is too widespread a market to affect the likelihood of confusion among customers. The possibility of expansion is irrelevant since MTM and Amazon both already sell high-end timepieces, as discussed below under the heading “relatedness of the goods.” We find that three of the remaining five relevant factors appear to weigh in favor of a finding of a likelihood of confusion, and we address these first. We hold that it is the province of a jury to determine how heavily each of these factors should weigh. MULTI TIME MACHINE V. AMAZON.COM 15 a. Strength of the Mark A mark’s strength is a measure of how uniquely identified it is with a product or service, and therefore how deserving of trademark protection. Fortune Dynamic, 618 F.3d at 1032. Two types of trademark strength are relevant: commercial and conceptual. Commercial strength refers to a mark’s recognition in the marketplace, that is, how widely recognized the mark is by customers. Id. at 1034. Neither MTM nor Amazon presented evidence of MTM’s commercial strength. We do not consider it; we turn to consideration of conceptual strength. “A mark’s conceptual strength depends largely on the obviousness of its connection to the good or service to which it refers. The less obvious the connection, the stronger the mark, and vice versa.” Id. at 1032–33. Conceptual strength is considered along a continuum, and in this circuit, marks may be classified as falling into one of five categories, from conceptually weak to conceptually strong: generic, descriptive, suggestive, arbitrary, or fanciful. Id. at 1033. Whether a mark is descriptive or suggestive is a question of fact. Id. at 1034. In an infringement suit, “the distinction [between a descriptive and suggestive mark] is important . . . because if the mark is suggestive, there is a stronger likelihood that a jury could reasonably conclude that the ‘strength of the mark’ factor favors the [plaintiff].” Id. Here, the district court found that “Amazon’s evidence is persuasive in showing that the marks are not strong; they are at best suggestive, and more likely descriptive.” 926 F.Supp.2d at 1139. However, the phrase “MTM Special Ops” requires “a mental leap from the mark to the product,” because the phrase does not expressly refer to watches. Fortune Dynamic, 618 F.3d at 1034. Indeed, by evoking elite military forces 16 MULTI TIME MACHINE V. AMAZON.COM (“Special Ops”), the goods suggested by the phrase are as likely to be protective gear, binoculars, weapons, or boots as they are watches. A jury could find that the mark is suggestive and conceptually strong because it does not obviously refer to watches, or that it is merely descriptive because the watches are made in a military style. Either way, the weight of the evidence is a question of fact, and there is a genuine issue of fact as to the conceptual strength of the mark. As in Fortune Dynamic, “a jury should assess the conceptual strength of [plaintiff’s] mark in the first instance.” 618 F.3d at 1033. b. Similarity of the Goods Like MTM, Amazon sells specialized, military-style watches. The similarity of the goods Amazon is selling weighs in favor of a finding of infringement. The district court cited Network Automation for the proposition that “though the products were interchangeable . . . that fact would ‘become less important if advertisements are clearly labeled or consumers exercise a high degree of care, because rather than being misled, the consumer would merely be confronted with choices among similar products.’” 926 F.Supp.2d at 1137 (citing Network Automation, 638 F.3d at 1150). The district court found “the same is true in this case; although Amazon and MTM both sell watches, which are identical products, this is misleading only if the consumer is confused, not if the consumer simply has clearly marked options.” Id. This conclusion assumes that Amazon customers will not be confused, and that the options are clearly marked, which are questions of fact as to which both parties submitted evidence. The facts of this case are distinguishable from Network Automation, where the claimant trademark holder’s products were displayed MULTI TIME MACHINE V. AMAZON.COM 17 alongside the alleged infringers’ products, thereby presenting “clearly marked options.” MTM watches are not displayed at all on the Amazon website. Whether customers will believe the options on Amazon’s page, which do not include MTM products, are clearly marked as having no association with, or approval by, MTM, and whether they will be confused, is an open question, and its answer does not render the identity of the goods here moot. Rather, a jury could find that it weighs in favor of finding likelihood of confusion. On summary judgment, the court may not make assumptions about the sophistication of would-be purchasers. Fortune Dynamic, 618 F.3d at 1030. Some members of MTM’s target demographic, men of 22–55 years of age who like military-styled, rugged products, may not be frequent internet shoppers. Such purchasers “may incorrectly believe that [defendant] licensed [the mark] from [plaintiff] . . . . Other consumers may simply believe that [defendant or the manufacturers it features] bought out [plaintiff], or that they are related companies.” Brookfield Commc’n, Inc. v. West Coast Entm’t Corp., 174 F.3d 1036, 1057 (9th Cir. 1999). This is especially possible here because Amazon touts itself as offering “Earth’s Biggest Selection of products,” and, as noted above, manufacturers sometimes market luxury brands under distinct marks. Even if further internet research could clarify the matter for a customer who wondered if MTM had been acquired or had acquired its competitor watch-makers, it is incorrect to conclude that “likelihood of confusion exists only when consumers are confused as to the source of a product they actually purchase. It is . . . well established that the Lanham Act protects against many other forms of confusion.” Brookfield, 174 F.3d at 1057. Network Automation found that on the internet, initial interest confusion is an untenable theory where sponsored links 18 MULTI TIME MACHINE V. AMAZON.COM appear on search pages that have “partitioned their search results pages so that the advertisements appear in separately labeled sections for ‘sponsored’ links.’” 638 F.3d at 1154. Here, the competitors’ products are not clearly labeled as being BBS results rather than keyword searches. The similarity of the goods means that an Amazon customer who searches for “MTM Special Ops” and then investigates watches manufactured by Luminox or Chase-Durer, even if he later purchases such a watch without any confusion as to its source, will have been subject to “confusion, not mere diversion.”7 Network Automation, 638 F.3d at 1149. Even though his confusion may be “dispelled before an actual sale occurs, initial interest confusion impermissibly capitalizes on the goodwill associated with a mark and is therefore actionable trademark infringement.” Playboy, 354 F.3d at 1025. Therefore, the similarity of goods weighs in favor of MTM, and a jury should determine just how much it weighs in favor of MTM. c. Defendant’s intent “A defendant’s intent to confuse constitutes probative evidence of likely confusion.” Playboy, 354 F.3d at 1028. MTM submitted evidence that Amazon vendors and customers had complained to Amazon because they did not understand why they received certain non-responsive search results when they searched for products that are not carried by Amazon. The evidence showed that Amazon employees did not take action to address the complaints by explaining to the public how the BBS function works. One Amazon employee noted that explaining BBS to the public might draw 7 Amazon’s evidence that customers do not purchase competitors’ watches after searching “MTM SPECIAL OPS” is addressed below. MULTI TIME MACHINE V. AMAZON.COM 19 customers’ and vendors’ unwanted scrutiny to the matter. Amazon did not disclose to shoppers how its BBS worked. As in Playboy, this evidence suggests, “at a minimum, that defendants do nothing to alleviate confusion. . . . Although not definitive, this factor provides some evidence of an intent to confuse on the part of defendants.” Playboy, 354 F.3d at 1029. From evidence that “Earth’s most customer-centric company” took no action on these complaints, a jury could infer that Amazon intended to confuse its customers. We leave it to a jury to determine, if Amazon so intended, how important that intent is, and we turn to two factors that we think weigh in favor of Amazon. d. Evidence of Actual Confusion Where evidence of actual confusion is submitted, it is “strong support for the likelihood of confusion.” Network Automation, 638 F.3d at 1151 (internal quotation marks omitted). But actual confusion “is not necessary to a finding of likelihood of confusion under the Lanham Act. Indeed, proving actual confusion is difficult . . . and the courts have often discounted such evidence because it was unclear or insubstantial.” Id. (internal quotation marks, citation, and brackets omitted). MTM did not submit colorable evidence of actual confusion. MTM offered its president’s testimony that he had knowledge of actual confusion. The district court found this testimony was too vague to constitute evidence. The president testified that someone he had met named Eric said, in reference to Amazon’s page, “it’s confusing.” Such testimony does not suffice to give rise to a genuine issue of fact, even were such hearsay admissible, as the record does 20 MULTI TIME MACHINE V. AMAZON.COM not provide support for the speculation that Eric was a potential customer. Amazon submitted evidence that purports to show that no customers were confused, because customers who searched for “Luminox” were 21 times as likely to purchase a Luminox watch as were customers who searched for “MTM Special Ops.” We do not find it surprising that customers who search for an item (Luminox watches) are more likely to buy that item than customers who did not search for it but searched for another product (MTM watches). But in the absence of evidence of actual confusion, we agree that the factor weighs in favor of Amazon. However, we are not persuaded that a jury could not view this purported evidence of no actual confusion as flawed because a user researching watches might initially be confused about the availability of MTM watches online and so not purchase a Luminox the same day.8 Further, some users did search for “MTM Special Ops” and purchase a competitor’s watch the same day, which a jury could find probative of some confusion. 8 In response to MTM’s critique that the data did not fully account for consumer behavior, the district court opined that “Amazon persuasively responds that the value of the data is not absolute but relative; there is no reason to think that those consumers searching for Luminox would exhibit different behaviors from those searching for MTM Special Ops.” 926 F.Supp.2d at 1140. However, the very relativity of the data makes its value a question for a jury, who might determine that Luminox customers on Amazon are different from would-be MTM Special Ops purchasers: Luminox customers make same-day purchases because the product they sought is available on Amazon. MTM Special Ops customers may wait a few days to buy a Luminox watch because it is not what they sought, but their interest in a Luminox watch was piqued because they were uncertain whether or how Luminox is affiliated with or approved by MTM. MULTI TIME MACHINE V. AMAZON.COM 21 e. Degree of Care As to the degree of care expected of a purchaser, when goods are expensive, purchasers can be expected to exercise greater care, though confusion may still be likely. Network Automation, 638 F.3d at 1152. MTM’s watches are priced between several hundred dollars to two thousand dollars. The district court did not err in finding that consumers could be presumed to use a high degree of care in purchasing such watches. However, in light of our determination that other factors give rise to genuine issues of fact, we note that a jury may find that Amazon presented evidence that “same day sales” are high for Luminox. This could be interpreted, by a jury, as proof that at least some persons who seek military watches are impulse buyers who do not spend as much time comparing products as careful buyers might. If so, the jury might accord the price of the watches little weight compared to the other factors. This factor and its relative importance are matters for a jury. B. Use in Commerce This court has held that use of a trademark as a search engine keyword that triggers the display of a competitor’s advertisement is a “use in commerce” under the Lanham Act. Network Automation, 638 F.3d at 1144–45. Amazon contends that the user-generated search term “MTM Special Ops” is not a use in commerce within the meaning of the Lanham Act. We hold that the customer-generated use of a trademark in the retail search context is a use in commerce. As the district court correctly observed, though Network Automation is distinguishable because the search engines were selling the use of competitor’s trademarks, Amazon’s purpose is not less commercial just because it is selling 22 MULTI TIME MACHINE V. AMAZON.COM wares, not advertising space. Therefore, we decline to affirm the district court on the alternative ground that Amazon’s use is not a use in commerce. Conclusion We are by no means certain that MTM will be able to prove likelihood of confusion under an initial interest confusion theory, but we are confident the matter can be determined only by resolving genuine issues of material fact. REVERSED AND REMANDED. SILVERMAN, Circuit Judge, dissenting: Live! From New York! It’s Saturday Night! . . . and the scene is the Olympia Restaurant, Chicago, January, 1978. Dan Aykroyd is manning the grill, Bill Murray is working prep, and John Belushi is up front taking orders. A customer, Jane Curtin, walks in and orders two cheeseburgers. Belushi yells to the grill: “Cheezborger, cheezborger.” Curtin then orders a Coke. Without looking up, Belushi replies: “No Coke. Pepsi.” Pause it right there. Would anyone seriously contend that the diner violated Coke’s trademark by responding to the customer’s order that it doesn’t carry Coke, only Pepsi? Now, fast-forward to the present. A customer goes online to Amazon.com looking for a certain military-style MULTI TIME MACHINE V. AMAZON.COM 23 wristwatch – specifically the “MTM Special Ops” – marketed and manufactured by Plaintiff Multi Time Machine, Inc. The customer types “mtm special ops” in the search box and presses “enter.” Because Amazon does not sell the MTM Special Ops watch, what the search produces is a list, with photographs, of several other brands of military style watches that Amazon does carry, specifically identified by their brand names – Luminox, Chase-Durer, TAWATEC, and Modus – sort of like what happens when you order a Coke, and are clearly told that they only have Pepsi. The particular search results page at issue is displayed below: 24 MULTI TIME MACHINE V. AMAZON.COM Case: 13-55575, 11/26/2013, ID: 8878836, DktEntry: 17-2, Page 13 of 37 -10- 10 MULTI TIME MACHINE V. AMAZON.COM 25 MTM brought suit alleging that Amazon’s response to a search for the MTM Special Ops watch on its website is trademark infringement in violation of the Lanham Act. MTM contends that Amazon’s search results page creates a likelihood of confusion, even though there is no evidence of any actual confusion and even though the other brands are clearly identified by name and each product is displayed with a photograph. The district court granted summary judgment in favor of Amazon. I would affirm. “The core element of trademark infringement” is whether the defendant’s conduct “is likely to confuse customers about the source of the products.” E. & J. Gallo Winery v. Gallo Cattle Co., 967 F.2d 1280, 1290 (9th Cir. 1992). Because Amazon’s search results page clearly labels the name and manufacturer of each product offered for sale and even includes photographs of the items, no reasonably prudent consumer accustomed to shopping online would likely be confused as to the source of the products. Thus, summary judgment of MTM’s trademark claims was proper. I. “Although disfavored in trademark infringement cases, summary judgment may be entered when no genuine issue of material fact exists.” Id. Indeed, in several trademark cases, we have concluded that there is no likelihood of confusion as a matter of law, and affirmed the district court’s grant of summary judgment in favor of the defendant. See, e.g., One Indus., LLC v. Jim O’Neal Distrib., 578 F.3d 1154, 1162–65 (9th Cir. 2009); M2 Software, Inc. v. Madacy Entm’t, 421 F.3d 1073, 1080–85 (9th Cir. 2005); Surfvivor Media, Inc. v. Survivor Prods., 406 F.3d 625, 631–34 (9th Cir. 2005). 26 MULTI TIME MACHINE V. AMAZON.COM To prevail on a claim of trademark infringement under the Lanham Act, “a trademark holder must show that the defendant’s use of its trademark ‘is likely to cause confusion, or to cause mistake, or to deceive.’” Fortune Dynamic, Inc. v. Victoria’s Secret Stores Brand Mgmt., 618 F.3d 1025, 1030 (9th Cir. 2010) (quoting 15 U.S.C. § 1125(a)(1)-(a)(1)(A)). “The test for likelihood of confusion is whether a ‘reasonably prudent consumer’ in the marketplace is likely to be confused as to the origin of the good or service bearing one of the marks.” Dreamwerks Prod. Group v. SKG Studio, 142 F.3d 1127, 1129 (9th Cir. 1998). “The confusion must ‘be probable, not simply a possibility.’” Murray v. Cable NBC, 86 F.3d 858, 861 (9th Cir. 1996). Here, the district court was correct in ruling that there is no likelihood of confusion. Amazon is responding to a customer’s inquiry about a brand it does not carry by doing no more than stating clearly (and showing pictures!) of what brands it does carry. To whatever extent the Sleekcraft factors1 apply in a case such as this – a merchant responding to a request for a particular brand it does not sell by offering other brands clearly identified as such – the undisputed evidence shows that confusion on the part of the inquiring buyer is not at all likely. Not only are the other brands clearly labeled and accompanied by a photograph, there is no evidence of actual confusion by anyone. 1 The eight-factor test from our decision in AMF Inc. v. Sleekcraft Boats, 599 F.2d 341, 348-49 (9th Cir. 1979), abrogated in part on other grounds as recognized in Mattel, Inc. v. Walking Mountain Prods., 353 F.3d 792 (9th Cir. 2003). MULTI TIME MACHINE V. AMAZON.COM 27 To analyze likelihood of confusion, we utilize the eightfactor test set forth in Sleekcraft.2 However, “[w]e have long cautioned that applying the Sleekcraft test is not like counting beans.” One Indus., 578 F.3d at 1162; see also Network Automation, Inc. v. Advanced Sys. Concepts, 638 F.3d 1137, 1145 (9th Cir. 2011) (“The Sleekcraft factors are intended as an adaptable proxy for consumer confusion, not a rote checklist.”). “Some factors are much more important than others, and the relative importance of each individual factor will be case-specific.” Brookfield, 174 F.3d at 1054. Moreover, the Sleekcraft factors are not exhaustive and other variables may come into play depending on the particular facts presented. Network Automation, 638 F.3d at 1145–46. This is particularly true in the Internet context. See Brookfield, 174 F.3d at 1054 (“We must be acutely aware of excessive rigidity when applying the law in the Internet context; emerging technologies require a flexible approach.”). Indeed, in evaluating claims of trademark infringement in cases involving Internet search engines, we have found particularly important an additional factor that is outside of the Sleekcraft test: “the labeling and appearance of the advertisements and the surrounding context on the screen displaying the results page.” Network Automation, 638 F.3d at 1154. In the present case, the eight-factor Sleekcraft test is not particularly apt. This is not surprising as the Sleekcraft test 2 The eight Sleekcraft factors are: “1. strength of the mark; 2. proximity of the goods; 3. similarity of the marks; 4. evidence of actual confusion; 5. marketing channels used; 6. type of goods and the degree of care likely to be exercised by the purchaser; 7. defendant’s intent in selecting the mark; and 8. likelihood of expansion of the product lines.” 599 F.2d at 348–49. 28 MULTI TIME MACHINE V. AMAZON.COM was developed for a different problem – i.e., for analyzing whether two competing brands’ marks are sufficiently similar to cause consumer confusion. See Sleekcraft, 599 F.2d at 348. Although the present case involves brands that compete with MTM, such as Luminox, Chase-Durer, TAWATEC, and Modus, MTM does not contend that the marks for these competing brands are similar to its trademarks. Rather, MTM argues that the design of Amazon’s search results page creates a likelihood of initial interest confusion because when a customer searches for MTM Special Ops watches on Amazon.com, the search results page displays the search term used – here, “mtm special ops” – followed by a display of numerous watches manufactured by MTM’s competitors and offered for sale by Amazon, without explicitly informing the customer that Amazon does not carry MTM watches. Thus, the present case focuses on a different type of confusion than was at issue in Sleekcraft. Here, the confusion is not caused by the design of the competitor’s mark, but by the design of the web page that is displaying the competing marks and offering the competing products for sale. Sleekcraft aside, the ultimate test for determining likelihood of confusion is whether a “reasonably prudent consumer” in the marketplace is likely to be confused as to the origin of the goods. Dreamwerks, 142 F.3d at 1129. Our case can be resolved simply by an evaluation of the web page at issue and the relevant consumer. Cf. Brookfield, 174 F.3d at 1054 (“[I]t is often possible to reach a conclusion with respect to likelihood of confusion after considering only a subset of the factors.”). Indeed, we have previously noted that “[i]n the keyword advertising context [i.e., where a user performs a search on the internet, and based on the keywords contained in the search, the resulting web page displays certain advertisements containing products or services for sale,] the MULTI TIME MACHINE V. AMAZON.COM 29 ‘likelihood of confusion will ultimately turn on what the consumer saw on the screen and reasonably believed, given the context.’” Network Automation, 638 F.3d at 1153. In other words, the case will turn on the answers to the following two questions: (1) Who is the relevant reasonable consumer?; and (2) What would he reasonably believe based on what he saw on the screen? Turning to the first question, we have explained that “[t]he nature of the goods and the type of consumer is highly relevant to determining the likelihood of confusion in the keyword advertising context.” Network Automation, 638 F.3d at 1152. “In evaluating this factor, we consider ‘the typical buyer exercising ordinary caution.’” Au-Tomotive Gold, Inc. v. Volkswagen of Am., Inc., 457 F.3d 1062, 1076 (9th Cir. 2006) (quoting Sleekcraft, 599 F.2d at 353). “Confusion is less likely where buyers exercise care and precision in their purchases, such as for expensive or sophisticated items.” Id. Moreover, “the default degree of consumer care is becoming more heightened as the novelty of the Internet evaporates and online commerce becomes commonplace.” Network Automation, 638 F.3d at 1152. The goods in the present case are expensive. It is undisputed that the watches at issue sell for several hundred dollars. Therefore, the relevant consumer in the present case “is a reasonably prudent consumer accustomed to shopping online.” Toyota Motor Sales, U.S.A., Inc. v. Tabari, 610 F.3d 1171, 1176 (9th Cir. 2010). Turning to the second question, as MTM itself asserts, the labeling and appearance of the products for sale on Amazon’s web page is the most important factor in this case. This is because we have previously noted that clear labeling can 30 MULTI TIME MACHINE V. AMAZON.COM eliminate the likelihood of initial interest confusion in cases involving Internet search terms. See, e.g., Playboy Enters. v. Netscape Communs. Corp., 354 F.3d 1020, 1030 n.44 (9th Cir. 2004) (explaining that clear labeling “might eliminate the likelihood of initial interest confusion that exists in this case”); Network Automation, 638 F.3d at 1154 (same). Indeed, MTM itself argues: “The common thread of [the Ninth Circuit’s decisions in Brookfield, Playboy, and Network Automation] is that liability under the Lanham Act can only be avoided as a matter of law where there is clear labeling to avoid the possibility of confusion – including initial interest confusion – resulting from the use of another’s trademark.” Thus, MTM agrees that summary judgment of its trademark claims is appropriate if there is clear labeling that avoids likely confusion. Here, the products at issue are clearly labeled by Amazon to avoid any likelihood of initial interest confusion by a reasonably prudent consumer accustomed to online shopping. When a shopper goes to Amazon’s website and searches for a product using MTM’s trademark “mtm special ops,” the resulting page displays several products, all of which are clearly labeled with the product’s name and manufacturer in large, bright, bold letters and includes a photograph of the item. In fact, the manufacturer’s name is listed twice. For example, the first result is “Luminox Men’s 8401 Black Ops Watch by Luminox.” The second result is “Chase-Durer Men’s 246.4BB7-XL-BR Special Forces 1000XL Black Ionic-Plated Underwater Demolition Team Watch by Chase-Durer.” Because Amazon clearly labels each of the products for sale by brand name and model number accompanied by a photograph of the item, it is simply bizarre to suppose that a reasonably prudent consumer accustomed to MULTI TIME MACHINE V. AMAZON.COM 31 online shopping would be confused about the source of the goods. MTM argues that initial interest confusion might occur because Amazon lists the search term used – here the trademarked phrase “mtm special ops” – three times at the top of the search page. MTM argues that because Amazon lists the search term “mtm special ops” at the top of the page, a consumer might conclude that the products displayed are types of MTM watches. But, a review of Amazon’s search results page shows that such consumer confusion is highly unlikely. None of these products are labeled with the word “MTM” or the phrase “Special Ops,” let alone the specific phrase “MTM Special Ops.” Further, some of the products are not even watches. The sixth result is a book entitled “Survive!: The Disaster, Crisis and Emergency Handbook by Jerry Ahem.” The tenth result is a book entitled “The Moses Expedition: A Novel by Juan Gómez-Jurado.” It is perplexing how one could assume that a book entitled “The Moses Expedition” is a type of MTM watch or is in any way affiliated with MTM watches. It is hard to fathom how a reasonably prudent consumer accustomed to shopping online would view Amazon’s search results page and conclude that the products offered are MTM watches. Some of the products are not even watches! And the watches that are offered for sale are clearly labeled as being manufactured by Luminox, Chase-Durer, TAWATEC, or Modus – not by MTM. It is possible that some dolt somewhere might be confused by the search results page. But, “[u]nreasonable, imprudent and inexperienced web-shoppers are not relevant.” Tabari, 610 F.3d at 1176; see also Network Automation, 638 F.3d at 1153 (“[W]e expect consumers searching for expensive products online to be even more sophisticated.”). 32 MULTI TIME MACHINE V. AMAZON.COM The majority hypothesizes, without any evidence to support it, that a reasonable jury could infer that initial interest confusion is possible here because consumers might view these search results and wonder whether a competitor has acquired MTM or is otherwise affiliated with MTM. There is no evidence in the record that anyone, anywhere, has ever labored under the mistaken impression that Luminox or the other brands offered are in any way, shape, or form affiliated with MTM. Moreover, to establish likelihood of confusion, MTM must show that confusion is likely, not just possible. See Murray, 86 F.3d at 861. MTM argues that in order to eliminate the likelihood of confusion, Amazon must change its search results page, so that it explains to customers that it does not offer MTM watches for sale before suggesting alternative watches to the customer. “No MTM, Luminox” is essentially what MTM says is required. I disagree. The search results page makes clear to anyone who can read English that Amazon only carries the brands of watches that are clearly and explicitly listed on the web page. The search results page is unambiguous. In light of the clear labeling Amazon uses on its search results page, no reasonable trier of fact could conclude that Amazon’s search results page would likely confuse a reasonably prudent consumer accustomed to shopping online as to the source of the goods being offered. See Playboy, 354 F.3d at 1030 n.44 (Clear labeling “might eliminate the likelihood of initial interest confusion that exists in this case.”); Network Automation, 638 F.3d at 1154 (same). Therefore, summary judgment of MTM’s trademark claims was appropriate. MULTI TIME MACHINE V. AMAZON.COM 33 MTM attempts to argue that summary judgment of its claims is inappropriate because there are numerous factual disputes related to Amazon’s search results page. But, to the extent there are any factual disputes between the parties, none of them are material to the analysis. MTM cannot dispute the fact that the watches at issue all sell for hundreds of dollars. Therefore, as a matter of law, the relevant consumer would be a reasonably prudent consumer accustomed to shopping online. See Tabari, 610 F.3d at 1176; Network Automation, 638 F.3d at 1152–53. Further, MTM cannot dispute the contents of the web page at issue. A review of Amazon’s web page shows that each product listed for sale is clearly labeled with the product’s name and manufacturer and a photograph, and none of the products are labeled with MTM’s mark. Thus, the undisputed facts show that it is highly unlikely that a reasonably prudent consumer accustomed to shopping online would be confused as to the source of the goods offered for sale on Amazon’s web page. It is true that likelihood of confusion is often a question of fact, but not always. In a case such as this, where a court can conclude that the consumer confusion alleged by the trademark holder is highly unlikely by simply reviewing the product listing/advertisement at issue, summary judgment is appropriate. Cf. M2 Software, 421 F.3d at 1085 (explaining that summary judgment of a trademark claim is appropriate where the plaintiff has failed to present “sufficient evidence to permit a rational trier of fact to find that confusion is ‘probable,’ not merely ‘possible’”). Indeed, in the similar context of evaluating allegations of consumer deception when dealing with false advertising claims, we have at least twice concluded – after a review of the label or advertisement at issue – that there was no likelihood of consumer deception as a matter of law because no reasonable consumer could have 34 MULTI TIME MACHINE V. AMAZON.COM been deceived by the label/advertisement at issue in the manner alleged by the plaintiff. See, e.g., Davis v. HSBC Bank, 691 F.3d 1152, 1162 (9th Cir. 2012); Freeman v. Time, Inc., 68 F.3d 285, 289–90 (9th Cir. 1995). II. In light of Amazon’s clear labeling of the products it carries, by brand name and model, accompanied by a photograph of the item, no rational trier of fact could possibly find that a reasonably prudent consumer accustomed to online shopping would likely be confused by the Amazon search results. I would hold that the district court correctly granted summary judgment in favor of Amazon, or as John Belushi might have put it, “No reversal. Affirm.”