Trademark Use in Advertising

What three appellate courts have now decided about buying a competitor's name

A keyword is the word a user types that an advertiser bids on to trigger an ad. Whether a competitor may bid on a brand's name is the question that brings most of these matters in the door. Three federal appellate courts have answered it on the merits within two years, and all three answered the same way.

The Second Circuit affirmed judgment for Warby Parker in 1-800 Contacts, Inc. v. JAND, Inc. (8 October 2024), holding that "the mere act of purchasing a competitor's trademarks in the context of keyword search advertising does not constitute trademark infringement." The ads and landing pages carried Warby Parker branding, the ad was labeled "Ad," and the display URL was the defendant's own. The plaintiff offered no consumer survey, which the court noted is how claimants typically prove actual confusion.

Two weeks later the Ninth Circuit affirmed summary judgment in Lerner & Rowe PC v. Brown, Engstrand & Shely LLC, 119 F.4th 711 (22 October 2024). The arithmetic is the useful part: 236 phone calls mentioning the plaintiff's name against 109,322 ad impressions from 2017 to 2021, a rate of 0.216 percent, alongside a 6.82 percent click-through rate and a defense survey reporting 0 to 3 percent confusion. Judge Desai concurred in full but urged en banc reconsideration of whether buying a keyword is "use in commerce" at all — the live doctrinal edge here.

The teaching case, because one opinion did both things

On 4 August 2026 the Eleventh Circuit published Deltona Transformer Corp. v. The NOCO Company. The keyword-bidding theory failed: the court held that "keyword bidding doesn't constitute trademark infringement" because "consumers don't — indeed, can't — see the plaintiff's mark — all that's visible here is a NOCO ad," and that likelihood of confusion "turn[s] on what the consumer s[ees] on the screen." That framing tells an expert which artifact to reconstruct.

The infringement verdict was nonetheless upheld, on a different theory. A jury could find infringement where the defendant "used Deltona's protectable mark in the text of NOCO's Amazon ads," including one reading "The most advanced battery tender for any vehicle," combined with internal communications showing an intent to appropriate goodwill while knowing the term was a trademark. The false advertising count was separately reversed on pleading grounds, as unfair competition pleaded and false advertising tried.

The disposition matters as much as the holdings. A $1.3 million lump-sum verdict spanned five theories and three were eliminated, so the court remanded for a new trial on damages, unable to "determine the extent to which the $1.3 million damages award reflects liability for any of those claims or theories." Apportionment is a damages expert's problem, but factual findings organized theory by theory are what make it possible at all.

The most cited case in this area decided the least

Rescuecom Corp. v. Google Inc., 562 F.3d 123 (2d Cir. 2009), is routinely described as having held that keyword advertising on a competitor's trademark is infringement. It held nothing of the kind.

The appeal was from a Rule 12(b)(6) dismissal. The Second Circuit held that Google's sale and recommendation of the mark to advertisers is a "use in commerce" within the Lanham Act, sufficient to state a claim, and vacated the dismissal. On the element that actually decides infringement it said, three separate ways, that it was deciding nothing: it had "no idea whether Rescuecom can prove" that Google's use caused "likelihood of confusion or mistake"; whether the practice was "in fact benign or confusing" was not for the court to judge at that stage; and it "express[ed] no view as to whether Rescuecom can prove a Lanham Act violation," holding only that "an actionable claim is adequately alleged." No merits ruling on confusion ever followed.

Two further traps. The opinion's appendix on the history of "use in commerce" labels itself "dictum and not a binding opinion of the court." And at least four reported decisions carry 1-800 Contacts in the caption — WhenU.com (2d Cir. 2005, pop-up software), Lens.com (10th Cir. 2013, keywords and affiliates), the FTC (2d Cir. 2021, antitrust) and JAND (2d Cir. 2024, keywords) — and they are about four different things. Never write "the 1-800 Contacts case" without the year and the adversary.

The Ninth Circuit's framework, and a results page that defeated confusion

Network Automation, Inc. v. Advanced Systems Concepts, Inc., 638 F.3d 1137 (9th Cir. 2011), is a framework case, not a holding that keyword advertising is lawful; it reversed a preliminary injunction. It identified the four factors most relevant in keyword cases: "(1) the strength of the mark; (2) the evidence of actual confusion; (3) the type of goods and degree of care likely to be exercised by the purchaser; and (4) the labeling and appearance of the advertisements and the surrounding context on the screen displaying the results page." It narrowed initial interest confusion, holding that "the owner of the mark must demonstrate likely confusion, not mere diversion," and the fourth factor is an open invitation to evidence: labeling and appearance "includes more than the text of the advertisement, and must be considered as a whole."

Multi Time Machine, Inc. v. Amazon.com, Inc., 804 F.3d 930 (9th Cir. 2015), applied that to a marketplace results page and affirmed summary judgment for Amazon: labeling by brand, model and photograph meant "no reasonably prudent consumer" would likely be confused about source. The procedural history is where citations go wrong. The panel first issued an opinion on 6 July 2015 reversing summary judgment, then amended and superseded it on 21 October 2015, affirming, with the two judges' positions swapped between them. Citing "Multi Time Machine, 9th Cir. 2015" without specifying the amended October opinion risks citing a superseded result that reached the opposite conclusion.

An antitrust decision that keeps being filed under trademark

1-800 Contacts, Inc. v. Federal Trade Commission, 1 F.4th 102 (2d Cir. 2021), is the most mis-described case in this area. It is an antitrust decision under Section 5 of the FTC Act, and no trademark claim was adjudicated.

The Commission had alleged that trademark settlement agreements between 1-800 Contacts and rival sellers, in which the parties agreed not to bid on each other's marks and to use negative keywords, unreasonably restrained trade. The Second Circuit held the wrong analytical framework had been applied: the agreements "cannot be classified as inherently suspect," and a full rule-of-reason analysis was required. The order was vacated and the matter remanded with instructions to dismiss the administrative complaint.

Five errors show up in the wild, each worth catching in an opposing report:

• That it held agreements not to bid on a competitor's mark are lawful. It held the Commission used the wrong mode of analysis and did not carry its burden on that record.
• That it is a trademark case. The question was whether Section 5 was violated.
• That it makes trademark settlements antitrust-proof. The opinion says the opposite: "trademark settlement agreements are not immune from antitrust scrutiny."
• That it settles whether a brand can demand a competitor stop bidding on its name. That belongs to the Network Automation line, which cuts against the brand.
• That it is part of the FTC's fake-review or endorsement work. Different part of the agency's docket entirely.

Where the evidence sits: ad text versus keyword targeting

The doctrine has converged on what appeared on the screen. That record divides in two.

Targeting configuration lives in the ad account: keyword lists with match types, negative keyword lists, the search terms report showing the queries that actually triggered the ad, audience and placement settings, and the change history logging who changed what and when. That answers whether a mark was bid on, when it was added, whether it was ever removed, and what queries the campaign in fact served against.

Creative is the half that carries the liability: headlines and descriptions, the display URL, extensions and assets, the combinations the platform actually served rather than the ones drafted, the landing page, and on marketplaces the listing title and detail page. In Deltona the mark in the Amazon ad text is what survived. In 1-800 Contacts, Inc. v. Lens.com, 722 F.3d 1229 (10th Cir. 2013), keyword purchase alone did not infringe, but summary judgment was reversed as to contributory liability for affiliate advertisers whose ad text displayed the mark, the court noting the defendant could have sent "an email blast to its affiliates forbidding such use" without identifying the culprit. For affiliate programs and agency-managed accounts, that second half is the operative one.

The platforms' own rules, and what a violation of one is worth

A platform policy is a private operational rule, not law. A finding that an advertiser violated Google's trademark policy is not a finding of infringement, and Google declining to act on a complaint is not a finding of non-infringement. That distinction is worth explaining to a jury.

Google's Trademarks policy does not restrict trademarks as keywords. It restricts trademarks in ad text on complaint, specifically for ads from a direct competitor and ads using the mark in a confusing, deceptive or misleading way, and it is explicit that "[t]he trademark must be used in the ad, not only on the ad's landing page." Reseller and informational-site exceptions turn on what the landing page primarily does.

Microsoft Advertising is harder to pin down. Its complaint intake asks for the ad title, ad copy and display URL, so the process is built around creative elements, and its complaint page states that Microsoft "does not remove organic search results in response to a trademark complaint" because that content is controlled by the site owner. Its substantive policy documentation moved during a help-center migration, and I could not retrieve current text on whether keyword complaints are accepted — which is the lesson: capture and date-stamp the policy page you rely on.

What the advertising record does not settle

I do not opine that confusion was likely. That is the legal test the finder of fact applies under the circuit's multi-factor framework, and testifying to it trades a factual expertise for a legal conclusion. Nor do I supply the survey; survey design and interpretation is a distinct discipline with its own training expectations, described in the Reference Manual on Scientific Evidence, fourth edition (2025). Surveys are not a trump card in any event: Lens.com gave a 7.4 percent net confusion survey "no more than minimal weight," while JAND faulted a plaintiff for offering none.

Platform metrics measure delivery, not belief. A click-through rate speaks to the magnitude of diversion, which is how both Lens.com and Lerner & Rowe used it; converting it into a confusion rate invents a bridge the cases do not build.

Reconstruction is partial: ads are personalized, auction-dependent and asset-combined, so what one user saw is not what another saw, and retention windows limit what can be recovered. And the law here is circuit-specific and moving: a conclusion drawn from Ninth Circuit authority does not travel automatically to the Second, Tenth or Eleventh; and the doctrine moved twice in 2024 and again in August 2026. Any statement of the law in this area carries a date.

Frequently Asked Questions

Published 2026-08-12