The great click fraud debate hasn’t gone away, but it has entered a new phase, thanks largely to new efforts by Google and Yahoo! to add some visibility into advertisers’ results and to the overall performance of their pay-per-click ad networks.
Visibility has been one of the contentious issues between search marketers and the engines that deliver their ads. While they’ve been quick to tout their proactive stance to detecting bogus clicks and rooting out suspect domains in their Web content networks, the big search powers haven’t been equally forthcoming with details about the extent of the problem of “invalid clicks”—their preferred term—on their networks or to give advertisers tools to fight them, or even, some marketers say, efficient ways to file reimbursement claims.
But in recent weeks both Google and Yahoo! have taken significant steps to lift some of the secrecy surrounding fraudulent clicks on their networks. These include going on record about the percentage of total ad clicks that they flag as not coming from legitimate visitors, integrating new data into the reporting they provide to marketers and, in Yahoo’s case, creating an executive-level post with responsibility for safeguarding click quality across a number of Yahoo! ad products including sponsored listings.
The moves don’t add up to anything near a total solution, some marketers and SEM professionals say; and indeed the overall fraud statistics the search engines have publicized have raised as many questions as they’ve answered. But the various initiatives do seem calculated to bridge the trust gap with search advertisers, perhaps in an effort to forestall class-action click-fraud suits like the ones both Google and Yahoo! ended with settlements last year.
Google broke the statistical ice at the beginning of March with an announcement that 10% of the total clicks on all Google’s networks—search and content pages—are flagged by its detectors as invalid and not charged to the advertiser. That information came in a post on the Google AdWords blog by Shuman Ghosemajumder, business product manager for trust and safety, who actually pointed out that the Google “invalid click” rate has fluctuated below 10% since the company launched the AdWords pay-per-click program in 2002. He also notes that the overall rate will be higher or lower for specific industry verticals.
One step further along, of those clicks that Google’s processes deem valid, about 0.02% are later judged to be fraudulent and credited to advertiser accounts. “Put another way, for every ten thousand clicks on Google AdWords ads, fewer than two are reactively detected cases of possible click fraud,” Ghosemajumder writes.
To reiterate, Google says it finds about one in ten incoming clicks on its network to be suspect and filters them without applying any charge to advertisers. Another two clicks in ten thousand are caught by looking back over log files, usually as a result of advertiser inquiries.
A few weeks later Yahoo! released the nugget that its own proactive “network discard rate” for invalid clicks is between 12% and 15% of all clicks on its networks. Once again, that’s an aggregated figure and will run higher or lower for specific verticals. No word on what proportion of clicks makes it through the Yahoo! detectors only to be found fraudulent later.
While those percentages are aggregated and thus not very useful to a specific marketer wondering about the fraud content in his ad clicks, search industry professionals seemed to accept the generalized stats as the first hints from the engines on the size of the bogus click problem. The proportions are largely in line with many widely accepted third-party fraud estimates, and much lower than some studies that set the click fraud rate as high as 30%.
In a dispute where the search engines have always been reluctant to divulge their internal processes, many search marketers seemed to applaud the effort to apply boundary markers to the problem without necessarily buying into their specific placement.
Enquiro CEO and Search Engine Marketing Professional Organization chairman Gord Hotchkiss wrote in his blog that the only reasons Google would have for producing incorrect click fraud network stats would be financial, either for a share price boost from investors or to entice more advertisers into the pay-per-click fold—and he doesn’t believe the company would deflate fraud figures for either group. Analysts have already factored in a certain amount of click fraud and moved on, he wrote. Advertisers too have become less preoccupied with the question of fraud; in the recent SEMPO State of the Industry survey, anxiety about click fraud declined among search marketers.
“Google had to release these numbers because everyone was asking for them,” Hotchkiss wrote, in a post before Yahoo! released its own click fraud proportions. “If we can accept those numbers, then perhaps we can get on with looking at our overall campaign performance.” Focusing on conversion rates, for example, is a more productive use of search marketers’ time than fretting that click fraud is more widespread than the engines are saying, he said.
But other analysts expressed skepticism about Google’s contention that 0.02% is a realistic measure of the amount of fraud its systems don’t detect in advance. Jokob Nielsen points out in a comment on Search Engine Land that a true measure of unfiltered but bogus clicks would be “all the manually-discovered fraudulent clicks divided by the number of clicks in those campaigns that complained and were investigated.”
“Presumably, most campaigns don’t complain and are thus not investigated, meaning that they hide some additional clicks that would have been found,” he wrote.
“Let’s say for argument’s sake that one out of 100 advertisers ever submit to Google for a claim of fraud reimbursement,” says Tom Cuthbert, CEO of click auditing firm Click Forensics. “If that [0.02% percentage] holds true, then what would that amount to for the masses of advertisers in their customer base who don’t submit a claim?”
So the size of the undiscovered fraud problem remains elusive, even if the engines have given some indication of the total proportion of bogus clicks.
As for tools, Google and Yahoo! have promised very different measures to help alleviate fraud. Google will introduce the ability to let advertisers block specific IP addresses, whether because their logs tell them these addresses are producing lots of fraudulent clicks or simply because clicks from these points rarely convert into sales.
Google says it will also beef up fraudulent click reporting on campaigns. In mid-2006, it began reporting the absolute numbers of detected bogus clicks per campaign along with the percentage those made up of all clicks on that campaign. Now Google says it will add another stat to the report: the dollar figure those invalid, uncharged clicks represent. The company will also launch a tutorial Web site to educate advertisers about preventing and detecting click fraud, and later this year will roll out a standardized automated system for reporting suspected click fraud for investigation.
Yahoo! has taken a different tack, opting for people over systems. The search engine has appointed in-house attorney Reggie Davis to the newly-created post of vice president of marketplace quality.
Davis will hire and lead a team that will respond to advertiser concerns over bogus clicks on pay-per-click ads on Yahoo!’s search results pages and within its network of Web content sites. Other areas relating to online ad quality, such as traffic quality and network placement, will also be part of Davis’ responsibilities.
In his new capacity, Davis will help guide development of tools designed to make Yahoo online ads more attractive to advertisers. These will reportedly include the introduction later this year of a tiered pricing system based on click quality and a domain blocking function that will let marketers bar clicks from suspect Web sites.
Davis says the company is also working on plans to institute an advisory council of advertisers and publishers who will offer feedback on issues and enhancements in Yahoo’s performance ad programs. Last year’s settlement of a class-action click-fraud lawsuit against Yahoo! by Checkmate Strategic Group called for the formation of such an outside panel. In fact, Davis managed the Yahoo! legal team in that litigation.
“I give credit to Yahoo! for creating this post and hiring this group,” says Rob Murray, president of search marketing firm iProspect. “I don’t see anyone in a similar position at Google or MSN.” He also gives props to both Google and Yahoo! for taking steps to add transparency to the fraud problem. But he’s also careful to distinguish these moves toward greater visibility from anything that would actively enhance fraud prevention.
“None of this is aimed at prevention—it’s aimed at identification,” Murray says. “It’s symbolic in nature, and it will add transparency for after-the-fact analysis. But I don’t see anyone doing a lot to prevent fraud in the first place.”