pay-for-click advertisers

Citysearch Click Fraud Class Action Certified — but Proving Meaningful Damages May Remain a Problem for Plaintiffs

The recent certification of a national class action in the Citysearch click fraud case represents a major victory – at least for the plaintiffs’ counsel. But whether adjudication of the case will produce significant recoveries for the plaintiffs is an open question.

The Citysearch click fraud class action (Menagerie Productions v. Citysearch, C.D. Cal., No. 2:08-cv-04263) was brought on behalf of some 10,000 advertisers on websites. Citysearch operates dozens of websites that provide information about restaurants, shops, hotels and other services in individual cities around the U.S. For example, at, Citysearch provides information geared towards the DFW metroplex.

To earn revenue through these sites, Citysearch sells advertising. Much of this advertising is “pay-for-click”, in which advertisers only pay when visitors clicked on their ads. The complaint claims that Citysearch entered into a standard form advertising agreement for these ads which claimed that: “We connect you to customers. You pay only for results.” In its FAQ page for the agreement, Citysearch also stated as follow:

Q: How do I know that clicks to my website are legitimate?
A: Citysearch proactively researches and develops processes, policies, and technologies to identify invalid click activity with respect to our customers’ advertising. Citysearch employs advanced security filters and blocks out clicks from spiders and robots.

The two named plaintiffs, Menagerie and Redwolf, claimed that despite paying up to $1,900 in advertising fees for pay-for-click ads over a period of several months, they received no new customers. There are many legitimate reasons that an ad campaign may not generate identifiable new revenue. However, the complaint alleged that the plaintiffs’ failure to generate new customers was because of click-fraud. In click fraud, an on-line media source that is party to a click-through ad contract inflates the number of ad clicks to fraudulently increase its ad revenues.

The plaintiffs allege that Citysearch failed to track fraudulent clicks originating with its employees and “partner sites” and failed to inform advertisers that it did not employ methods to track fraudulent clicks — but nevertheless charged customers for invalid clicks. The plaintiffs also allege that Citysearch falsely claims that customers will not be charged for invalid clicks, even though it knows or should know that these claims are false. The plaintiffs seek to recover the advertising fees they paid under breach of contract theories, as well as under California’s unfair competition law.

To qualify as a class action, the named plaintiffs must meet two sets of requirements: First, they must first meet four requirements in Federal Rule of Civil Procedure 23(a) that test whether the class is sufficient numerous and whether the claims brought by the named plaintiffs adequately represent the rest of the plaintiffs in the class. The main dispute here centered on whether the claims by the named plaintiffs were typical of all of Citysearch’s 10,000 pay-for-click advertisers.

Citysearch argued that the plaintiffs had presented no evidence that they were charged for fraudulent clicks, and hence, had no actual injury. In response, the plaintiffs presented a report from their expert which identified two clicks received by named plaintiff Redwolf within the same second on the same day. The plaintiffs also presented evidence that Citysearch had charged Redwolf for both clicks. The second named plaintiff, Menagerie Productions presented similar evidence.

Unfortunately for the public, the documents that contain this evidence were filed under seal – so it is not possible to analyze the general prevalence of such possibly invalid double clicks. The impression from the Court’s ruling is that the prevalence of even potentially invalid clicks was not extensive. Nevertheless, even though the evidence appears scant, the Court held that the two named plaintiffs had presented sufficient evidence they had suffered an actual injury – both had paid for invalid clicks that Citysearch had failed to detect.

The Court further held that even if some of the representations made by Citysearch to pay-for-click advertisers differed, the typicality requirement was met: “plaintiffs’ claims are based on an alleged common course of conduct by Citysearch to (1) charge it advertisers for invalid clicks, and (2) make material omissions regarding the existence and quality of its click filters.”

To qualify as a class action, the named plaintiffs must also show that the proposed class fits into one of several categories of cases specified in FRCP 23(b) that test whether certifying the case as a class action would make for efficient adjudication. Here, the named plaintiffs attempted to fit the case into FRCP 23(b)(2) which requires a showing that interests of all of the parties would best be served by settling their differences in single action. To do this, the named plaintiffs needed to show that the issues common to the class predominated over issues unique to class members, and that the proposed class was a superior method of adjudicating the matter.

The Court had little difficulty in finding that common issues of fact and law predominated for the breach of contract claims – because the claim arose from a standard form contract to which all of Citysearch’s pay-for-click advertisers had agreed. The Court also found that common issues predominated for the breach of the covenant of good faith and fair dealing claim – because the focus would be on whether the procedures Citysearch employed to identify click fraud were objectively reasonable.

The Court had a little more trouble with the UCL claims. California’s UCL statute permits claims against any “unlawful, unfair or fraudulent business act or practice.” Cal. Bus & Prof. Code ยง 17200. Here, the plaintiffs claimed that Citysearch’s advertising practices were both unfair and fraudulent. However, the Court found that common issues of law and fact would only predominate as to the fraudulent practices claims.

The Court held that to recover for fraudulent business practices under the UCL, a plaintiff does not need to establish the common-law elements of fraud, such as proof of deception, reliance and injury. Rather, UCL fraud is governed by a “reasonable consumer” test, which only requires that plaintiffs show that members of the public are likely to be deceived” [citing Williams v. Gerber Products Co., 523 F.3d 934, 938 (9th Cir. 2008)]. Because the UCL claim arose from Citysearch’s common course of conduct to all class members and the “reasonable consumer” standard would be used to adjudicate this claim, the predominance standard was met.

On the other hand, that Court found that to recover for unfair business practices under the UCL, “a plaintiff’s individual expectations about the business practice are relevant to determining the extent of its harm.” The need to examine the expectations of each plaintiff in the expected 10,000 member class made this claim unsuitable for resolution via a class action.

To show that a class action was a superior method of resolving the claims of all litigants, the plaintiffs also needed to show that they had a plausible method for calculating damages. The plaintiffs claimed that their expert would be able to identify fraudulent clicks (defined as clicks as having no probability of creating value) by merely examining the Citysearch click logs. Accordingly, the plaintiffs proposed a three-step method to calculate damages: (1) the plaintiffs’ expert would establish categories of clicks that were “objectively” invalid and that Citysearch should have filtered out, (2) he would examine Citysearch’s click logs to identify the invalid clicks for which each member of the plaintiff class was charged, and (3) the damages incurred by each plaintiff would be computed as a matter of simple math.

Citysearch claimed that this was bosh – because “it is impossible . . . to measure the subjective intent of the users who click on an online advertisement” merely from examining click logs.

The Court found that to satisfy FRCP 23 requirements, the plaintiffs’ damages methodology only had to be plausible – a low standard that she found they had met. Ultimately, in ruling issued on November 8, 2009, the Court certified a class action of “All persons or entities in the United States who entered into form contracts for pay-for-click advertising through, paid money for this advertising service, and experienced click fraud by reason of double clicks or Citysearch’s failure to apply automatic filters to traffic from its syndication partners up through March 23, 2007.”

The latest action in this case is that on November 25, 2009, Citysearch filed a notice of appeal of the class certification ruling. This appeal may indicate that Citysearch believes it is at risk of significant damages. However, the evidence that has filtered from the court records suggests that this risk may be moderate.

The plaintiffs’ primary damages will be recovery of the dollars they spent on clicks that the court finds were invalid. However, the two named plaintiffs seem to have only been able to show that they paid for one invalid double-click each. It seems to me that if there had evidence of pervasive invalid clicks, this evidence would have made it into the court’s ruling. Moreover, according to Citysearch’s papers, over 90% of its on-line advertisers renew or extend their pay-for-click contracts – suggesting that they are satisfied with the results and that the prevalence of invalid clicks may be low. If this is the case, the plaintiffs’ victory here could be pyrrhic.

What is more frustrating for online media businesses like Citysearch is that the precision in determining an advertiser’s ROI from its advertising dollars sought in this suit is only possible in Internet media. Broadcast media traditionally have only been able to use surveys (which is what “ratings” really are) that constitute only an educated guess at viewership of advertising.