Anyone who thought click fraud could be solved with a blanket settlement, take a step forward. Not so fast, Google.
Right now, there’s no way of knowing if Google’s $90 million offer will settle its part in the Lane’s Collectibles click-fraud class action suit. Search marketing advertisers will have 30 days to opt out of the settlement once they’ve been officially notified of the offer by Google; but at press time, those notices had apparently not gone out. They should be poised to mail soon: At the moment, a court hearing on that final okay is slated for late July.
But another disgruntled Google advertiser has filed a complaint intended to block the Arkansas settlement. Meanwhile, a second class-action click fraud suit is on hold in federal court in California pending the Arkansas decision. And lawyers in that case, who also represent the plaintiff in the new Arkansas complaint, have launched a Web site, www.clickfraud-legal-center.com, that tells Google search marketers they “may have already paid thousands of dollars in fraudulent charges” and urges them to opt out of the settlement.
Google responded by calling the complaint inappropriate and impugned the motivations of the lawyers in gaming the class-action process. They countered in a release yesterday that Google was the one abusing the legal system by “cutting itself a sweetheart deal in Arkansas that provides almost nothing to class members.”
In other words, pop some corn and grab a seat, because this show’s just getting started.
And remember, these fireworks involve past click-fraud claims against one search engine. Admittedly, it’s the one advertisers spend the most on, and the settlement is meant to cover claims stretching from the first pay-per-click ad on Google to the moment the judge brings his gavel down on a final approval. Still, the legal wrangling does nothing to answer the question: What about future fraud?
That was on the mind of an audience at last month’s ad:tech San Francisco session on click fraud. They were compelled to think in future terms, because the first panelist, Jessie Stricchiola, president of Alchemist Media, made it clear up front that she was not going to speak about the present, or the Google offer. She couldn’t, since she’s been retained as a consultant by the plaintiffs in the Lane’s case.
But the experts assembled were ready to discuss the systemic conditions that permit or foster fraud in the pay-per-click model, and possible ways to plug those gaps.
One large stumbling block is that the search engines don’t have access to data from enough advertisers to detect low-level but widespread acts of fraud. They’re left analyzing click patterns of traffic that comes to a Web site through their pay-per-click ads, but they don’t have a way to track actions on the Web site. According to Stricchiola, one of the earliest trackers of click fraud, it’s like the search engines have security cameras trained on the exterior of a bank.
“If every year a million dollars gets stolen from that bank, the search engines can say, ‘We know who stole it, because we know who went in,’” she said. “But the advertisers have video cameras inside the bank. So they see not only who came in, but what that person did when they got in.” Some Web operators are voluntarily submitting that web traffic data to the engines, but many don’t want to, figuring that business suppliers shouldn’t have that much visibility into their customers’ operations.
The upshot is that the search engines are the lone data source—and a faulty one– for determining which clicks should be billed and which credited as fraudulent. Lori Weiman, director of KeywordMax, pointed out that firms like hers have the advantage of being able to look at fraud across multiple search engines. That’s insight the engines themselves don’t have, and ammunition that advertisers can use against the engines in their click-fraud battles.
If judging the amount of click fraud is a problem, so is determining its cost; if 15% of clicks are bogus but those bad clicks occur on high-value keywords, then much more than 15% of total pay-per-click spending could be wasted. Tom Cuthbert, president and CEO of click-audit firm Click Forensics, says that indeed, fraud is more likely to occur on the costly search terms.
Earlier this year, Cuthbert’s firm launched the Click Fraud Network, a voluntary monitoring group that tracks the potential fraud on its members’ PPC ads using either software tags or their log files. The network reported in April that click fraud levels over the whole search industry ran at about 13.7% in March—substantially lower than the whisper numbers put out by other analysts, which have run as high as 40%. And its tiered breakouts found that March fraud was even lower than the average at Google and Yahoo!, accounting for 12.7% of all clicks.
At the time of the ad:tech meeting, the Click Fraud Network had about 500 Web marketers as members. But Cuthbert said later that another 500 joined the network immediately following the conference. That may help give the data more statistically validity; some search industry observers have complained that the network is too heavily weighted toward small Web advertisers who receive fewer than 100,000 click a month.
But detecting fraud is only part of the problem. Another issue is the way search engines handle disputes over fraudulent clicks and the often inscrutable means they use to arrive at fraud percentages. Several panelists pointed out that advertisers are made to pay for all their clicks, including potentially illegitimate ones, and then have to wait for ad credits (not refunds), which can be delivered on the engines’ timetable.
That’s different treatment than the engines give to the Web publishers in their networks, Stricchiola put up a slide comparing the terms of the search engines’ agreements with their advertisers against their contracts with the Web publishers in their networks. The language differences were illuminating: Google’s agreement with PPC advertisers calls for payment on “actual clicks”, but the deals it signs with Web publishers specify that it will pay them for “valid clicks”—a world of difference. Yahoo! too issues contracts to advertisers that make them pay up for “all clicks” on their sponsored listings; but Web sites in its network get paid out of revenue from “human users”.
It’s a complicated problem; and the remedies may be more complex still. One possibility is some kind of click scoring process to give advertisers a better view of how suspicious their clicks were. Both Stricchiola and Cuthbert are involved in initiatives to develop such scoring systems. Those fixes may be a tough sell to advertisers who opted for PPC advertising because it’s measurable and easily understood.
At one point, Stricchiola asked the audience how many would favor a click pricing system that linked advertiser payments to click quality, and charged lower-than-premium prices for clicks that were more or less dubious. No hands went up.
Then Weiman jumped in and restated the question succinctly in a sentence. More than half the audience in the double-wide conference hall raised their hands.