Trademark Liability Online: Keyword Advertising, Marketplaces, and Contributory Infringement After Tiffany v. eBay

Summary. Trademark law on the internet is mostly a question of who is responsible when someone else's mark shows up in a search ad, a marketplace listing, or a print-on-demand storefront. This article explains the two doctrines that answer that question: the likelihood-of-confusion analysis as courts have adapted it to keyword advertising and search results pages (Rescuecom, Network Automation, 1-800 Contacts v. Lens.com, Rosetta Stone, Multi Time Machine, and the slow retreat from Brookfield's initial interest confusion), and the Inwood test for contributory liability as the Second Circuit applied it to eBay and later courts applied it to web hosts, flea markets, landlords, and Redbubble. It explains why generalized knowledge of counterfeiting is not enough, how willful blindness fills the gap, why there is no trademark DMCA and why Section 230 does not help platforms, and what the INFORM Consumers Act and the proposed SHOP SAFE Act change. It then turns practical: platform programs (Amazon Brand Registry, Project Zero, Transparency, eBay VeRO, and the rest), test purchases and evidence, takedown notices, Schedule A litigation, statutory damages for counterfeits, and the defenses available to sellers, advertisers, and platforms. Two worked examples, a cosmetics brand fighting counterfeits across three marketplaces and a competitor buying your brand as a keyword, show how the doctrine plays out in practice.

Type the name of almost any consumer brand into a search engine and the first thing you see is an advertisement, often for a competitor, triggered by the brand name you just typed. Scroll a marketplace for the same brand and you will find the genuine article listed next to knockoffs, gray-market imports, and a seller in another hemisphere whose storefront was created last Tuesday. The brand owner's instinct is to sue the biggest name on the page: Google, Amazon, eBay. That instinct is usually wrong, and understanding why is the key to enforcing trademarks online without wasting a year and a legal budget.

The platforms are rarely the infringers. They are intermediaries, and American trademark law holds intermediaries liable only on a narrow, fact-specific showing that the Supreme Court set out in 1982 and the Second Circuit refined for the internet in 2010. The advertiser who bought your mark as a keyword may not be liable either, unless the ad itself is confusing. This article explains both halves of that picture, the search half and the marketplace half, and then explains what actually works.

The short answer

• Buying a competitor's trademark as a search keyword is a "use in commerce" that can support an infringement claim, Rescuecom Corp. v. Google Inc., 562 F.3d 123 (2d Cir. 2009), but it is rarely infringing by itself. Liability turns on whether the resulting ad is likely to confuse, and clearly labeled ads that do not use the mark in their text usually are not. Network Automation, Inc. v. Advanced Systems Concepts, Inc., 638 F.3d 1137 (9th Cir. 2011); 1-800 Contacts, Inc. v. JAND, Inc., 119 F.4th 234 (2d Cir. 2024) (the bare purchase of a competitor's mark as a keyword is not infringement).
• A marketplace, host, or other intermediary is liable for its users' infringement only under the Inwood test: it intentionally induced the infringement, or it continued to supply its service to someone it knew or had reason to know was infringing. Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844 (1982).
• Generalized knowledge that counterfeits are sold on the platform is not enough. The plaintiff must show knowledge of specific infringing listings or sellers, or willful blindness to them. Tiffany (NJ) Inc. v. eBay Inc., 600 F.3d 93 (2d Cir. 2010).
• Willful blindness counts as knowledge. A defendant who suspects infringement and deliberately avoids confirming it is treated as if it knew. Omega SA v. 375 Canal, LLC, 984 F.3d 244 (2d Cir. 2021).
• There is no trademark DMCA and Section 230 does not apply. The copyright safe harbors in 17 U.S.C. § 512 have no trademark counterpart, and 47 U.S.C. § 230(e)(2) carves intellectual property claims out of the immunity. Platforms are protected by Tiffany, not by statute.
• Practically, the fastest relief comes from platform programs (Amazon Brand Registry, eBay VeRO, and their equivalents), followed by demand letters to sellers, and, for counterfeits at scale, Schedule A litigation with asset freezes and statutory damages under 15 U.S.C. § 1117(c).

The legal framework: three statutes and no safe harbor

Every claim discussed in this article rests on a short list of Lanham Act provisions, so it helps to see them together.

Notice what is missing. Nothing in the Lanham Act says when an intermediary is liable for someone else's infringement. Secondary liability in trademark law is entirely judge-made, derived from the common law of torts, and the Supreme Court has said it is narrower than its copyright cousin. Sony Corp. of America v. Universal City Studios, Inc., 464 U.S. 417 (1984), at 439 n.19, called Inwood's test a "narrow standard" and refused to borrow it for copyright, citing "the fundamental differences between copyright law and trademark law." The Restatement (Third) of Unfair Competition § 27 (1995) states the common-law rule: one who supplies goods to a distributor is liable for the distributor's infringement if it intentionally induces it or fails to take reasonable precautions where infringement can reasonably be anticipated. J. Thomas McCarthy, McCarthy on Trademarks and Unfair Competition ch. 25 (5th ed.), especially §§ 25:17 through 25:20.50, collects the cases and is the treatise courts cite most.

Two statutory absences shape everything else. First, there is no trademark analog to the DMCA's notice-and-takedown safe harbor. A service provider that follows § 512(c) of the Copyright Act (designated agent, expeditious removal, repeat-infringer policy) earns statutory protection from copyright damages; a provider that does exactly the same things for trademark complaints earns only a good set of facts for a Tiffany defense. See Digital Millennium Copyright Act Safe Harbors for Online Service Providers for how the copyright regime works. Second, Section 230's immunity for publishing third-party content, broad as it is, expressly excludes "any law pertaining to intellectual property." Courts have divided on whether that exclusion reaches state-law IP claims such as the right of publicity (the Ninth Circuit in Perfect 10, Inc. v. CCBill LLC, 488 F.3d 1102 (9th Cir. 2007) read it to cover only federal IP, and the Third Circuit later disagreed), but nobody disputes that federal Lanham Act claims fall outside the immunity. For the statute's actual reach, see Section 230 and Platform Liability: What the Statute Actually Says and Where It Stops.

Keyword advertising: is buying a competitor's mark a "use in commerce"?

Keyword advertising is the auction system by which search engines sell the right to have an ad appear when a user types a particular term. Advertisers bid on terms, including competitors' trademarks; the search engine displays the winning ads above or beside the "organic" results and labels them, more or less visibly, as sponsored. The practice generated a decade of litigation on a threshold question: when Google sells "Rescuecom" as a keyword, or when a competitor buys it, is anyone "using" the mark "in commerce" within the meaning of § 1127 and § 1114?

The Second Circuit initially said no in a software case, holding that pop-up ads triggered by a competitor's URL in an internal directory were not a trademark use because the mark was never displayed to the consumer. That decision, 1-800 Contacts, Inc. v. WhenU.com, Inc., 414 F.3d 400 (2d Cir. 2005), was read by several district courts to immunize keyword sales altogether, although others disagreed; the Eastern District of Virginia, for one, refused to dismiss GEICO's Lanham Act claims against Google and Overture over their sale of GEICO's marks as keywords. Government Employees Insurance Co. v. Google, Inc., 330 F. Supp. 2d 700 (E.D. Va. 2004). Four years later the same court corrected course. In Rescuecom Corp. v. Google Inc., 562 F.3d 123 (2d Cir. 2009), it held that Google's recommendation and sale of the plaintiff's mark as a keyword through its AdWords program and Keyword Suggestion Tool was a use in commerce, distinguishing WhenU on the ground that Google displayed, offered, and sold the mark itself. The court was careful to say it was deciding only the threshold question; whether the resulting ads were likely to confuse remained open. (Rescuecom later dropped the case.)

After Rescuecom, the "use" question largely disappeared from keyword litigation. The courts of appeals that have addressed it treat the purchase or sale of a trademark as a keyword as a use in commerce (the Ninth Circuit adopted Rescuecom's reasoning in Network Automation), and the fight moved to likelihood of confusion. The question is not entirely closed, though. Concurring in a 2024 Ninth Circuit keyword case, Judge Desai urged her court to reconsider en banc whether an advertiser that merely bids on a keyword, without ever displaying the mark, "uses" it at all, pointing out that Rescuecom addressed the search engine that sold the keyword, not the advertiser that bought it. Lerner & Rowe PC v. Brown Engstrand & Shely LLC, 119 F.4th 711 (9th Cir. 2024) (Desai, J., concurring). A keyword defendant should preserve the argument even where binding precedent forecloses it today.

Likelihood of confusion in a search results page

The multifactor confusion tests, Polaroid in the Second Circuit and Sleekcraft in the Ninth, were built for products on shelves, not ads on screens. Polaroid Corp. v. Polarad Electronics Corp., 287 F.2d 492 (2d Cir. 1961); AMF Inc. v. Sleekcraft Boats, 599 F.2d 341 (9th Cir. 1979). The leading keyword case explains how to adapt them.

Network Automation: the factors that actually matter

In Network Automation, Inc. v. Advanced Systems Concepts, Inc., 638 F.3d 1137 (9th Cir. 2011), Advanced Systems Concepts sold job-scheduling software under the registered mark ActiveBatch, and its competitor Network Automation, which sold a rival product called AutoMate, bought "ActiveBatch" as a keyword on Google and Bing so that its ads appeared when users searched for the term. After a cease-and-desist exchange, Network Automation sued for a declaration that the practice was lawful, and the mark owner counterclaimed and moved for a preliminary injunction. The district court granted the injunction by mechanically applying the three Sleekcraft factors an earlier Ninth Circuit case had called most important for internet disputes (similarity of marks, relatedness of goods, and shared marketing channels). The Ninth Circuit vacated. It held that those three factors were a poor fit for keyword advertising, because on the internet everyone shares the same marketing channel and the marks are by definition identical, and that the factors most relevant to keyword cases are instead:

1. The strength of the mark, since consumers searching for a strong, distinctive mark are more likely to expect results from that source.
2. Evidence of actual confusion, which in keyword cases is usually thin.
3. The type of goods and the degree of care a purchaser is likely to exercise; the court observed that internet users are increasingly sophisticated about sponsored links.
4. The labeling and appearance of the advertisements and the surrounding context on the results page: whether the ad is set apart and marked as sponsored, and whether its text uses the mark or clearly identifies a different source.

The court also warned against treating initial interest confusion (discussed below) as a shortcut. Diverting a user's attention is not infringement unless the diversion is likely to confuse as to source, and a clearly labeled ad for a competitor is more like a comparison display than a bait-and-switch.

1-800 Contacts v. Lens.com: the numbers that sink a keyword claim

1-800 Contacts, Inc. v. Lens.com, Inc., 722 F.3d 1229 (10th Cir. 2013), is the most useful keyword decision for defendants because it quantifies the problem. Lens.com bought nine keywords that were variants of the plaintiff's mark (such as "1-800 contact lenses"), and two of its commission-paid affiliates bought "1800Contacts" itself. None of Lens.com's own ads displayed the mark. Over eight months, Lens.com's keywords generated 1,626 ad impressions and just 25 clicks, a rate of 1.5 percent, and the Tenth Circuit reasoned that even if every one of those 25 users had been confused, initial interest confusion occurred at most 1.5 percent of the time, which could not support relief. The affiliates' numbers were lower still: more than 448,000 impressions for one affiliate's ads with clicks at about 0.7 percent, and about 243,000 for the other's with clicks under 1 percent. The court affirmed summary judgment on every claim based on ads that did not display the mark. One affiliate, however, had published at least one ad that put "1800 Contacts" in the ad copy, and as to that ad the court reversed summary judgment on contributory infringement. Lens.com learned of the ad from the complaint, which reproduced a screenshot of it, yet waited three months before having its affiliate network tell the affiliate to stop, when a single email blast to all of its affiliates would have ended the practice without first identifying the culprit. That is a useful gloss on Tiffany: where a defendant can stop an infringement through a channel that reaches the unknown infringer without disturbing anyone's lawful conduct, ignorance of the infringer's identity is no excuse. The lesson on both sides: keyword purchase plus a clean ad is almost never actionable, keyword purchase plus the mark in the ad copy frequently is, and an advertiser who learns its affiliates are doing the latter must act within days.

Rosetta Stone v. Google: the search engine as defendant

Rosetta Stone Ltd. v. Google, Inc., 676 F.3d 144 (4th Cir. 2012), is the one appellate case in which claims against a search engine survived summary judgment. Rosetta Stone showed that after Google changed its policy in 2009 to allow trademarks in ad text for resellers and others, sponsored links for counterfeit Rosetta Stone software multiplied; it had evidence of actual confusion from consumers who bought counterfeits through sponsored links, internal Google studies finding that the likelihood of confusion "remains high" when trademarks appear in the title or body of a sponsored link, testimony from two of Google's own in-house trademark lawyers (testifying as corporate designees) that they could not tell from a results page which sponsored links came from authorized resellers, and records of roughly 200 notices it had sent Google between September 2009 and March 2010 identifying sponsored links for counterfeit software. The Fourth Circuit vacated summary judgment for Google on direct infringement, contributory infringement, and dilution, holding that the evidence created triable issues, that the district court had wrongly applied the functionality doctrine to Google's use of the mark, and that Tiffany (decided after a full bench trial) could not be mechanically applied at summary judgment. The case settled before trial, so it stands for no more than this: a search engine that keeps selling a mark as a keyword to advertisers it has been told are counterfeiters has a Tiffany problem. A search engine that takes the notices seriously does not.

Multi Time Machine v. Amazon: labeling as the whole case

Keyword logic extends to a retailer's own site search. In Multi Time Machine, Inc. v. Amazon.com, Inc., 804 F.3d 930 (9th Cir. 2015), a watchmaker whose products Amazon did not carry complained that an Amazon search for "MTM Special Ops" returned a page of competitors' watches without saying "no results for MTM." A divided panel initially sided with the watchmaker, then on rehearing affirmed summary judgment for Amazon. The majority held that because the results page clearly labeled each watch with its own brand name and photograph, a reasonably prudent consumer shopping online for an expensive product would not be confused about who made what. The court said that in cases of this kind, the Sleekcraft factors are a poor fit and the question reduces to two things: who are the relevant consumers, and what would they reasonably believe based on what they saw on the screen. Judge Bea's dissent argued that a jury should decide. For retailers, Multi Time Machine is the design principle: label results clearly, show each product's brand, and the search page itself will not be the basis of liability.

Brookfield, initial interest confusion, and its decline

Initial interest confusion is the theory that a defendant infringes by using a mark to capture a consumer's attention even if the consumer realizes the truth before buying. The Ninth Circuit embraced it for the internet in Brookfield Communications, Inc. v. West Coast Entertainment Corp., 174 F.3d 1036 (9th Cir. 1999), where the defendant used "MovieBuff" in website metatags. The court's analogy was a highway billboard that falsely advertises a competitor's store at the next exit; the driver who takes the exit and finds a different store may stay and shop, and the mark owner has lost the sale. The theory was later applied to banner ads keyed to the plaintiff's marks in Playboy Enterprises, Inc. v. Netscape Communications Corp., 354 F.3d 1020 (9th Cir. 2004), where unlabeled ads triggered by "playboy" created a triable issue.

Courts have since narrowed the doctrine almost to vanishing in keyword cases. Network Automation held that the billboard analogy fails when the ad is labeled, because the consumer who sees a labeled competitor ad is not misled about who is behind it. Multi Time Machine reduced the inquiry to clear labeling. And in a 2024 dispute between two Arizona personal-injury firms, the Ninth Circuit affirmed summary judgment for a firm that had bought its rival's name as a Google keyword from 2015 to 2021, a tactic marketers call "conquesting" (and one the plaintiff had used itself in other contexts). Lerner & Rowe PC v. Brown Engstrand & Shely LLC, 119 F.4th 711 (9th Cir. 2024). The plaintiff's best evidence was 236 intake calls in which a caller mentioned the plaintiff's name, but Google's data showed 109,322 impressions of the defendant's ads on searches for that name, a confusion rate of roughly 0.2 percent. The ads never used the mark, and the consumers' degree of care and the ads' labeling both favored the defendant. The court's image is worth borrowing: with the numerator and the denominator both in the record, it could "see the entire iceberg." McCarthy § 23:6 traces the initial interest confusion doctrine's rise and retreat, and McCarthy ch. 25A collects the internet cases. The practical rule today: diversion is not confusion, and a keyword claim needs the mark in the ad, a misleading ad, or real evidence of confused buyers.

The advertising-side regulators: the FTC and the antitrust episode

Two developments outside trademark law bear on keyword practice. In June 2013 the Federal Trade Commission's staff sent letters to the major search engines updating its 2002 guidance on distinguishing paid search results from natural ones. The letters observed that the visual cues separating ads from organic results had become less distinct and asked search engines to use clear and prominent labels, shading, or borders so that consumers can tell advertising from search results. That guidance is enforced, if at all, under Section 5 of the FTC Act, 15 U.S.C. § 45, but it matters to trademark cases because the Network Automation and Multi Time Machine labeling analysis assumes the labels work.

The second development is a warning to brand owners who try to solve keyword disputes by contract. Between 2004 and 2013, 1-800 Contacts entered into thirteen settlement agreements (and one related sourcing agreement) with competitors under which each side promised not to bid on the other's marks and to use negative keywords, instructions that keep an ad from showing on particular searches, to suppress its ads. The FTC found those agreements to be unfair methods of competition that suppressed advertising and made it costlier for consumers to compare prices. The Second Circuit vacated the Commission's order in 1-800 Contacts, Inc. v. Federal Trade Commission, 1 F.4th 102 (2d Cir. 2021), and remanded with instructions to dismiss the complaint. It held that trademark settlement agreements are not immune from antitrust scrutiny, but that the Commission erred in treating these as "inherently suspect" and erred again in concluding that they failed the rule of reason. The case ended well for the brand owner, but only after years of agency litigation, and the lesson is that a keyword non-bidding agreement is an agreement among competitors about advertising, exactly the kind of agreement that Section 1 of the Sherman Act, 15 U.S.C. § 1, polices, and it should be drafted with antitrust counsel in the room. For a broader look at platform-related antitrust, see Antitrust for Technology Companies: Monopolization, Platform Conduct, and the IP-Antitrust Interface.

What the advertiser can and cannot do

Secondary liability: the Inwood test

Contributory trademark infringement is the doctrine that makes a non-infringer answer for an infringer's conduct because of what the non-infringer did or knew. The Supreme Court announced the test in a pharmaceutical case that had nothing to do with the internet. In Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844 (1982), generic drug makers sold capsules that copied the colors of the branded drug, and some pharmacists filled branded prescriptions with the generics and labeled the bottles with the brand name. Ives sued the generic makers for the pharmacists' passing off. The Court held that a manufacturer or distributor is contributorily liable if it either (1) intentionally induces another to infringe, or (2) continues to supply its product to one whom it knows or has reason to know is engaging in trademark infringement. On the record, the generic makers had done neither.

Three features of the test matter for everything that follows:

• There must be a direct infringement by someone. Contributory liability is derivative. If the seller's listing is not infringing (because the goods are genuine, or the use is nominative), the platform cannot be liable for it either. Rosetta Stone, 676 F.3d at 163.
• Knowledge is about particular infringers, and it is tested at the time of continued supply. The second prong speaks of continuing to supply "one whom" the defendant knows is infringing. That phrasing is the root of Tiffany's specific-knowledge requirement.
• The test was written for suppliers of goods. Applying it to suppliers of services required an additional step, which the courts of appeals supplied in the flea market and domain registrar cases.

From products to services: Hard Rock, Fonovisa, and Lockheed

Hard Rock Cafe Licensing Corp. v. Concession Services, Inc., 955 F.2d 1143 (7th Cir. 1992), was the first appellate case to apply Inwood to a service provider: the operator of Chicago-area "Swap-O-Rama" flea markets where a vendor sold counterfeit Hard Rock t-shirts. Writing for the Seventh Circuit, Judge Cudahy reasoned that the operator stood in the position of a landlord or licensor and, under Restatement (Second) of Torts § 877(c), is responsible for the torts of those it permits on its premises if it knows or has reason to know of them. The court added two refinements that remain the law: willful blindness (suspecting wrongdoing and deliberately failing to investigate) is equivalent to actual knowledge, but the operator has no affirmative duty to take precautions against counterfeits it does not know about. Because the trial judge's findings blurred willful blindness with a failure to take reasonable precautions, which is only negligence, the court vacated the judgment against the flea market and remanded. It admitted that "finding the line between 'willful blindness' and 'reason to know' may be like finding the horizon over Lake Michigan in a snowstorm," then drew the line anyway: willful blindness is subjective (what did this defendant suspect, and what did it do about it?), while reason to know is objective (what would a reasonably prudent person in its position have known?). The court also addressed vicarious liability and held that it requires an actual or apparent partnership, authority to bind one another in transactions with third parties, or joint ownership or control of the infringing product, none of which a flea market has with its vendors.

Fonovisa, Inc. v. Cherry Auction, Inc., 76 F.3d 259 (9th Cir. 1996), reached the same result for a California swap meet where vendors sold counterfeit recordings and the county sheriff had already raided the premises and notified the operator. The Ninth Circuit held that the operator, which supplied the space, parking, advertising, and customers, had "supplied the necessary marketplace" and could be contributorily liable for continuing to do so with knowledge. Then in Lockheed Martin Corp. v. Network Solutions, Inc., 194 F.3d 980 (9th Cir. 1999), the court drew the other boundary. A domain name registrar that merely translated domain names into IP addresses and had no control over the content of the registrants' sites was not liable for cybersquatters' registrations of "skunkworks" variants, even after notice. The test the court announced for service providers is the one still used: the plaintiff must show that the defendant exercised direct control and monitoring over the instrumentality used to infringe. A registrar's "rote translation service" did not qualify. The Anticybersquatting Consumer Protection Act, enacted about a month after the decision, now addresses domain disputes directly, and the Lanham Act limits registrar liability expressly in § 1114(2)(D); see Domain Name Disputes: Cybersquatting, the ACPA, and the UDRP.

Tiffany v. eBay: the case that defined marketplace liability

Tiffany sued eBay in 2004 after its own buying program found that most of the "Tiffany" silver jewelry it purchased on eBay was counterfeit (73.1 percent in a 2004 sample, although the trial court found the programs methodologically flawed). The theory was straightforward: eBay knew, in general, that a great deal of the Tiffany merchandise on its site was fake; it profited from every sale; so it should be contributorily liable for continuing to provide its service to the sellers. After a bench trial, the district court rejected the claim, and the Second Circuit affirmed in Tiffany (NJ) Inc. v. eBay Inc., 600 F.3d 93 (2d Cir. 2010).

What eBay did

The facts mattered enormously, and anyone who runs or sues a platform should know them. eBay spent as much as $20 million a year on trust and safety tools and ran a trust and safety department of some 4,000 employees, more than 200 of whom worked only on infringement. It operated a Verified Rights Owner program (VeRO) through which rights holders reported listings on a Notice of Claimed Infringement form, and its practice was to remove a reported listing within 24 hours; in fact it removed 70 to 80 percent within 12 hours. Its fraud engine screened listings automatically, including with Tiffany-specific filters built around some 90 keywords. It suspended hundreds of thousands of sellers a year, tens of thousands of them for suspected infringement, generally under a three-strikes rule. The trial court found that eBay never refused to remove a reported Tiffany listing and always gave Tiffany the seller's contact information. Tiffany filed thousands of these notices. What eBay would not do was what Tiffany's demand letters asked: treat every seller offering five or more Tiffany items as a presumptive counterfeiter. Both courts found that presumption factually unfounded, because genuine Tiffany goods really were sold on eBay.

What the court held

No direct infringement. eBay's own use of "Tiffany" in advertising and on its site to describe genuine Tiffany goods actually offered there was a lawful nominative use. It was true that genuine Tiffany goods were sold on eBay, and eBay did not suggest Tiffany's sponsorship.

No contributory infringement, and the standard is specific knowledge. Because eBay no longer disputed the point on appeal, the court assumed without deciding that Inwood governs a service provider like eBay; the trial court had applied it because eBay retained "significant control" over the listings and transactions on its site. The court then held that "for contributory trademark infringement liability to lie, a service provider must have more than a general knowledge or reason to know that its service is being used to sell counterfeit goods. Some contemporary knowledge of which particular listings are infringing or will infringe in the future is necessary." Tiffany's demand letters and buying-program results told eBay that counterfeiting was a problem; they did not tell eBay which listings were counterfeit, and when Tiffany did identify listings, eBay took them down.

Willful blindness fills the gap, but eBay was not willfully blind. The court was explicit that a provider "may not shield itself from learning of the particular infringing transactions by looking the other way," and that willful blindness is knowledge. But a provider that invests heavily in detection, responds to notices, and suspends offenders is the opposite of willfully blind. The court also noted, in a footnote that defendants quote often, that a service provider is not liable "merely for failing to anticipate that others would use its service to infringe."

The burden of policing stays with the brand owner. This is the holding's practical core. The court declined to shift the cost of identifying counterfeits from Tiffany, which knows its own products, to eBay, which does not. A platform that gives rights holders an effective reporting tool and uses it in good faith has done what the law requires.

False advertising remanded. The court did remand a narrow false advertising claim concerning eBay's sponsored-link ads for "Tiffany" items, directing the district court to consider whether the ads misled consumers into believing the goods were all genuine. On remand the district court found no false advertising, and that was the end of the case.

Why Tiffany is read as a platform charter

Tiffany did not create a safe harbor. It applied a knowledge standard to a defendant with an unusually good record. But because its facts became the template, it functions as a charter: a marketplace that (1) provides a notice channel, (2) removes identified listings promptly, (3) escalates against repeat offenders, (4) invests in proactive detection it is not legally required to build, and (5) does not ignore red flags about particular sellers, will almost never be contributorily liable. Each later case tests one of those five elements.

After Tiffany: hosts, landlords, flea markets, and willful blindness

Louis Vuitton v. Akanoc: web hosts that ignore notices

In Louis Vuitton Malletier, S.A. v. Akanoc Solutions, Inc., 658 F.3d 936 (9th Cir. 2011), web hosting companies and their principal hosted websites, run by customers in China, that sold counterfeit Louis Vuitton goods. Louis Vuitton sent at least eighteen notices of infringement in 2006 and 2007 identifying the sites and got no response. A jury found willful contributory trademark and copyright infringement and awarded statutory damages of $10.5 million on the trademark claims and $300,000 on the copyright claims against each of three defendants, about $32.4 million in all. The district court then set aside the verdict against an affiliated company that merely owned and leased the server hardware, and the Ninth Circuit affirmed that ruling along with the liability findings against the company that operated the servers and its principal. It held, however, that the awards could not simply be multiplied by the number of defendants, and directed entry of single awards totaling $10.8 million for which the host and its principal were jointly and severally liable. The host argued, under Lockheed, that it did not control the infringing websites. The court disagreed: the host physically housed the sites on its servers, controlled the "master switch" that could take them offline, and therefore directly controlled and monitored the instrumentality. The case is Tiffany's mirror image. Same test, opposite facts, opposite outcome.

Coach v. Goodfellow and the flea market line

Coach, Inc. v. Goodfellow, 717 F.3d 498 (6th Cir. 2013), affirmed a summary judgment of contributory liability and a $5.04 million jury award of statutory damages ($240,000 for each of 21 marks, on a finding of willfulness) against the operator of a Memphis flea market whose vendors sold counterfeit Coach goods. Coach sent a demand letter in January 2010, the county district attorney general's office sent another two months later, and law enforcement raided the market three times; the last raid involved some 150 officers, seized more than 4,600 counterfeit items, and shut the market down for good. The operator's response had been to hand out pamphlets at random, post signs that were actually aimed at counterfeit currency, and hold a voluntary vendor meeting on a day the market was closed. The Sixth Circuit held that he had actual knowledge, that he engaged in "ostrich-like practices," and that his efforts did not compare with eBay's. Together with the Eleventh Circuit's decision affirming a verdict against the owners of a discount mall whose subtenants sold counterfeit Ray-Ban eyewear through three law enforcement raids, Luxottica Group, S.p.A. v. Airport Mini Mall, LLC, 932 F.3d 1303 (11th Cir. 2019), the flea market cases show that physical and online marketplaces are judged by the same rules and that doing something is not the same as doing enough.

Omega v. 375 Canal: willful blindness without a named vendor

The Second Circuit's most important post-Tiffany decision is Omega SA v. 375 Canal, LLC, 984 F.3d 244 (2d Cir. 2021). The defendant owned the building at 375 Canal Street in Manhattan, which came with a record: nuisance suits by the City of New York settled in 2006 and 2009 (each time with promises to keep counterfeits off the premises), a 2006 consent injunction in a Louis Vuitton case, and a 2010 police sting in which a counterfeit Omega watch was sold inside the building. Omega's owner wrote to the landlord in 2011, an investigator bought a counterfeit Omega Seamaster there in 2012, and Omega sued the landlord for its tenants' sales. The landlord argued that under Tiffany Omega had to prove it knew of a specific vendor selling specific counterfeits and continued to lease to that vendor. The Second Circuit rejected that reading. Tiffany itself had said that if eBay had reason to suspect counterfeiting and "intentionally shielded itself from discovering the offending listings or the identity of the sellers behind them," it could be charged with knowledge. So, the court held, a defendant "may be willfully blind either to particular transactions or to the identities of infringers," and the plaintiff need not identify the vendor by name. The court also restated Tiffany's limits: no liability for general knowledge of counterfeiting on the property, and no affirmative duty to police. It affirmed the jury's award of $275,000 per mark for four marks, $1.1 million in statutory damages. The decision reconciles the two strands of Tiffany: specific knowledge is required, but a defendant cannot manufacture ignorance by refusing to look.

Print-on-demand: the two Redbubble cases

Redbubble operates a marketplace where independent artists upload designs that Redbubble's contracted manufacturers print onto shirts, stickers, and mugs when a customer orders; Redbubble handles payment, shipping, and customer service, and the goods arrive in Redbubble packaging. That business model tests the line between a passive marketplace and a seller. Two circuits reached different stages of the question.

In Ohio State University v. Redbubble, Inc., 989 F.3d 435 (6th Cir. 2021), the district court granted summary judgment to Redbubble on direct infringement, reasoning that it was a mere facilitator like eBay. The Sixth Circuit reversed. It held that direct liability under § 1114 extends beyond the party that physically applies the mark to those who "use" the mark in commerce, and that a company that brings infringing goods into being, markets them under its own brand, and ships them in its own packaging may be more than a passive intermediary. The court remanded for a fact-intensive inquiry into Redbubble's degree of involvement.

In Y.Y.G.M. SA v. Redbubble, Inc., 75 F.4th 995 (9th Cir. 2023), brought by the owner of the Brandy Melville marks, a jury found Redbubble liable for willful contributory counterfeiting and contributory infringement; the trial judge then set aside part of the verdict and let the rest stand. On appeal Brandy Melville argued that a platform with general awareness of infringement has a duty to take reasonable corrective action. The Ninth Circuit disagreed. Agreeing with the Second and Fourth Circuits, it held that contributory liability requires knowledge of specific infringers or specific instances of infringement, and that general knowledge of infringement on a platform, even infringement of the plaintiff's own marks, does not establish willful blindness. Willful blindness can supply the knowledge, but only where the defendant subjectively believed infringement was likely occurring and took deliberate actions to avoid learning about it. The panel vacated the post-trial ruling and remanded for the district court to apply that standard. It also held that the trial court had analyzed confusion on the counterfeiting claim the wrong way by asking whether the goods were "stitch-for-stitch" copies, vacated the denial of a permanent injunction because the court had discounted future harm under the statutory presumption of irreparable harm, and affirmed the denial of prejudgment interest on statutory damages. Read together, the Redbubble cases mean that a print-on-demand platform faces exposure on two fronts: a direct claim if its role in creating and branding the product is active enough, and a contributory claim if it does not act on specific notices.

Payment processors and the "instrumentality" question

The Lockheed control requirement produced a split on payment intermediaries. In Perfect 10, Inc. v. Visa International Service Ass'n, 494 F.3d 788 (9th Cir. 2007), the Ninth Circuit held that credit card networks that processed payments for infringing websites were not contributorily liable for either copyright or trademark infringement, because the infringement (display of images) happened whether or not anyone paid, and the networks had no control over the sites. A district court in New York distinguished Visa in Gucci America, Inc. v. Frontline Processing Corp., 721 F. Supp. 2d 228 (S.D.N.Y. 2010), denying a motion to dismiss claims against credit card processors and an intermediary whose services were necessary to complete sales of counterfeit goods and whose business model allegedly catered to high-risk counterfeiters. The distinction is whether the service is essential to the infringing transaction, and plaintiffs now routinely name payment processors in counterfeiting cases on that theory.

Vicarious liability: why it almost never works

Vicarious trademark liability asks not what the defendant did but what relationship it had with the infringer. Most courts, following Hard Rock, require an actual or apparent partnership, authority to bind one another in transactions with third parties, or joint ownership or control over the infringing product. See Kelly-Brown v. Winfrey, 717 F.3d 295 (2d Cir. 2013) (affirming dismissal of vicarious-liability claims against corporate sponsors of an allegedly infringing event); Rosetta Stone, 676 F.3d at 165. The Tenth Circuit in Lens.com analyzed the question in ordinary agency terms: even assuming the commission-paid affiliates were the advertiser's agents, the advertiser was not vicariously liable for the one ad that used the plaintiff's mark, because the affiliate (whose employee had placed the ad without his knowledge) never believed he had authority to do that. A marketplace and its sellers, a search engine and its advertisers, a host and its customers: none of these is a partnership, and courts have consistently declined to import copyright's broader "right and ability to supervise plus financial benefit" test into trademark law. Hard Rock, 955 F.2d at 1150. Vicarious liability belongs in a complaint only where the defendants genuinely operate as one enterprise (for instance, a corporate parent that jointly controls the infringing product, or affiliated entities that share ownership of a storefront).

A note on the copyright analogs

Copyright's secondary liability doctrines are broader and should be cited carefully. Metro-Goldwyn-Mayer Studios Inc. v. Grokster, Ltd., 545 U.S. 913 (2005), established inducement liability for distributing a product with the object of promoting infringement, which overlaps with Inwood's first prong. BMG Rights Management (US) LLC v. Cox Communications, Inc., 881 F.3d 293 (4th Cir. 2018), held that an ISP forfeited its DMCA safe harbor by failing to implement its repeat-infringer policy. Trademark cases cite these decisions for willful blindness and for the proposition that a paper policy is not a real one, but the trademark test remains Inwood, and the Supreme Court's observation in Sony that trademark secondary liability is the narrower doctrine is still good law. The same caution applies to district court decisions on music streaming sites, which address copyright's direct-liability volitional-conduct rule rather than anything in the Lanham Act.

How courts actually apply it: where the circuits differ

The doctrine is national in name and regional in practice. Four divergences matter when you are choosing a forum, sizing exposure, or predicting how a judge will read the same facts.

Initial interest confusion. The Ninth Circuit built the internet version of the doctrine in Brookfield and has since confined it, for labeled keyword ads, almost to nothing. The Seventh Circuit applied it to a competitor that put a misspelled version of a rival's mark in its website metatags, affirming an injunction that required corrective language on the competitor's site. Promatek Industries, Ltd. v. Equitrac Corp., 300 F.3d 808 (7th Cir. 2002). The Tenth Circuit recognized it where unauthorized resellers used the plaintiffs' marks on their websites, in metatags, and in paid premium search placements to draw buyers who were then offered other brands. Australian Gold, Inc. v. Hatfield, 436 F.3d 1228 (10th Cir. 2006). The Second Circuit recognizes the theory but requires a plaintiff to plead intentional deception in internet cases, because "consumers diverted on the Internet can more readily get back on track than those in actual space." JAND, 119 F.4th 234 (quoting Savin Corp. v. Savin Group, 391 F.3d 439, 462 n.13 (2d Cir. 2004)). The Fourth Circuit has said flatly that it has "never adopted the initial interest confusion theory," preferring to examine a use "in the context in which it is seen by the ordinary consumer." Lamparello v. Falwell, 420 F.3d 309 (4th Cir. 2005). A diversion-only keyword claim is weak everywhere; in the Fourth Circuit it is weaker still.

What counts as knowledge. Every circuit to decide the question requires knowledge of specific infringers or specific instances of infringement, and every one accepts willful blindness as a substitute. They differ at the margins. The Second Circuit lets a defendant be willfully blind to infringers' identities (Omega). The Ninth Circuit requires a subjective belief that infringement was likely plus deliberate steps to avoid learning more (Y.Y.G.M.). The Tenth Circuit dispenses with knowing the infringer's identity where a simple, targeted remedy, such as an email to every affiliate, was available (Lens.com). The Sixth and Eleventh Circuits have applied the doctrine to a flea market operator and to mall owners whose notice came from demand letters and police raids (Coach; Luxottica).

Control of the instrumentality. The Ninth Circuit makes "direct control and monitoring of the instrumentality used by a third party to infringe" an express element for service providers (Lockheed; Perfect 10 v. Visa; Akanoc). The Second Circuit in Tiffany assumed without deciding that Inwood governs a service provider, after the trial court found that eBay retained significant control over its listings and transactions, and district courts in that circuit have treated control as part of the test (Frontline). Where control is contested, as with payment processors, registrars, and advertising networks, the forum can decide the case.

Section 230 and state-law claims. Federal Lanham Act claims fall outside Section 230 everywhere. State-law claims are different. The Ninth Circuit reads the intellectual-property exception in § 230(e)(2) to cover only federal intellectual property, so a platform can invoke Section 230 against state-law claims such as right of publicity and state unfair competition (CCBill). The Third Circuit holds that the exception reaches state intellectual property laws too. Hepp v. Facebook, 14 F.4th 204 (3d Cir. 2021). A complaint against a platform that pairs the Lanham Act with state claims should be drafted with that split in mind.

Platform programs and platform regulation: the enforcement layer Tiffany created

Because Tiffany rewards platforms that build notice-and-removal systems, every major marketplace has built one, and the programs are now the first line of enforcement for most brands. They are private contractual systems, not legal procedures; a platform can change the rules, and a removal decision is not an adjudication of anything. But they are fast, free, and effective against the great majority of infringing listings.

Three points of practice. First, enrollment requires a registered mark almost everywhere, which is the single strongest argument for registering before you have a problem; see Benefits of Federal Trademark Registration. Second, platforms track the accuracy of your reports, and a brand that reports lawful resale of genuine goods as "counterfeit" will lose its expedited privileges and may face a seller's counter-complaint; see the first sale discussion below. Third, every removal notice is a record. If the matter ends up in litigation, the sequence of notices and the platform's responses is the evidence that establishes either the platform's good faith (as in Tiffany) or its willful blindness (as in Akanoc), and the seller's repeat-offender status is what supports willfulness and enhanced damages against the seller.

Two further developments are worth knowing. Some marketplaces now sue counterfeiters themselves, sometimes jointly with the brand; Amazon formed a Counterfeit Crimes Unit in 2020 for that purpose. And bad actors have learned to work the programs from the other direction, filing trademark applications (occasionally with doctored specimens) to obtain registry access and then reporting the genuine seller. The answers to those filings are the Trademark Modernization Act's tools: a letter of protest during examination under 37 C.F.R. § 2.149, and expungement or reexamination petitions after registration under 15 U.S.C. §§ 1066a and 1066b. See Expungement, Reexamination, and Letters of Protest: The Trademark Modernization Act in Practice.

The INFORM Consumers Act and the SHOP SAFE proposals

Congress has so far declined to legislate a trademark liability standard for platforms, but it has imposed transparency duties. The Integrity, Notification, and Fairness in Online Retail Marketplaces for Consumers Act, codified at 15 U.S.C. § 45f and effective June 27, 2023, requires an online marketplace to collect and verify identifying information (bank account, government ID, tax ID, and contact information) from any "high-volume third party seller," defined as a seller with 200 or more discrete sales of new or unused consumer products totaling $5,000 or more in gross revenue in any continuous twelve-month period during the preceding two years. Sellers above $20,000 in annual revenue must have their name, address, and contact information disclosed to consumers on the product listing or order confirmation, with limited exceptions. Marketplaces must suspend sellers who do not comply and must provide consumers a mechanism to report suspicious activity. The FTC and state attorneys general enforce the Act; a violation is treated as a violation of an FTC rule defining an unfair or deceptive practice, which exposes a marketplace to civil penalties.

For brand owners, INFORM changes the economics of anonymous selling. A high-volume seller's identity exists in the platform's records and can be subpoenaed; a seller who gave false information to the platform has committed a separate wrong that supports willfulness; and a marketplace that failed to verify at all has a regulatory problem independent of trademark law. For the mechanics of compelling a platform to identify a seller, see Identifying an Anonymous Defendant Online: A Practical Guide.

The SHOP SAFE Act, introduced in several Congresses since 2020 and not enacted as of this writing, would go further. It would make a platform contributorily liable for a third-party seller's use of a counterfeit mark on goods that implicate health or safety unless the platform follows a statutory list of best practices: verifying seller identity, requiring sellers to consent to U.S. jurisdiction, screening listings proactively, terminating repeat offenders, and so on. In effect it would codify Tiffany's facts as the price of a safe harbor, and shift the policing burden toward platforms for the most dangerous categories of counterfeit. Whether it passes, and in what form, is worth checking before relying on the current allocation of burdens. Separately, legislation to let Customs seize goods that infringe design patents (the Counterfeit Goods Seizure Act, introduced in 2019) has not become law; Customs recordation and seizure remain available for registered trademarks and copyrights under 19 C.F.R. Part 133, which the border enforcement documents cover: Stopping Gray Market and Counterfeit Imports: A Practical Guide and Border Enforcement Toolkit: Recordations, Seizure Responses, and Gray Market Programs.

Enforcement mechanics: from test purchase to judgment

Test purchases and chain of custody

Every online enforcement matter begins with a purchase. A test purchase, meaning a purchase of the suspect product made for the purpose of examination, does four jobs: it proves the product is counterfeit or otherwise infringing (a listing photograph proves nothing about the physical goods), it establishes that the seller ships to the forum (which supports personal jurisdiction), it identifies the shipper and payment recipient, and it creates the exhibit for the preliminary injunction motion. The purchase should be made by an investigator or paralegal, not the lawyer who will argue the case, using an account that does not identify the brand owner; the listing should be captured in full before purchase (seller name, item number, price, photos, shipping terms, the page URL and timestamp); the package should be photographed unopened, then opened on camera, with the contents compared against a genuine exemplar; and each person who handled it should sign a chain-of-custody log. Authentication of the resulting screenshots and goods follows Fed. R. Evid. 901, which requires only evidence sufficient to support a finding that the item is what it purports to be, usually the testimony of the person who made the capture. The Authenticating Website Evidence: A Practical Checklist walks through the capture and authentication steps, and the companion toolkit in this cluster supplies the log forms: Online Trademark Enforcement Toolkit: Platform Notices, Test Purchase Records, and Contributory Liability Demands.

Notices, counter-notices, and escalation

A platform notice should identify the registration, the listing, the reason the listing infringes, and the sender's authority, and should be accurate, because a knowingly false notice exposes the sender to the seller's claims (tortious interference, unfair competition, and under some platforms' terms, loss of reporting privileges) and because, if litigation follows, the notice will be an exhibit. Unlike the DMCA, trademark takedowns have no statutory counter-notice procedure; most platforms nonetheless allow the seller to dispute a removal, and a disputed removal usually ends with the platform reinstating the listing and telling the parties to resolve it elsewhere. Escalation runs: individual notices; a request that the platform act against the seller account rather than the listing; a demand letter to the seller if the seller is identifiable; and litigation. For the demand letter stage, see Drafting a Trademark Cease-and-Desist Letter and, for the recipient's side, Responding to a Trademark Cease-and-Desist Letter.

Schedule A litigation and asset freezes

For counterfeiting at scale, the dominant tool is the "Schedule A" case: a single complaint in a federal district (historically the Northern District of Illinois, more recently also Florida, New York, and Texas) against dozens or hundreds of sellers identified only by their storefront names on a sealed schedule; an ex parte temporary restraining order under Fed. R. Civ. P. 65 and § 1116 that enjoins the sales and, critically, directs the marketplaces and payment processors to freeze the sellers' accounts; alternative service by email under Fed. R. Civ. P. 4(f)(3), which the Ninth Circuit approved for an elusive foreign online business in Rio Properties, Inc. v. Rio International Interlink, 284 F.3d 1007 (9th Cir. 2002); and, for the many defendants who never appear, default judgments awarding statutory damages under § 1117(c) that are satisfied from the frozen funds. Courts have grown skeptical of the practice, scrutinizing joinder under Fed. R. Civ. P. 20, personal jurisdiction over sellers whose only forum contact is a single test purchase (the Seventh Circuit upheld jurisdiction on that basis in NBA Properties, Inc. v. HANWJH, 46 F.4th 614 (7th Cir. 2022), where the seller's interactive store accepted and shipped an order to Illinois), and the proportionality of freezes to the likely judgment. The freezes themselves rest on the equitable character of an accounting of profits, which most courts hold takes these cases outside the rule of Grupo Mexicano de Desarrollo, S.A. v. Alliance Bond Fund, Inc., 527 U.S. 308 (1999), that a court may not freeze assets before judgment in an action for money damages alone. See Reebok International Ltd. v. Marnatech Enterprises, Inc., 970 F.2d 552 (9th Cir. 1992); Levi Strauss & Co. v. Sunrise International Trading Inc., 51 F.3d 982 (11th Cir. 1995). The practice, its mechanics, and the pushback are covered in Counterfeiting, Seizure Orders, and Schedule A Litigation: Fighting Fakes in Federal Court, and the companion guide in this cluster, Enforcing Trademarks Against Online Sellers and Platforms: A Practical Guide, gives the step-by-step. The ex parte seizure remedy of § 1116(d), which authorizes the U.S. Marshals to seize goods and records, remains available but is used far less often than account freezes, because the goods in an online case are usually overseas and the money is not.

Damages

Against a direct infringer, § 1117(a) allows recovery of the defendant's profits, the plaintiff's damages, and costs, with fees in exceptional cases; after Romag Fasteners, Inc. v. Fossil, Inc., 590 U.S. 212 (2020), willfulness is important but not a precondition to a profits award. Where the mark is a counterfeit (a spurious mark identical with or substantially indistinguishable from a registered mark, used on the registered goods, § 1127), § 1117(b) mandates treble damages and fees absent extenuating circumstances, and § 1117(c) lets the plaintiff elect statutory damages instead, which is what Schedule A plaintiffs invariably do because no seller produces sales records. Contributory infringers are jointly and severally liable with the direct infringer for the same remedies, which is how the host and its principal in Akanoc and the landlord in Omega ended up paying statutory damages for counterfeits they never sold. Extraterritorial sales are not reachable: Abitron Austria GmbH v. Hetronic International, Inc., 600 U.S. 412 (2023), limits § 1114 and § 1125(a) to infringing uses in commerce in the United States, so the sales that count are the ones shipped here.

Evidence that wins and loses these cases

Online trademark cases are unusually data-rich, and the side that controls the data usually wins.

Keyword cases turn on numbers. The ad platform records every impression, click, and conversion, and those figures are discoverable. Lens.com and Lerner & Rowe both ended on the ratio of possibly confused consumers to exposed consumers, and both treat rates in the low single digits, or below, as de minimis. Anecdotes need context: in Lens.com a customer-service note about a caller who canceled an order after realizing she had not reached the plaintiff got little weight because nothing showed how she had found the defendant. A plaintiff should log calls, chats, and misdirected reviews with enough detail (date, channel, what the customer searched for, what she said) to tie each one to the ads; a defendant should insist on the denominator. Discovery fights over keyword data are common and costly to lose: the Tenth Circuit affirmed sanctions against Lens.com for dilatory and obstructive responses to requests for its own and its affiliates' keyword purchase records.

Surveys must replicate the screen. A confusion survey in a keyword case has to show respondents what a searcher actually saw and ask unambiguous questions. The Lens.com survey was excluded under Fed. R. Evid. 702 because its key question (whether the ad "originates from 1-800-CONTACTS") could be read as asking about the search term rather than the company, and because the questions suggested a connection. Even if admitted, its net result (19.4 percent in the test group against 11.9 percent in a control group shown a search for "contact lenses") was entitled to minimal weight. McCarthy states the conventional view that results below 10 percent can become evidence that confusion is not likely, 6 McCarthy § 32:189, and the Restatement (Third) of Unfair Competition § 20 cmt. g notes that surveys showing 7 to 15 percent confusion have been held adequate when supported by other evidence. For survey design, see Consumer Survey Expert Methodology in Trademark Cases: A Practical Guide.

Knowledge cases turn on the notice file. In a contributory case the decisive exhibit is the chronology of notices and responses: what the platform was told, how specifically, when, and what it did. Tiffany won on that chronology; Akanoc and Omega lost on theirs. Plaintiffs should send notices that identify listings and sellers, keep proof of delivery, and log every response and every silence. Platforms should log every notice, the action taken, and the time to action, because evidence of a routine practice is admissible to show what was done on a particular occasion under Fed. R. Evid. 406. Internal documents showing suspicion about particular sellers are the raw material of willful blindness; in Rosetta Stone the search engine's own studies and its lawyers' testimony created the triable issues.

Counterfeit status needs a witness. A listing photograph proves nothing about the goods. Courts expect a test purchase authenticated under Fed. R. Evid. 901 by the person who made it, an examination by someone qualified to say why the item is not genuine (packaging, batch codes, materials, construction), and, where volumes are large, a summary chart of listings under Fed. R. Evid. 1006 with the underlying captures available for inspection. A seller's own listing statements are admissions of a party opponent under Fed. R. Evid. 801(d)(2), and records a platform produces in response to a subpoena under Fed. R. Civ. P. 45 can usually be authenticated by a records custodian's certification under Fed. R. Evid. 902 rather than live testimony.

Identity and money evidence comes from third parties. Seller identities and payment data sit with the marketplaces and payment processors, and a plaintiff usually needs a court order for expedited discovery under Fed. R. Civ. P. 26(d) before serving subpoenas. Since June 2023 the INFORM Consumers Act has required marketplaces to collect and verify bank, tax, and contact information for high-volume sellers, which makes those subpoenas more productive than they used to be.

What it costs and how long it takes

Platform takedowns cost staff time and little else; a brand with a few hundred listings a month can run a program with a paralegal and a monitoring vendor for a few thousand dollars a month. A demand letter to an identifiable domestic seller runs from a few hundred to a few thousand dollars and resolves most cases in which the seller has anything to lose. A Schedule A case against a hundred anonymous sellers typically costs in the tens of thousands of dollars through default judgment, is funded in practice from the frozen balances, and yields recoveries that range from nothing to six figures depending on what the freezes catch. A contested keyword case through summary judgment, with survey experts and ad-platform data, can easily cost each side several hundred thousand dollars. A contested contributory-liability case against a platform is a different animal entirely: Tiffany was litigated for six years through a bench trial and appeal, and Rosetta Stone for roughly three before it settled. Timing: platform removals, one to five days; a TRO in a Schedule A case, one to two weeks from filing; default judgments, three to six months; a contested platform case, years.

Defenses: the seller's and the platform's side of the ledger

First sale. A reseller of genuine goods does not infringe by using the mark to identify them, even if the brand owner would prefer it not. The rule traces to Prestonettes, Inc. v. Coty, 264 U.S. 359 (1924), and Champion Spark Plug Co. v. Sanders, 331 U.S. 125 (1947), which permitted the resale of reconditioned plugs with adequate disclosure. The exception, and the brand owner's usual response, is the material-differences doctrine: goods that differ materially from the authorized version (different formula, missing warranty, no quality control, repackaged, expired) are not "genuine" for first sale purposes. A brand that reports a lawful reseller's listing as counterfeit is asserting a false claim; a reseller that strips serial numbers or sells units diverted from a quality-controlled channel has likely lost the defense. See Gray Market Goods and the First Sale Doctrine: Parallel Imports, Material Differences, and Exhaustion.

Nominative fair use. An advertiser, reseller, or repair shop may use the mark to refer to the brand owner's product where the product is not readily identifiable without it, only so much of the mark is used as is reasonably necessary, and nothing suggests sponsorship or endorsement. New Kids on the Block v. News America Publishing, Inc., 971 F.2d 302 (9th Cir. 1992). Tiffany applied the doctrine to eBay's own advertising. Keyword ads that say "Compatible with BrandX" or "We repair BrandX phones" live here, and the line is crossed when the ad adopts the brand's logo, stylization, or trade dress, or uses the mark more than necessary. After Jack Daniel's Properties, Inc. v. VIP Products LLC, 599 U.S. 140 (2023), use of a mark as a source identifier for the defendant's own goods gets no special First Amendment screening, which matters for sellers who adopt a brand's name as their storefront name. The companion cluster covers the defenses in depth: Asserting or Defeating a Trademark Fair Use Defense: A Practical Guide.

Descriptive fair use. Where the term is descriptive and used descriptively (a seller of "special ops" watches describing them that way), KP Permanent Make-Up, Inc. v. Lasting Impression I, Inc., 543 U.S. 111 (2004), holds that the defendant need not negate confusion to invoke the defense.

The platform's defenses, in order. No underlying direct infringement (the goods are genuine, or the use is nominative). No control over the instrumentality (Lockheed; weaker for marketplaces and hosts than for registrars and payment networks). No specific knowledge (Tiffany): the notices were general, or identified nothing, or were acted upon. No willful blindness (Omega, Y.Y.G.M.): the platform investigated what it was told. Adequate remedial measures: removals, suspensions, detection investment. And, against a direct-infringement theory of the Ohio State kind, that the platform's role was passive: no production, no branding, no packaging, no promotion of the specific goods.

Strategy for each side

The same doctrine points each party toward a different playbook.

For the brand owner

1. Register first. Platform programs and counterfeiting remedies both depend on a registration covering the goods actually being sold. 15 U.S.C. § 1057(b) gives the registration prima facie evidentiary weight that shortens every later step.
2. Build the notice record deliberately. Specific, accurate, listing-level notices are the currency of contributory liability. A letter saying "your site is full of counterfeits" accomplishes nothing after Tiffany; a spreadsheet of fifty listings with the reasons each is counterfeit, sent weekly, builds both a removal history and, if the platform stops responding, a willful blindness case.
3. Target the seller, not the platform, unless the platform fails. The platform is the defendant of last resort, and naming it in the first complaint usually buys an expensive motion to dismiss and a hostile relationship with the entity whose cooperation you need.
4. Follow the money. Payment processors and marketplace account balances are where recoveries actually come from. Identify them through test purchases and platform disclosures before filing.
5. Separate counterfeit from gray market from merely unauthorized. Each has a different legal theory, a different remedy, and a different risk of a false-report claim. The cosmetics example below shows the sorting.
6. Keep keyword disputes in proportion. Most are resolved by a letter asking the competitor to add your mark as a negative keyword and to keep it out of ad text. Litigation makes sense only where the ad uses the mark, impersonates you, or diverts measurable sales.

For the seller, advertiser, or platform

1. Advertisers: bid freely, keep the mark out of headlines and display URLs, label comparative claims as comparisons, and document that your landing page clearly identifies you. Monitor affiliates and act on complaints within days, not months (Lens.com).
2. Resellers: keep purchase records proving the goods are genuine and the supply chain intact; do not alter packaging or remove codes; describe condition accurately; respond to takedowns with proof of authenticity rather than silence, since silence reads as guilt to platforms and courts alike.
3. Platforms and hosts: build a notice channel, respond within a defined period, keep records of every notice and action, implement a written repeat-infringer policy and actually apply it (BMG v. Cox is the cautionary tale), investigate red flags about particular sellers rather than waiting for a brand to spell them out, and do not design products or promotions around third-party content in a way that makes you a seller rather than a venue (Ohio State). Comply with INFORM's verification duties; a seller you cannot identify is a seller for whom you may end up answering. The platform-side documents cover the architecture: Running a Platform That Hosts User Content: A Practical Guide and Platform Liability Toolkit: Terms of Service, Notice Procedures, and Litigation Defenses.

Common mistakes

• Sending general complaints and expecting specific results. After Tiffany, "you know there are fakes on your site" is not notice of anything.
• Reporting genuine goods as counterfeit. It forfeits credibility with the platform, invites a counterclaim, and can be false advertising or tortious interference in its own right.
• Suing the search engine over a labeled competitor ad. The Network Automation line makes that claim close to hopeless absent the mark in the ad text.
• Treating the DMCA process as a trademark process. A § 512 notice for a trademark complaint is defective; platforms have separate forms, and a knowingly false DMCA notice carries § 512(f) liability.
• Neglecting the chain of custody on test purchases. The counterfeit in the exhibit box has to be the one that arrived from the seller, and someone has to be able to say so under oath.
• Platforms: writing a policy and not applying it. Courts look at what was done with the notices, not at the terms of service.
• Forgetting Section 230 does not apply. Platform counsel who lead with § 230 in a Lanham Act case lose the motion and the court's patience.
• Pleading vicarious liability by reflex. Without an actual or apparent partnership, authority to bind, or joint control over the infringing product, the claim invites a motion to dismiss and tells the court the complaint came from a form. Hard Rock; Kelly-Brown.
• Ignoring the affiliate channel. An advertiser answers for affiliates it knows are putting a competitor's mark in ad copy if it does not act promptly (Lens.com). A no-trademark clause in the affiliate agreement, plus a way to message every affiliate at once, is cheap insurance.
• Treating a platform's decision as a ruling. A takedown, or a refusal to take one down, decides nothing about infringement. A seller whose listing comes down should answer with evidence of authenticity; a brand whose notice is rejected should ask whether the platform is right before escalating.
• Waiting. Months of tolerating known counterfeits invites the argument that the statutory presumption of irreparable harm in 15 U.S.C. § 1116(a) has been rebutted, and courts may weigh that delay (Y.Y.G.M.). See Preliminary Injunctions and Temporary Restraining Orders: Emergency Relief in Federal Court.
• Litigating where a cheaper forum exists. A counterfeit storefront operating on its own domain is often faster to reach through the registrar, a Uniform Rapid Suspension complaint, or an in rem action than through a lawsuit against an anonymous operator. See Choosing the Right Domain Name Dispute Procedure: A Practical Guide.

Worked examples

Example 1: A cosmetics brand fighting counterfeits on three marketplaces

Fernhill Skin (an invented company) sells a $68 vitamin C serum under the registered word mark FERNHILL GLOW and a registered label design, through its own website, about two hundred authorized salons, and its own Amazon storefront. In March its monitoring vendor reports 420 listings for "Fernhill Glow" across Amazon, eBay, and TikTok Shop, many priced between $19 and $29, along with a spike in one-star reviews complaining of a sour smell and a burning sensation. The founder wants every listing gone by Friday.

Step 1: Sort before you shoot. Fernhill's counsel resists the urge to report everything as counterfeit. Fourteen test purchases from the cheapest and highest-volume sellers sort the listings into four piles.

The sorting matters because the remedies and the risks differ by pile. Reporting the salons' genuine units as counterfeit would be a false notice. Reporting the Korean-market units as counterfeit would be inaccurate even though they may well infringe. And the expired-goods pile requires Fernhill to prove what Warner-Lambert demands: legitimate, substantial, and nonpretextual quality-control procedures that it actually follows, and sales of nonconforming goods that will diminish the value of the mark. Fernhill's shelf-life policy, its lot-tracking records, and its practice of pulling aged stock from salons are the evidence.

Step 2: Use each platform's program, and keep the file. Fernhill is enrolled in Amazon Brand Registry, so it reports counterfeit listings through the brand tools, applies Transparency codes to its next production run, and asks Amazon to act against the eleven seller accounts whose test buys were counterfeit, not just against their listings. On eBay it files VeRO notices for the counterfeit listings and asks for the sellers' contact information. On TikTok Shop it reports the listings and the affiliate videos promoting them through the platform's intellectual property portal. Every notice goes into a log with the listing ID, seller name, date, reason, test-buy reference, and the platform's response. If one platform keeps reinstating a storefront Fernhill has documented three times, the log becomes the willful-blindness record that Omega and Akanoc describe. If the platforms respond promptly, the same log is their Tiffany defense, and Fernhill will not need to sue them.

Step 3: Separate the sellers by who and where they are. INFORM disclosures on the high-volume sellers' listings identify nine U.S. businesses. Each gets a demand letter written for its pile: counterfeit sellers are asked for an accounting, the name of their supplier, and destruction of inventory; gray market sellers are told what makes the Korean units materially different; the expired-goods seller is shown the shelf-life policy. For the letters themselves, see Drafting a Trademark Cease-and-Desist Letter. Sixty-one storefronts sell counterfeits from overseas under accounts that disclose only a business name and a foreign address. For those, Fernhill files a single Schedule A complaint in the Northern District of Illinois after confirming that each storefront shipped a test purchase to Illinois, the jurisdictional theory upheld in NBA Properties. It obtains a sealed TRO under Fed. R. Civ. P. 65 freezing the storefronts' marketplace balances, serves by email under Rule 4(f)(3) where the Hague Service Convention does not stand in the way (see Serving a Foreign Defendant Under the Hague Convention: A Practical Checklist), and moves for default judgments seeking statutory damages under 15 U.S.C. § 1117(c). Counsel keeps the joinder story honest: the complaint pleads the shared product photographs, identical misspellings, and overlapping payment accounts that tie the sellers together, because a court that doubts joinder under Rule 20 can sever the case under Fed. R. Civ. P. 21 and multiply the filing fees.

Step 4: Close the border and the domains. Fernhill records its registrations with U.S. Customs and Border Protection under 19 C.F.R. Part 133, which lets officers detain and seize counterfeit shipments at entry under 19 U.S.C. § 1526 and 15 U.S.C. § 1124. It also files a Uniform Rapid Suspension complaint against a lookalike domain that a counterfeiter used to sell the serum directly, a remedy explained in Beyond the UDRP: Uniform Rapid Suspension, Country-Code Domain Disputes, In Rem ACPA Actions, and Reverse Domain Name Hijacking.

Step 5: Decide what not to do. Fernhill does not sue Amazon, eBay, or TikTok; all three responded to specific notices within days, which under Tiffany is close to a complete defense. It does not report the salons. And it does not send letters to dupe sellers merely for naming its product, because Smith v. Chanel allows a seller to use a competitor's mark to identify the product it copied, so long as the comparison does not confuse buyers about source. Two dupe sellers are different: one claims its serum has "the same formula," which is a false-advertising problem if untrue (see False Advertising Under the Lanham Act: Literal Falsity, Materiality, and Competitor Standing), and the other uses Fernhill's label design in its product images, which is a likelihood-of-confusion problem. Those two get letters.

Result. Within ninety days the counterfeit listings drop by roughly three quarters, most of the domestic sellers comply, the Schedule A case produces default judgments with damages collected from the frozen balances, and the remaining fight is with two gray-market resellers over whether Korean labeling is a material difference. That is a realistic outcome. Online enforcement is never finished, but it can be made expensive enough that most counterfeiters move on to softer targets.

Example 2: A competitor buys your brand as a keyword

Quillstone (also invented) sells scheduling software to dental practices under the registered mark QUILLSTONE. Its marketing director notices that searches for "quillstone" and "quillstone pricing" now show, at the top of the page, a sponsored ad for a competitor reading "TaskForge Scheduling: Built for Busy Practices." Two prospects tell Quillstone's sales team they "thought TaskForge was Quillstone's new version." The CEO wants to sue.

The first conversation. The keyword purchase alone is not a winning claim. In the Second Circuit it would likely be dismissed on the pleadings (JAND). In the Ninth, the Network Automation factors would focus on the ad's labeling and content, and an ad that never uses QUILLSTONE and names TaskForge looks like Lerner & Rowe. In the Tenth, Lens.com would ask for the numbers, and two confused prospects against several thousand impressions is what those courts call de minimis. Counsel also flags the risk running the other way: aggressive litigation over a clean ad could produce an adverse ruling and a fee motion under 15 U.S.C. § 1117(a) (see Lanham Act Attorney's Fees Under 15 U.S.C. § 1117(a): A Practical Guide). The practical options are cheaper: bid on QUILLSTONE so Quillstone's own ad appears first, strengthen the organic result, and record every confused-prospect report in detail in case the facts change.

The facts change. Three months later TaskForge's ads read "Quillstone Login | Official Scheduling Site" and click through to a landing page in Quillstone's shade of teal with the headline "Welcome back, Quillstone users." That is a different case. The mark is now in the ad text, "Login" and "Official" assert a connection that does not exist, and the landing page borrows the look of Quillstone's site. Under Lens.com, Rosetta Stone, and the labeling factor of Network Automation, ad copy that tells consumers they are reaching the mark owner is exactly what makes a keyword claim viable. Counsel captures the ads and the landing page with dates and a process that will satisfy Fed. R. Evid. 901 (see Capturing the Web: A Practitioner's Guide to Authenticating Website Screenshots as Evidence in Federal Court), files the ad platform's trademark complaint about the ad text, and sends a cease-and-desist letter. The letter demands that TaskForge remove QUILLSTONE from ad text and display URLs, stop using "official" and "login," and change the landing page. It does not demand that TaskForge stop bidding on the keyword. That demand would be weak on the merits, and an agreement among competitors to stop bidding raises the antitrust questions that 1-800 Contacts v. FTC took years to resolve.

TaskForge's side. TaskForge's counsel advises it to fix the ads within days rather than fight about them. Comparative advertising is lawful: "Switching from Quillstone? TaskForge imports your data in minutes" uses only the word mark, uses it to refer to Quillstone, and suggests no sponsorship, which is the nominative-use pattern of New Kids on the Block v. News America Publishing, Inc., 971 F.2d 302 (9th Cir. 1992) and Toyota Motor Sales, U.S.A., Inc. v. Tabari, 610 F.3d 1171 (9th Cir. 2010). "Official site" and "login" are not. TaskForge also preserves the argument from Judge Desai's Lerner & Rowe concurrence that bidding on a keyword is not a use at all, and documents each ad-copy change so that any damages period ends on a provable date.

Resolution. The parties settle on ad-copy rules rather than bidding rules. TaskForge may bid on QUILLSTONE but may use it in ad text only in an explicitly comparative phrase, may never use "official," "login," or Quillstone's colors, and must name TaskForge in every headline. Quillstone gets what it actually needed, the end of the impersonation, without a restraint on bidding that an antitrust enforcer might second-guess. Had TaskForge refused, Quillstone would have sued under 15 U.S.C. § 1114 and 15 U.S.C. § 1125(a), moved for a preliminary injunction with the benefit of the rebuttable presumption of irreparable harm in § 1116(a), and commissioned a survey that replicates the results page and uses a proper control instead of repeating the Lens.com survey's mistakes.

Frequently asked questions

Can I sue a competitor for buying my trademark as a search keyword? You can, but the purchase alone rarely wins. Courts treat the sale of a mark as a keyword as a use in commerce (Rescuecom Corp. v. Google Inc., 562 F.3d 123 (2d Cir. 2009)), then ask whether the ad the keyword triggers is likely to confuse consumers about who is behind it. A labeled ad that names the competitor and keeps your mark out of its text usually is not, and the Second Circuit now dismisses bare keyword claims on the pleadings (1-800 Contacts, Inc. v. JAND, Inc., 119 F.4th 234 (2d Cir. 2024)). The claim becomes real when the ad puts your mark in the headline, calls itself "official," lands on a page dressed in your trade dress, or produces confusion that is measurable against the impressions it generated. Before suing, capture the ads properly, collect your own call and chat logs, and try the cheaper tools first: bid on your own mark, file an ad-text complaint with the ad platform, and send a letter that targets ad copy rather than bidding.

Is Amazon, eBay, or Etsy liable for counterfeits sold by third-party sellers? Usually not, and the reason is a knowledge standard rather than a statute. Under Inwood Laboratories, Inc. v. Ives Laboratories, Inc., 456 U.S. 844 (1982), as Tiffany applied it to a marketplace, the platform is contributorily liable only if it induced the infringement or kept serving a seller it knew or had reason to know was infringing, and the knowledge has to concern particular listings or sellers. A marketplace that removes reported listings quickly, acts against repeat offenders, and does not deliberately avoid confirming what it suspects about specific sellers has a strong defense. Exposure grows when it ignores specific notices (Akanoc), keeps restoring storefronts it has been told are selling fakes, or takes on the seller's role by producing, branding, and shipping the goods itself (Ohio State University v. Redbubble, Inc., 989 F.3d 435 (6th Cir. 2021)).

Don't the DMCA or Section 230 protect platforms anyway? No. The safe harbors in 17 U.S.C. § 512 cover copyright claims only, and 47 U.S.C. § 230(e)(2) withholds the publisher immunity from any law pertaining to intellectual property, which every court reads to include the Lanham Act. A platform's protection against trademark claims is therefore factual: what it was told, what it did, and how quickly. One narrow statutory limit does exist. Under 15 U.S.C. § 1114(2)(B), an innocent publisher or distributor of infringing paid advertising, including advertising carried in an electronic communication, faces only an injunction against future publication. The limit applies only to innocent infringers, and a platform that has received specific notices will have trouble claiming innocence for long.

How specific does a takedown notice have to be? Specific enough that the recipient can find the infringement and evaluate it without doing the brand owner's investigation for it. A good notice identifies the registration, the listing or seller by ID or URL, the reason the item is counterfeit or infringing (a test-buy result, a batch code that matches no production run, an unauthorized logo), and the sender's authority. A statement that "most of the X merchandise on your site is fake" is the kind of generalized knowledge Tiffany held insufficient. The outer edge is marked by Omega SA v. 375 Canal, LLC, 984 F.3d 244 (2d Cir. 2021), and 1-800 Contacts, Inc. v. Lens.com, Inc., 722 F.3d 1229 (10th Cir. 2013): a defendant cannot ignore red flags about a known location or a known advertising channel merely because the brand has not named the individual infringer, particularly where one targeted step would have stopped the conduct.

Can a platform or landlord owe statutory damages for counterfeits it never sold? Yes. A contributory infringer answers for the direct infringer's remedies, including statutory damages for counterfeits under 15 U.S.C. § 1117(c). The web host and its principal in Akanoc ended up jointly liable for $10.5 million on the trademark claims, the flea market operator in Coach, Inc. v. Goodfellow, 717 F.3d 498 (6th Cir. 2013), for $5.04 million, and the Canal Street landlord in Omega for $1.1 million. Each had specific notice and did too little with it. Y.Y.G.M. SA v. Redbubble, Inc., 75 F.4th 995 (9th Cir. 2023), shows the same theory reaching a jury against a print-on-demand marketplace, with the Ninth Circuit insisting on knowledge of specific infringers or infringements, or genuine willful blindness, before liability attaches.

I resell genuine products and my listings keep coming down. What can I do? Answer with documents. Platforms usually reinstate listings when a seller shows invoices from an authorized source, unaltered packaging, and intact serial or batch codes, and a brand that reports genuine goods as counterfeit puts its own reporting privileges at risk. Know the limits of the first sale defense, though. Goods that differ materially from what the brand sells in the United States are not "genuine" for Lanham Act purposes, and courts count more than physical differences: in Beltronics USA, Inc. v. Midwest Inventory Distribution, LLC, 562 F.3d 1067 (10th Cir. 2009), the Tenth Circuit affirmed an injunction against an eBay seller of radar detectors whose serial labels had been removed or replaced, because buyers lost the warranty, upgrades, and service that came with the genuine article. If the brand writes to you, Responding to a Trademark Cease-and-Desist Letter covers the response. A seller who believes the reports are knowingly false may have claims of its own, such as tortious interference, but those are slow and costly; a documented appeal to the platform is usually faster.

I was named in a Schedule A case and my marketplace account is frozen. Now what? Act within days and retain counsel who can appear in that court. The TRO usually restrains the whole marketplace balance, not only the proceeds of the accused product, and the plaintiff's settlement demand often assumes you will never appear. The defenses that work are concrete: the product is genuine or does not infringe; the court lacks personal jurisdiction because nothing was ever sold or shipped into the forum (jurisdiction in NBA Properties, Inc. v. HANWJH, 46 F.4th 614 (7th Cir. 2022), rested on an actual sale shipped to Illinois); the defendants were improperly joined under Fed. R. Civ. P. 20; the freeze far exceeds any plausible profits from the accused sales; or the asserted registration does not cover the goods. A growing number of judges have severed misjoined defendants, limited restraints to amounts tied to the accused sales, or required larger bonds. Counterfeiting, Seizure Orders, and Schedule A Litigation: Fighting Fakes in Federal Court covers the defense playbook.

Can I find out who an anonymous seller really is? Often, and more easily than before 2023. The INFORM Consumers Act requires marketplaces to collect and verify high-volume sellers' bank, tax, and contact information and to show the name and address of the larger ones on their listings. Where the listing does not reveal the seller, a subpoena to the marketplace or payment processor can, but subpoenas require a pending case, and discovery before the Rule 26(f) conference requires leave of court under Fed. R. Civ. P. 26(d). Courts grant leave for good cause, usually on a showing of a prima facie claim, a narrowly drawn request, and no other way to identify the defendant, using the factors from Sony Music Entertainment Inc. v. Does 1-40, 326 F. Supp. 2d 556 (S.D.N.Y. 2004), which the Second Circuit adopted in Arista Records, LLC v. Doe 3, 604 F.3d 110 (2d Cir. 2010), and the older approach of Columbia Insurance Co. v. Seescandy.com, 185 F.R.D. 573 (N.D. Cal. 1999). The procedure is explained in Identifying an Anonymous Defendant Online: A Practical Guide and, for the First Amendment overlay when the anonymous party is a speaker rather than a seller, Unmasking Anonymous Online Speakers: John Doe Subpoenas, the Dendrite and Cahill Tests, and the First Amendment.

Do I need a federal registration to enforce my mark online? In practice, yes. Section 43(a) protects unregistered marks, but almost every marketplace program requires a registration (or at least a pending application) for enrollment; the counterfeiting remedies, including treble damages, statutory damages, and ex parte seizure, apply only to counterfeits of a mark registered on the Principal Register for the goods in question; and Customs recordation under 19 C.F.R. Part 133 requires a registration. A registration also supplies prima facie evidence of validity and ownership under 15 U.S.C. § 1057(b) and 15 U.S.C. § 1115(a), which shortens every motion that follows. If a counterfeiter has registered your brand first to take over a marketplace listing, the answers are a petition to cancel under 15 U.S.C. § 1064 and, where the registration was procured by fraud, a damages claim under 15 U.S.C. § 1120; see Petitioning to Cancel a Trademark Registration: A Practical Checklist.

Practical takeaways

• In keyword disputes, the ad is the case, not the bid. Police ad text, display URLs, and landing pages, and leave the bidding alone. A non-bidding agreement between competitors is an antitrust question as much as a trademark one (1-800 Contacts v. FTC).
• Numbers decide keyword cases. Plaintiffs need confused-consumer evidence tied to the ads; defendants should insist on the denominator. Impressions, clicks, and a survey that replicates the results page with a proper control are the evidence that matters (Lens.com; Lerner & Rowe).
• Contributory liability is a knowledge standard. General awareness of counterfeiting on a platform is not enough; specific notice about particular listings or sellers is what converts a bystander into a potential defendant (Tiffany; Y.Y.G.M.).
• Willful blindness is the bridge between the two. A defendant that suspects specific infringement and deliberately avoids confirming it is treated as knowing, and it need not know the infringer's name (Omega).
• Control matters, especially in the Ninth Circuit. Hosts that hold the "master switch" are exposed (Akanoc); registrars and payment networks that provide routine services usually are not (Lockheed; Visa), though processors essential to counterfeit sales can be (Frontline).
• There is no safe harbor, only a record. Platforms should log every notice, action, and response time and actually apply their repeat-offender policies. Brand owners should keep the same log from their side; it is either the platform's Tiffany defense or the plaintiff's willful blindness case.
• Platforms that make and brand the goods risk direct liability. Print-on-demand and fulfillment models that put the platform's name on the package invite the Ohio State theory.
• Sort before you report. Counterfeit, gray market, expired or decoded, and lawful resale call for different theories and different tools, and a false counterfeit report carries real costs.
• Build every matter on authenticated test purchases. Capture the listing, buy, document the unboxing, compare with a genuine exemplar, and keep a chain-of-custody log a witness can swear to.
• Use Schedule A litigation for what it does well. It is efficient against anonymous foreign sellers at scale, but it now draws scrutiny: test-buy into the forum, plead joinder facts honestly, and keep asset restraints proportionate.
• Register, then record. A registration covering the goods actually sold unlocks platform programs, counterfeiting remedies, and Customs recordation all at once.
• Defendants should preserve their arguments. Keyword defendants should preserve the position in Judge Desai's Lerner & Rowe concurrence that bidding is not a use; resellers should keep the paperwork that proves authenticity and an intact supply chain.

This document is general information, not legal advice for any particular situation; the law changes, and the facts matter.

Related Documents

Related Articles

• Counterfeiting, Seizure Orders, and Schedule A Litigation: Fighting Fakes in Federal Court: the counterfeiting statutes, ex parte seizures, and the mechanics and critics of Schedule A practice.
• Brand Protection Online: A Strategic Guide for Businesses: building a monitoring and enforcement program across marketplaces, social media, and domains.
• Section 230 and Platform Liability: What the Statute Actually Says and Where It Stops: the immunity, its intellectual property exception, and the circuit split over state-law claims.
• Section 230 Reform and Platform Liability for User-Generated IP Infringement: reform proposals and how they would change platform exposure for infringing uploads.
• Gray Market Goods and the First Sale Doctrine: Parallel Imports, Material Differences, and Exhaustion: when resale of genuine goods is lawful and when material differences make it infringing.
• Trademark Fair Use: Descriptive Use, Nominative Use, and Expressive Works After Jack Daniel's: the defenses advertisers and resellers raise when they refer to another's mark.
• Unmasking Anonymous Online Speakers: John Doe Subpoenas, the Dendrite and Cahill Tests, and the First Amendment: the standards courts apply before ordering a platform to identify an anonymous user.
• Beyond the UDRP: Uniform Rapid Suspension, Country-Code Domain Disputes, In Rem ACPA Actions, and Reverse Domain Name Hijacking: the domain-side tools for counterfeit storefronts that operate on their own domains.
• Expungement, Reexamination, and Letters of Protest: The Trademark Modernization Act in Practice: clearing fraudulent or unused registrations that bad actors use to hijack marketplace brand programs.
• Indirect and Divided Patent Infringement: Inducement, Contributory Liability, and the Akamai Problem: the statutory patent analog to judge-made trademark secondary liability, useful for contrast.
• False Advertising Under the Lanham Act: Literal Falsity, Materiality, and Competitor Standing: the claim that reaches "same formula" dupe ads and misleading comparative copy.
• Consumer Survey Expert Methodology in Trademark Cases: A Practical Guide: designing confusion surveys that survive Rule 702, including results-page replicas.
• Capturing the Web: A Practitioner's Guide to Authenticating Website Screenshots as Evidence in Federal Court: making captures of ads, listings, and storefronts admissible.
• Antitrust for Technology Companies: Monopolization, Platform Conduct, and the IP-Antitrust Interface: the antitrust limits on trademark settlements and keyword agreements.

Related Guides

• Enforcing Trademarks Against Online Sellers and Platforms: A Practical Guide: the step-by-step companion to this article, from monitoring and test purchases through platform escalation and Schedule A litigation.
• Stopping Gray Market and Counterfeit Imports: A Practical Guide: Customs recordation, detentions, seizures, and gray market protection at the border.
• Running a Platform That Hosts User Content: A Practical Guide: the platform operator's side, including notice systems and repeat-infringer policies.
• Digital Millennium Copyright Act Safe Harbors for Online Service Providers: how the copyright safe harbor works and why trademark has none.
• Identifying an Anonymous Defendant Online: A Practical Guide: expedited discovery and platform subpoenas to unmask sellers.
• Asserting or Defeating a Trademark Fair Use Defense: A Practical Guide: building or attacking nominative and descriptive fair use defenses.
• Choosing the Right Domain Name Dispute Procedure: A Practical Guide: UDRP, URS, ccTLD procedures, and in rem actions compared.
• Filing or Responding to a Trademark Expungement or Reexamination Petition: A Practical Guide: the procedure for clearing registrations that block or hijack enforcement.

Related Checklists

• Online Marketplace Trademark Enforcement Checklist: A Practical Checklist: every step of a marketplace enforcement program in checklist form.
• Authenticating Website Evidence: A Practical Checklist: capture, preservation, and authentication steps for online evidence.
• Customs IP Enforcement Checklist: A Practical Checklist: recording marks with CBP and responding to detentions and seizures.
• Drafting a Trademark Cease-and-Desist Letter Checklist: A Practical Checklist: what a demand letter to an online seller or advertiser should contain.
• Trademark Likelihood-of-Confusion Analysis: A Practical Checklist: working through the confusion factors, including the internet-specific ones.
• Preliminary Injunction and TRO Application Checklist: A Practical Checklist: assembling emergency relief papers, including asset restraints.
• Trademark Fair Use Analysis Checklist: A Practical Checklist: screening comparative ads, reseller listings, and repair-shop uses.
• Anonymous Speaker Unmasking Checklist: A Practical Checklist: the steps and showings for a John Doe subpoena.

Related Toolkits

• Online Trademark Enforcement Toolkit: Platform Notices, Test Purchase Records, and Contributory Liability Demands: templates for notices, test-buy logs, chain-of-custody forms, and platform demand letters.
• Online Brand Protection Toolkit: A Strategic Roadmap and Resource Library: program design, monitoring, and escalation resources.
• Platform Liability Toolkit: Terms of Service, Notice Procedures, and Litigation Defenses: the platform-side documents that build a Tiffany record.
• Border Enforcement Toolkit: Recordations, Seizure Responses, and Gray Market Programs: recordation applications and responses to Customs notices.
• Trademark Enforcement and Litigation Toolkit: From Watching to Verdict and Appeal: the full litigation set, from watch notices to trial.
• John Doe Litigation Toolkit: Early Discovery Motions, Platform Subpoenas, and Motions to Quash: motions and subpoenas for identifying anonymous sellers and users.
• Trademark Fair Use Toolkit: Nominative Use Guidelines, Comparative Advertising Clearance, and Defense Briefing: clearance guidelines for comparative keyword ads and reseller copy.