Google AdWords' individual characteristics defeat class action

In re Google Adwords Litig., 2012 WL 28068 (N.D. Cal.)

Plaintiffs alleged false advertising of Google AdWords and sought class certification under California law.

The AdWords network includes parked domains (part of AdSense for domains, AFD) and error pages (AdSense for errors, AFE).  The ads display as a result of Google’s attempts to match ads with the terms users entered to arrive at a particular page.  Plaintiffs alleged that Google failed to disclose that, whether they chose to advertise in Google’s Search or Content networks, Google would place their ads on parked domains or error pages, and that Google was aware of the negative reputation of such domains/pages and deliberately concealed its involvement with those sites.  Indeed, Google allegedly published misleading AdSense policies which prohibited publishers from putting ads on pages lacking content, and obscured the URLs of AFD and AFE sites on Google's Placement Performance Reports so that advertisers could not see the actual sites upon which Google had placed their ads.

Google argued that plaintiffs mistakenly presumed that every ad on any parked domain or error page was worthless and harmful, whereas advertisers reaped demonstrable benefits from such ads (even if they had a negative perception of such ads).  Google also contended that its disclosures were fine, and that advertisers could exclude their ads from entire categories of web pages, including parked domains and error pages.

Plaintiffs moved to strike declarations by three Google employees because they were offering inadmissible expert testimony.  The court largely agreed with Google that these were lay witnesses testifying to the personal, particularized knowledge they had by virtue of their positions in the business, and thus their testimony was not problematic under Rule 701.  They provided key information about how AdWords pricing worked and about how many advertisers opted out of placing their ads on parked domains and error pages.  (Interestingly, the court used the term “behaves” to describe AdWords, suggesting a level of complexity and even agency beyond that of ordinary pricing systems; STS folks take note.)  To the extent that they opined on the merits, such as the viability of classwide restitution, the court disregarded their testimony.

Plaintiffs moved to certify a class of customers who were charged for clicks on ads placed on parked domains or error pages.

First, the court addressed standing.  Named plaintiffs have to show actual reliance under the UCL and FAL.  This can be done by showing that the defendant's misrepresentation or nondisclosure was an immediate cause of the plaintiff's injury-producing conduct.  For omissions, a plaintiff may show that, in the absence of the omission, the plaintiff in all reasonable probability wouldn’t have engaged in the injury-producing conduct.  An inference of reliance arises when a misrepresentation was material.

The named plaintiff alleged that it viewed the parked domain and error pages as low quality sites upon which it didn’t want to advertise, and that Google knew that plaintiff was likely to regard Google’s placement of ads on such pages as important.  Plaintiff further alleged that Google’s omissions led it to buy ads it otherwise wouldn’t have bought and to overpay for clicks on ads on such pages.  These allegations were sufficient to establish reliance.  Plaintiff also satisfied Article III with the same allegations, which indicated that it had lost money or property by buying ads it wouldn’t have wanted had it known the full truth.

Google argued that, even in UCL cases, absent class members needed Article III standing, and thus a fact-intensive individualized inquiry into injury was required for each member.  Since California provides relief under the UCL and FAL without individualized proof of deception, reliance, or injury, and since the governing objective test requires a plaintiff only “show that members of the public are likely to be deceived” by a defendant's representations about its product, no such fact-intensive inquiry was required.  If the class representative has Article III standing, no further analysis of unnamed class members need be undertaken.  (And anyway Article III was satisfied: consumers who bought AdWords were relieved of money in the transactions.)

Plaintiffs sought 23(b)(3) certification, which requires predominance of common questions of law or fact.

Google argued that the class wasn’t ascertainable, because there was no systematic way to exclude advertisers who benefited from ads on parked domains/error pages; most advertisers didn’t use Google’s tool to calculate conversion rates so there’s no classwide method to determine the benefits they received from ads on such pages, and anyway conversion rates alone might not be a satisfactory measure of overall benefits from such ads.  Given the class definition, the court saw this as an issue regarding entitlement to restitution, not ascertainability.

Numerosity wasn’t a problem.  On commonality, plaintiffs argued that whether Google's alleged omissions were misleading to a reasonable AdWords customer was capable of classwide resolution, while Google contended that whether advertisers were entitled to restitution wasn’t.  The court found both questions central to the litigation.  Though the second might generate individual answers, there was still a valid common question.  Plaintiffs’ claims were also sufficiently typical, and there was no evidence of conflicts of interest with proposed class members.

However, plaintiffs flunked predominance because questions of restitution would require individual inquiries.  (The court specifically rejected Google’s argument that individual questions of reliance and materiality also existed, because of the UCL/FAL reasonable consumer test.)  Though damages variations alone can’t defeat certification, the facts here were special.  AdWords uses an auction that means that each advertiser paid a separate amount per ad and per click, dependent also on the other bidders. “These intricacies make it more difficult to calculate what AdWords customers would have paid ‘but for’ the alleged misstatements or omissions.”  Thus, a reduction in demand for advertising on AdWords wouldn’t necessarily lead to a lower price for all advertisers. 

Moreover, advertisers have “widely varying goals,” making it difficult to calculate the value of what they received.  Conversion to purchase is one way of measuring performance, but in 2009 only a small percentage of active AdWords accounts opted into conversion tracking, and even for them that might not be the only measure of value.

Restitution need not be determined with exact precision, but it must be based on a specific measurable amount supported by substantial evidence.  With these parties and facts, the court wasn’t convinced that restitution could be reliably measured using common methods.  A full refund, for example, wouldn’t take into account benefits received, given that some advertisers actively sought to have their ads placed on parked domains and error pages.  A uniform discount for all such ads would also ignore that individual advertisers’ ads might outperform the same ads placed on other types of pages, as was the case for one named plaintiff.  Individual benefits from ads would need to be accounted for in any restitution calculation.  “Where, as here, proof of restitution due each class member cannot be proved with relative ease, the court finds good reason to deny class certification.”