On 22 May 2026 the Delhi High Court held that Google infringes the trademark HINDWARE by letting rival sanitary ware sellers bid on that word as a Google Ads keyword. Arul Murugan’s detailed post summarizing the dispute and holdings is here. This post is more about its critique.
In a rare final judgment or decree, the HC moves in three steps – allowing to bid on a keyword is “use” of the mark; that keyword is used by Google and not merely by the advertiser; and this use infringes under Section 29(8) with, in terms, no likelihood of confusion required. The advertisers who actually bought the keyword had settled, so the only conduct left holding up the decree is Google running an auction over a word. Importantly, after a full trial, the Court could not find a single confused consumer or a single diverted sale, which is why damages awarded are “nominal” and not towards the actual loss, if any.
In essence, the Court treats a trademark as a property right in a word, enforceable against anyone who profits from the word’s pulling power, and it reaches that result walking past the one requirement that keeps trademark law tethered to its purpose – that the mark be used “in relation to goods or services,” to indicate source. Once that requirement is loosened, a registration over HINDWARE for sanitaryware turns into a monopoly over the signifier “HINDWARE” in every commercial context, including the context of a search query and a labelled ad, de hors any cross class protection. The Trade Marks Act does not grant that, and the “investment function” theory the Court borrows to grant it has neither a historical nor a theoretical leg to stand on. Even the CJEUesque producer-centred tradition of trademark law disciplined itself with two limits where diversion was actionable only where it was brought about by deception, which is the passing-off core, and the property the courts recognized was a narrow one, tied to actual competition and to honestly acquired custom, never a right in the mark’s value at large. A Court that wants to protect promotional spend as a thing in itself has to earn that on its own terms, and it cannot keep borrowing the moral weight of “consumers must not be deceived” while quietly setting deception aside. Moreover, the investment function is affected only where there is substantial interference with the proprietor’s ability to acquire or preserve a reputation. The mere fact that keyword use leads some consumers to switch to a competitor is not, by itself, an injury to the investment function.
What the court has granted is essentially a semiotic monopoly, and semiotic monopoly is the one thing source-identification or consumer deception has never been about. It does not help the Court’s case that HINDWARE, which it keeps calling a coined word with “no dictionary meaning” is really “Hind” and “ware” pushed together; and even at its most distinctive, distinctiveness buys a stronger claim against confusion as to source, never a claim over the word at large.
Keywords doctrinally
I want to take the keyword question on its own terms. A keyword is a backend matching input. The advertiser hands Google a word, Google’s algorithm uses it to decide whether the advertiser’s labelled advertisement (clearly labelled as sponsored) is eligible to surface when someone searches that word, and if it surfaces and the searcher clicks, the advertiser pays. The searcher never sees the keyword. It is a string in a matching table, not a sign on a product.
The Court reads Section 2(2)(c)(i), use “in any other relation whatsoever” to goods, to cover invisible keyword use, and reads “advertising” in Section 29(6)(d) as a verb wide enough to swallow the whole process. But “is this use” and “is this use of a mark as a mark, to indicate source” are two different questions, and Section 2(1)(zb) answers the second – a trade mark is a sign used “for the purpose of indicating a connection in the course of trade between the goods and the proprietor.” A keyword indicates nothing to anyone, because nobody sees it in any relation to any good. Use of a word for the purposes of trademark law is tied to a product, to the word affixed to or presented with goods as their identifier. The sale of the word itself, as an input that an advertiser later buys, is the trading of a string, not the use of a mark on goods. Thus, a search engine that stores a sign and is paid for it “does not itself use the sign.”
The second holding, that the keyword is “use by Google,” rests on the same confusion of categories. Google does not sell sanitaryware. It sells an advertising service, and it is no competitor of HSIL in any market where HINDWARE identifies a source. HSIL are not proprietors of well-known marks and they do not have across-class to cover classes where they are not even registered – i.e., the class that covers offering internet advertising services. The Court bridges the gap by making diversion the hinge holding that the bid amount becomes payable to Google when there is diversion of traffic. The wrong the Court describes is the diversion of a searcher’s attention to a competitor, and it has assumed that diverting attention is the same as deceiving a consumer, without analyzing the harm involved.
Imagine, in a shop, I walk in and ask for Maggi. The shopkeeper, who earns a better margin on Knorr, says “we have this new Knorr soupy noodles, want to try it?” I say no, and he hands me my Maggi. Has anyone been deceived? Has Maggi’s mark been infringed? Of course not, and it makes no difference at all that the shopkeeper was paid by Knorr to make the offer. A Google results page is that shop. The rival’s advertisement sits under a label that reads “Sponsored,” the searcher’s own Hindware result sits in the organic list right below it, and both are on the same page at the same moment. A searcher who sees a labelled alternative and clicks it has made a choice, she has not been deceived about whose goods are whose. The Court’s contrary picture, that a search for HINDWARE is equivalent to the user entering the specific exclusive store of the trademark owner, simply assumes the searcher is a captured customer, whereas this consumer has only typed a word into Google, and one cannot read a single objective off a keyword. The people typing HINDWARE want the official site, or a price, or a review, or a complaint, or a spare part, or a comparison with rivals, and trademark law cannot presume that all of them wanted only the proprietor and must be sealed off from everything else.
Once we separate diversion from deception, the free-riding logic of the judgment collapses. The Court treats Google’s earning from the auction as “unfair advantage” because the advertiser “saved” on building its own brand. But every advertiser who places an ad where interested customers are looking saves that cost, and a benefit captured without confusion is no trademark wrong. Free-riding, standing alone, is mostly just competition, and competition that begins with the consumer’s awareness of the market leader is the ordinary condition of every market. It is what is commonly known as fair riding, and diversion of this sort is frequently a positive externality, because it widens the consumer’s field of choice at the very moment she is deciding. The “stealing the customer” anxiety that drives the judgment is a genuine worry for the proprietor, but it becomes a trademark concern only when the customer is stolen by confusion or being deceived, and there is no confusion or deception here.
A keyword is a string, and the same string can be a registered mark for different proprietors in unrelated classes. ABC can be a mark for pet food and, in another register entirely, a mark for diamonds, each perfectly valid and neither confusing the other. An injunction restraining Google from offering “HINDWARE” as a keyword is therefore not a remedy tailored to a confusing use by a rival sanitaryware seller. It is a monopoly over the string itself inside Google’s auction, granted on one proprietor’s say-so. The question the Court ought to have asked is – did a particular advertiser use the mark, in relation to its own goods, in an advertisement likely to make a reasonable consumer believe those goods came from or were connected with the proprietor? That asks about the advertiser’s same-class use, and about confusion, and it is answerable case by case. The Court could not ask it, because the advertisers had settled the dispute.
The American courts that began with a diversion theory have spent a decade walking it back, in Network Automation, in 1-800 Contacts v Lens.com, where a 1.5 per cent click-through was too thin to show confusion and the court observed that those users “may merely have wished to look at the offerings of a competitor,” and in Multi Time Machine v Amazon, where clearly labelled results showing rivals were held not to confuse as a matter of law. Even the Indian Division Bench decisions the judgment relies on were interim findings about descriptive conjugations, Agrawal Packers in DRS Logistics and MakeMyTrip. Hindware takes that tentative material and, because HINDWARE is coined, converts it into a post-trial per se ban, reasoning that Section 29(8) is “stricter” than Section 29(4) because it omits the words “without due cause.” That reading makes ordinary keyword advertising a harsher, no-defence, no-confusion wrong than the dilution of a famous mark, which cannot be what Parliament intended.
Read as a whole, Section 29(8) cannot bear the weight the Court puts on it. Sub-sections (1) to (4) carry the tests of liability, confusion in (1) to (3) and the hedged dilution route in (4). Sub-sections (6) to (9) only define what counts as “using” a mark, and 29(8) names one such mode, advertising. The Court instead reads it as a standalone infringement needing neither confusion nor 29(4)’s gates of reputation and due cause, a fourth category the legislature never enacted. Read that way it swallows the rest of Section 29 and the class system, since any advertiser reaching a rival’s customers would be caught without a word about confusion or similarity of goods.
Keyword advertising is a relevancy-and-competition mechanism, and trade mark law reaches it at one place only – where the advertisement, by what it shows the consumer about the goods, deceives her about their source. Take away the confusion or the deception the Court says it does not need, and what remains is a search engine punished for running an auction over a word, and a coined word turned into property its owner alone may profit from in every corner of the internet – even if it is not a well-known mark. That is not the work trade mark law was built to do. A mark cannot exist in vacuo; it lives only in connection with its goods. A keyword, divorced from any deception about those goods, is just a word a person typed to begin looking, or a monopoly over linguistics or semiotics itself.
Ultimately, I must concede that the instinct of those hailing the decision is, of course, intuitive. Every human being is averse to any kind of free riding, which is in fact a mental expression of people who live in competitive capitalist societies. That, however, does not mean property rights ought to be extended to all kinds of uses per se. Trademark policy, is only and only concerned with the signaling of a source through a statutory right that ensures buyers not be misled about who makes its goods or services the buyers want to purchase or avail. The concern of manipulation, if at all (without conceding keyword advertising to be anything more than fair riding and a positive externality, especially because of the labelling of ads as sponsored), is addressed through a different legal tool – consumer protection law. India already deals with it through the CCPA Guidelines on Dark Patterns, 2023. A clearly labelled “Sponsored” result shown beside the brand’s own listing is however, beyond the regulatory purview of a dark pattern. While an unlabeled or a disguised one would offend both the dark patterns regime and maybe even the law concerning deception, in any case, a per se monopoly over the use of a word is the wrong instrument for either case. Maybe the solution to the HINDWARE problem is an even clearer labelling regime under the CCPA Guidelines, within closed boxes ideally, like they used to in the better days of the internet.
Views are Personal