Important Ninth Circuit Ruling on Keyword Advertising, Plus Recaps of the Past 4 Months of Keyword Ad Decisions

By Eric Goldman

Network Automation, Inc. v. Advanced System Concepts, Inc., 2011 WL 815806 (9th Cir. March 8, 2011)

[warning: this blog post is nearly 5,000 words]


We’ve had surprisingly few appellate decisions involving keyword advertising generally, and almost none involving trademark owners’ lawsuits against keyword advertisers (as opposed to suing keyword sellers like search engines). On that basis alone, this ruling is important. The case is also remarkable because the opinion, written by highly regarded Judge Wardlaw, gets so many things right. Perhaps that sounds like damning with faint praise, but the reality is that the Ninth Circuit’s Internet trademark law has become horribly tortured due to deeply flawed opinions like the 1999 Brookfield case. This opinion deftly cuts through the accumulated doctrinal cruft and lays a nice foundation for future Internet trademark jurisprudence.

The only sour note is that the opinion makes some unnecessary and empirically shaky “presumptions”–exactly the kind of unfortunate appellate court fact-finding that got the Ninth Circuit into trouble into the first place. Still, given how this opinion could have turned out, I still give this opinion very high marks.


The litigants both make software for job scheduling and management. This is reasonably expensive ($1k-$10k) software targeted at businesses. The advertiser (Network Automation) purchased the trademark owner’s trademark as keywords (at both Google AdWords and Bing) for comparative advertising. Thus, this case deals with a nice, clean example of comparative competitive keyword advertising.

The ad copy read:

The text of Network’s advertisements begin with phrases such as “Job Scheduler,” “Intuitive Job Scheduler,” or “Batch Job Scheduling,” and end with the company’s web site address, The middle line reads: “Windows Job Scheduling + Much More. Easy to Deploy, Scalable. D/L Trial.”

The ad copy doesn’t reference the trademark, presumably because the trademark owner blocked it via the search engines’ trademark policies.

The lower court proceedings appear to be fairly typical (other than the fact the advertiser initiated the litigation with a declaratory judgment; hence why its name is first). The trademark owner argued that the comparative competitive ads created initial interest confusion; the court used a bastardized form of the Sleekcraft multi-factor likelihood of consumer confusion test to slam the advertiser; and the court issued a preliminary injunction.

Use in Commerce

The court actually addresses this factor explicitly, a vast improvement over the garbled words in Playboy v. Netscape. Unsurprisingly, the court says that buying keyword ads constitutes a use in commerce. I say unsurprisingly only because no court outside the Second Circuit has ruled otherwise, and the Second Circuit said that selling keyword ads was a use in commerce in the Rescuecom case.

The court doesn’t explore the potential differences between selling keywords (a la Rescuecom) and buying keywords (this case). Even so, it continues to be clear that courts aren’t going to adopt the use in commerce defense to either buying or selling keyword advertising. Oh well.

A Side Note About Metatags

In recounting the history of the Brookfield case and its discussion of metatags, the court drops FN3: “Modern search engines such as Google no longer use metatags. Instead they rely on their own algorithms to find websites. See McCarthy at § 25:69.” Metatag plaintiffs, take note. I don’t think this footnote puts the nail in the coffin of judicial overreactions to metatags, but it’s a nice incremental step retreating from Brookfield.

Likelihood of Consumer Confusion

As a procedural matter, the court addressed the “Internet trinity/Internet troika” variation of the standard Sleekcraft test. In Brookfield, and then again in the 2000 GoTo case, the Ninth Circuit said that 3 of the 8 Sleekcraft factors were more important in Internet trademark cases and thus should get priority. This expedited version of Sleekcraft tended to work in plaintiffs’ favor. Here, the court tries to kill the Internet trinity variation, saying:

we did not intend Brookfield to be read so expansively as to forever enshrine these three factors — now often referred to as the “Internet trinity” or “Internet troika” — as the test for trademark infringement on the Internet. Brookfield was the first to present a claim of initial interest confusion on the Internet; we recognized at the time it would not be the last, and so emphasized flexibility over rigidity….Given the multifaceted nature of the Internet and the ever-expanding ways in which we all use the technology, however, it makes no sense to prioritize the same three factors for every type of potential online commercial activity. The “troika” is a particularly poor fit for the question presented here.

The court also does not expressly kill off initial interest confusion. Instead, it sidesteps that issue altogether. For example, it doesn’t define initial interest confusion or explain when it may or may not be present. Nevertheless, it subtly tries to merge initial interest confusion into the standard Sleekcraft test:

when we examine initial interest confusion, the owner of the mark must demonstrate likely confusion, not mere diversion.

Well, if you’re going to have to use the Sleekcraft test to evaluate likely confusion, exactly what work does the initial interest confusion doctrine do? It would have been great if the court had just gone ahead and said that initial interest confusion is worthless, but I’ll take this. I especially like that the court say diversion isn’t enough. Although that is not an express repudiation of the initial interest confusion standard in Brookfield, the Brookfield case was all about diversion, and here the court implicitly undercuts it.

The court then proceeds to work through a standard Sleekcraft test:

Mark Strength. This is the first place (of several) where the court makes unnecessary and unfounded factual assumptions. The court says “a user searching for a distinctive term is more likely to be looking for a particular product, and therefore could be more susceptible to confusion when sponsored links appear that advertise a similar product from a different source. The court continues “Because the mark is both Systems’ product name and a suggestive federally registered trademark, consumers searching for the term are presumably looking for its specific product, and not a category of goods.”

Uh, no. As I explained in lengthy detail here, we can’t accurately infer a searcher’s objectives when they use a trademark as a search term. In fact, I give examples of circumstances where searchers may use a trademark as the search query for a class of goods. The court’s presumption here, an empirical question that the court doesn’t defend, is off-base.

The court partially redeems itself when it says “if the ordinary consumers of this particular product are particularly sophisticated and knowledgeable, they might also be aware that Systems is the source of ActiveBatch software and not be confused at all.” True, but I don’t think a high degree of sophistication is required to make this type of source distinction. Even poorly educated consumers can distinguish Coke and Pepsi in the marketplace and will not be confused if a Pepsi ad appears in response to a keyword search for Coke. It’s not the consumer sophistication that matters; it’s whether or not the consumer already has a mental map of the various existing brands in the market niche. Ironically, because Google and Microsoft don’t allow a comparative competitive ad to explain the relationship between the brands, it may be harder for comparative advertisers to teach consumers in the ad copy about the relationship between competitive brands.

Proximity of Goods. The court adds a new twist: “the proximity of the goods would become less important if advertisements are clearly labeled or consumers exercise a high degree of care.”

Mark Similarity. The court says this factor also depends on ad labeling and consumer sophistication.

Evidence of Actual Confusion. No evidence was introduced for the preliminary injunction, so the court weighs this as a non-factor. This is actually good news, because many courts have counted this factor against defendants by hypothesizing the existence of initial interest confusion as a substitute for any evidence of actual confusion.

Marketing Channels. Given that most companies have an Internet presence now, the court said the district court erred by counting this factor against the defendant.

Purchaser Care. The district court said that Internet consumers categorically exercise low care. Given the rich information on the Internet and the ability of consumers to do more research than ever, this has always been a dumb standard (see, e.g., Ann Bartow’s Likelihood of Confusion article).

This court rightly shreds that assumption. The court says we should not rely on “a conclusion reached by our court more than a decade ago in Brookfield and that Internet users on the whole exercise a low degree of care.”

Intent. The court says the lower court improperly assumed deceptive intent by the advertiser without considering the advertiser’s desire for comparative advertising.

Product Line Expansion. Unimportant when the litigants are already in direct competition, such as in this case.

Other Factors. In a footnote, the court rejects the bonus 7 factor test from the Hearts on Fire case. However, going back to language from Playboy v. Netscape, the court says the “appearance of the advertisements and their surrounding context on the user’s screen” are important, and the search engines’ presentation of ads–separated and labeled–should also be considered.

Instead of the Internet trinity or the Hearts on Fire supplemental test, the court possibly offers up a Internet quadrangle of Sleekcraft factors:

the most relevant factors to the analysis of the likelihood of confusion are: (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.

I’m not sure a new expedited form of Sleekcraft avoids the problems we saw with the Internet trinity. But these factors are a step forward.


After dissolving the preliminary injunction, the court remands the case to the district court. It’s not clear to me what will happen there. On the one hand, the district court judge showed that it was moved by the plaintiff’s story, so it still may be sympathetic to the trademark owner. On the other hand, the Ninth Circuit opinion has a lot of language favoring the advertiser, and the district court judge might interpret that language as an imperative to rule for the advertiser lest it get reversed again. I think this is a close call.


I am often asked by other Internet Law professors for a single keyword advertising case they should consider teaching. Until now, I haven’t had a good answer. I’ve taught several keyword ad cases over the years. The last two years I’ve taught the Hearts on Fire case, which has been pretty good. Other folks have taught the Second Circuit’s Rescuecom case, a theoretically interesting case but a lousy teaching case. In my opinion, this ruling is clearly the best keyword advertising teaching case now available. Unless something better comes along, I’ll be substituting this case for the Hearts on Fire case in my Internet Law reader. Assuming many of my colleagues make the same choice, I expect this opinion will be an instant classic.

For more on the opinion, see Paul Levy’s take.

UPDATE: Rebecca’s cogent critique of the case.


I have accrued a bunch of other keyword advertising cases over the past 4 months that I simply haven’t had time to blog. In the remainder of this post, I’ll catch up with recaps of those cases as well. However, for the most part, this nicely written Ninth Circuit opinion trumps the remaining precedential import of these other cases.

Defense Wins

Montana Camo, Inc. v. Cabela’s Inc., 2011 WL 744771 (D. Mont. Feb. 23, 2011). Cabela’s buys fabric from Montana Camo and manufactures clothes using the fabric. In a hangtag, Cabela’s indicates that the fabric is from Montana Camo. Cabela’s buys “Montana Camo” as keywords.

The court rejects Montana Camo’s 1125(a)(1)(B) false statement of fact claim because “the purchasing of a sponsored link is not a statement of fact. Further, considering that Montana Camo products were sold on Cabela’s website, it was not a false statement of fact.” The court rejects the 1125(a)(1)(A) unfair competition claim because Montana Camo didn’t marshal enough evidence of confusion.

Thus, this case indicates that a manufacturer may be able to bid on the trademarks of its component suppliers without running afoul of Lanham Act false advertising rules., Inc., v. One Techs., LP, CV-09-3783-VBF (MANx) (C.D. Cal. jury verdict Jan. 12, 2011).

The TM owner asserted its purported TM rights in “,” a problematic domain name designed to take advantage of misdirected consumers who were really seeking, the government-mandated website that lets consumers get free access to their credit reports. Consumers at get coopted into credit monitoring services that they may not want and probably don’t need.

Given the marginal legitimacy of, you’d think it would lay low legally. Instead, like other owners of crappy trademarks (see, e.g., 1-800 Contacts, discussed below), they tend to be more bare-knuckled litigious than typical trademark owners. In this case, they sued businesses that registered typosquatting domain name variations of I trust you see the irony– plays on consumer misrecollections of, yet they don’t like anyone doing the same to their purported trademark. Nice. The jury awarded a big cybersquatting judgment under the ACPA to the tune of $1.9M; however, the jury found that the defendants’ keyword bidding did not create a likelihood of consumer confusion.

We don’t have many jury verdicts about keyword advertising. The two I can think of are College Network v. Moore and Fair Isaac v. Experian. This would make the third time a jury has found in favor of the keyword advertiser over the trademark owner when the jury finally gets the question asked to them. This reinforces that juries may be more tolerant of keyword advertising than judges (and are certainly more tolerant than trademark owners!). This particular jury ruling is especially noteworthy because the jury thought the defendants were bad guys (hence the very large ACPA judgment), yet the jury still approved the keyword advertising.

1-800 Contacts, Inc. v., Inc., 2010 U.S. Dist. LEXIS 132389 (D. Utah Dec. 14, 2010).

This case, another suit over competitive keyword bidding, got stuck in my blogging queue. It’s a tremendously important ruling and a terribly embarrassing one for 1-800 Contacts, so I planned to devote a lengthy blog post exploring its interstices. Unfortunately, the time never materialized in my schedule. Why was this case so high on my list? Three highlights:

1) It was a resounding loss for 1-800 Contacts, a company that has earned my ire over the years for their duplicity and pugnaciousness about trademarks and keywords. (For my blog coverage of them, see here). Some lowlights in 1-800 Contacts’ track record:

* they are hyper-aggressive about protecting a marginal trademark. In my mind, it’s not a trademark at all, it’s a phone number. Frankly, I think we should categorically declare phone numbers as ineligible for trademark protection, just like we no longer recognize trademarks in [noun].[tld].

* they buy third party competitors’ trademarks as keyword triggers, yet they sue competitors for buying their name (I can’t really call it a trademark) as keyword triggers. Indeed, the court recounts that 1-800 Contacts bought “1 800 lens; 1 800 lense; 1 800 lenses; 1 800 the lens; 1 800 Lens; 1-800 lens; 1800lenses; 1800lens; 1800lenses; 1-800-lenses; 800 lens; 800 lenses; 800lens. These keywords generated 91,768 impressions, 8,477 clicks, and about $219,314 in profits for Plaintiff.” HYPOCRITE ALERT. (BTW, their $26 of profits per click is mind-bogglingly impressive).

* they flip-flopped on the Utah legislature’s efforts to ban keyword advertising, helping to kibosh the first law and then trying to sneak in a second law that favored their interests–aided by the fact that their in-house lobbyist is also a legislator and voted in favor of the bill her employer advocated. Yet, on its site, 1-800 Contacts claims “1-800 CONTACTS engages on public policy issues related to ocular health and the right of contact lens wearers to choose where they fill their prescriptions. We have not and will not get involved in public policy outside of the scope of this interest.” Sorry, I’m going to have to call BS on that.

2) The case rejects 1-800 Contacts’ attempt to hold the defendant liable for keyword ad buys made by’s affiliates. Trademark owners have been angling to establish a legal doctrine that online retailers are automatically liable for keyword ad buys by affiliates, but this case gives some additional reason to believe that trademark owners have been overreaching.

3) The case gets into details about how much money made and, in theory, 1-800 Contacts lost due to’s keyword ad buys. The court says bought the following keywords:

1 800 contact lenses; 1800 contact lenses; 800 contact lenses;;;;;; and These nine keywords generated about 1,626 impressions, 25 clicks, and $20.51 in profits

Wait, what? The parties are fighting over’s $20 of profits??? Hey, 1-800 Contacts, if you’ll stop bringing pitiful lawsuits, I’ll send you an Andrew Jackson out of my own pocket. Clearly, the real thrust of this lawsuit were the affiliates’ keyword ad buys, but even those weren’t voluminous: one affiliate bought 65,000 allegedly infringing impressions generating 352 clicks, and another affiliate allegedly bought 240,000 impressions generating 1,445 clicks.

Are ~1,800 allegedly misdirected clicks worth making a federal case out of? Even at 1-800 Contacts’ impressive (and probably overstated) $26 of profit per click, we’re talking about less than $40k of value that 1-800 Contacts purportedly lost. Yet, 1-800 Contacts was prepared to spend $1.1 MILLION on this lawsuit (and actually spent at least $650k). Great business decision there, guys. WHAT A WASTE. As I wrote in that earlier blog post, “I’m super-skeptical that the value of the consumers “diverted” (whatever that means) by’s competitive keyword advertising is more than $1.1M.” The financial details in the case reinforce that I was 100% right about that.

Substantively, the court says keyword ad buys are a use in commerce. The court correctly explores the effect of broad matching on searches like “1-800 Contacts”–due to broad matching, competitive ads keyed to “contacts” may show up. The court grants summary judgment to for its ads.

It suggests that some of affiliates’ ads may have infringed because they mention 1-800 Contacts in the ad copy. (The court later clarifies that it wasn’t the ad buy that infringed; it was the ad copy). However, those actions aren’t imputed to because got its affiliates through Commission Junction, and therefore didn’t know their identity and had little direct contact with them. The court also rejects 1-800 Contacts’ takedown notice to because 1-800 Contacts didn’t give enough information to find the affiliate who ran the ad.

Finally, 1-800 Contacts tried to argue that contractually agreed not to buy its trademarks as keywords during their various correspondences in response to 1-800 Contacts’ legal threats. This is similar to Barnes v. Yahoo and Scott P. v. Craigslist in that the plaintiff is arguing that the defendant promised to remediate and thus its failure to do so is a contract breach. The court rejects this bypass.

You can see why I love this opinion. It’s a long but rewarding read. Check it out.

(For people interested in Ben Edelman’s work, you might be interested in the court’s discussion about Ben’s expert report on pages 20-23. A sample: “parts of Edelman’s declaration are improper in that he presents evidence not within his personal knowledge by reciting what another said in deposition and stating that testimony as fact, he opines on facts for which no expert testimony is needed, and he draws legal conclusions that are outside his role as an expert”).

Plaintiff Wins

FTC v. Cantkier, 2011 WL 742647 (D.D.C. March 3, 2011). The court’s recap of the complaint:

The FTC has alleged that Lady and certain other defendants were running deceptive online advertisements featuring the names, phone numbers, and website addresses of federal homeowner relief and financial stability programs. The advertisements allegedly appeared on popular web search engines, such as Google and MSN, and were targeted to users using as search terms keywords related to the federal assistance programs. The Second Amended Complaint alleges that the advertisements represented that they were sponsored by federal homeowner relief and financial stability programs by featuring text and titles associated with those programs, including “” and “” When web users clicked these ads, they were not directed to the websites for the federal programs, but rather to private Internet websites (“lead collection websites”) that collected marketing leads for mortgage loan modification or foreclosure relief services. These lead collection websites had no actual connection with government programs; they solicited consumers to enter personal identifying and confidential financial information, and then the operators of the websites sold the consumers’ confidential information as marketing leads to persons who sell mortgage loan modification or foreclosure relief services….

Plaintiff alleges that Lady purchased advertisements on (“Google”). On Google, Lady bid on keywords “financial,” “,” “,” “hope now alliance,” “hope for homeowners,” “,” and “” On Google, his advertisements displayed titles “,” “Financial,” “Fha Gov,” “,” “,” and “” The FTC alleges that consumers who clicked on Lady’s advertisements were not directed to the government websites, but rather to his own websites that collected marketing leads for mortgage loan modification or foreclosure relief services. Lady’s websites prompted consumers to enter personal identifying and confidential financial information, which Lady then allegedly sold as marketing leads to persons who sell mortgage loan modification or foreclosure relief services. (cites omitted)

On this basis, the FTC alleged deceptive acts under the FTC Act. The court rejects the defendant’s motion to dismiss.

There are a number of interesting points in the discussion. Some highlights:

* the defendant argued that consumers understood they were clicking on ads. The court acknowledges this but says the FTC’s complaint is that the ad copy was deceptive.

* the defendant argued that his advertised websites didn’t look like official government websites. The court responds: “Internet users may not know what the real federal program website looks like until they successfully navigate to it. If they are diverted by advertisements bearing the name and web address of the federal program before ever reaching the program’s actual website, reasonable consumers could assume they have reached their intended destination, when, in fact, they have reached a commercial service.”

This is a little like the old Promatek v. Equitrac discussion of diversion, to which the “back button” is a solid retort. However, it feels qualitatively different to me that we’re dealing with allegedly false ad copy trying to mimic official government services. Contrast the rulings in the Consumerinfo case above, where the jury found no consumer confusion from keyword advertising for a website replicating a government-mandated website, and the recent Canadian decision in Private Career Training Institutions Agency v. Vancouver Career College (Burnaby) Inc., where the defendants’ websites mimicked community colleges. In the latter case, the court said that prospective students would figure out any confusion before enrolling in college. That case clearly expected consumers to be more sophisticated than the FTC did in this case. Also along this lines (but not a keyword ad case) is the lawsuit over

Rebecca’s post on the case.

Binder v. Disability Group, 2011 WL 284469 (C.D. Cal. Jan. 25, 2011). This is another lawyer-as-plaintiff suit, so you know we’re in trouble. The advertiser, a direct competitor, purchased the law firm’s name as keywords. The court breezily says that keyword purchases are a use in commerce. The district court found a likelihood of confusion by focusing on the Internet trinity of factors; the opinion also made a number of other statements inconsistent with the Network Automation case. Unlike Network Automation, in this case there was some evidence presented of actual confusion, including after users clicked on the ad (so the confusion was not solely attributable to the keyword ad). That might suggest the ruling would withstand further scrutiny, especially given that we’re talking about law firms competing with each other and clients could get into trouble by connecting with the wrong law firm.

In underdeveloped parts of the opinion, the court also finds Lanham Act false advertising and California unfair competition violations, saying “Plaintiffs have proven by a preponderance of the evidence that Defendants used Plaintiffs’ mark in their advertising campaign through Google to market their business in a manner that was likely to confuse potential clients and that deceived potential clients into thinking they were being led to Plaintiffs’ website” and “Plaintiffs have proven by a preponderance of the evidence that Defendants used Plaintiffs’ marks in their online campaign and in doing so attempted to pass off their website as Plaintiffs’, and/or infringed on Plaintiffs’ trademarks.” This deserved way more words than the court gave it. The court also has some garbled discussion that the TM owner did not need to mitigate harm by complaining to Google.

Using some questionable methodologies about conversion rates (18%!), revenue per case and costs of serviced cases (95% revenue margin!), the court calculated damages and then doubled them for willfulness to nearly $300k. Regarding willfulness, the court says:

Plaintiffs have established willfulness in this case. As described above, Defendants chose Plaintiffs’ marks based on the market. In doing so, Defendants intentionally misled potential clients and directed business away from Plaintiffs and to their own websites. Defendants had the deliberate intent to direct clients to their sites with the false impression that they were Binder and Binder. Defendants also intentionally chose Plaintiffs’ marks with knowledge that they were registered trademarks and in an attempt to profit from them.

Equating willfulness with exceptional, the court also awards attorneys’ fees and costs. The court also extended liability to the defendant’s principal personally. However, the court refused a request for corrective advertising and punitive damages (which were available for the CA unfair competition claim).

On the surface, this looks like a problematic case. Partially in response to this case, a Search Engine Land contributor asked if “Is It Time To Rethink Bidding On Trademarks?”. However, there are three mitigating factors that undercut its import:

1) the suggestion that the advertisers engaged in misleading activity after the keyword ad.

2) the court clearly disbelieved the defendant’s principal, never a good indicator of a successful defense.

3) I wonder how much of this case survives the Network Automation ruling. It appears potentially vulnerable to an appeal or rehearing request.

1-800 Contacts, Inc. v. Memorial Eye, PA, 2010 WL 5149269 (D. Utah Dec. 13, 2010).

In one of 1-800 Contacts’ multitudinous trademark lawsuits against competitors over competitive keyword ad bidding, the advertiser asserted an unclean hands defense (on the basis that 1-800 Contacts buys competitors’ trademarks for competitive keyword advertising itself) and a trademark misuse counterclaim. The court rejects both. In general, this ruling is trumped in importance by the ruling. However, it is interesting that the court thought 1-800 Contacts engaging in identical behavior as the behavior it was suing over wasn’t good enough for an unclean hands defense. In the court of popular opinion, 1-800 Contacts is unacceptably duplicitous.