Prevent Click Fraud from Eating You Alive.

Posted in Paid Marketing

This afternoon at SES San Jose, on the Issues Track, search marketers and audit firms gathered to discuss the click fraud problem. In this session issues surrounding fraud and tactics designed to safeguard media buys were explored. For many of us this session was the first time we’ve seen data to support our long-held suspicion: In terms of conversion, search networks outperform content networks at a rate of approximately 5/3 (Click Forensics).

How Much Does Fraud Hurt?
Traffic from Made for ad sites (MFAS) and parked domains convert at 1/20th the rate other PPC. 71.8% of sites that participate in the Yahoo publisher network are MFAS or parked domains as are approximately 60% of Google ad sense sites (Click Forensics). The victims of fraud are clients and advertisers who suffer lower conversion rates and higher CPAs, SEMs/Agencies whose usable budgets are lowered as a result, and search engines who make less money for clicks sent after writing off bad traffic. Average click-fraud rate across search advertising industries is 15.8% .

No Search Engines Allowed
Though click fraud sessions have been held at SES conferences for years this was the first panel which did not include representatives from search engines. Google, Yahoo, and MSN were all represented in the audience. The engines’ session on fraud prevention takes place later this afternoon. The panel segmentation this time around was designed ostensibly to allow for unfettered discourse among buyers (advertisers) without the sellers (engines) mucking up the dialog. The moderator was Jeffrey K. Rohrs, VP, Agency & Search Marketing, ExactTarget.

John Linden, Chief Technology Officer, Think Partnership explained that collection of data surrounding user behavior is key. Most fraud prevention companies separate out known problem-users but these days intelligent crime-bots and scripts use real IP ranges, variable traffic patterns, and a mixture of known user agents which can help bogus clicks blend in with legitimate traffic.

Did the User Actually Intend to Click?
The bad guys can scrape actual content into hidden iFrames and overlay hotspots to siphon unintended clicks from users that didn’t mean to click on any ad. Detection is difficult because the behavior must be prevented prior to the click. John also stressed that legitimate clicks can under perform for advertisers and that’s not fraud. Conversion analytics to establish performance baselines are very important for click quality and fraud detection to know what’s what.

The Bottom line: Click quality is every publisher’s responsibility, post click analysis has significant limitations, and client-side analytics uncover important malicious user behaviors.

Jessie Stricchiola, Founder, Alchemist Media Inc. has been an influential professional in the evolution of click fraud and solutions. She’s worked for clients, search engines, and participated in major litigation regarding Google and fraud. She noted that some companies with acceptable ROI choose not to pursue fraud. However if margins are important and significant expenditures are at risk for paid marketing it’s crucial to investigate suspected criminal behavior.

Tom Cuthbert, President & CEO, Click Forensics discussed various initiatives Google, Yahoo, and Ask have undertaken to educate, empower, and protect advertisers. “There’s been some growth and the signs are encouraging but the systems are still far from perfect.” Yahoo has implemented a traffic quality center to provide resources. Google is touting their Ad Traffic Quality Center and changed the paid search algorithm. Theoretically this shift will offer advertisers more control over achieving top placement while increasing the quality of ad results for users. Many of us believe the most important outcome of the change is increasing CPCs on Google…sound familiar?

Jon Myers, Director of Search Insight, Latitude Group says the take-away is that click fraud is an Internet crime to steal money. “There is a lot of hype, confusion, and concern which leads to a lot of questions.” He wants to know what the search engines are doing to protect us, how prevalent is click fraud on our paid search campaigns, and how can we protect ourselves.

Search Engine Efforts
Google operates a 3 stage process starting with the Invalid Clicks Process. This can result in a manual review. Microsoft adCenter offers a very nice new tool. Yahoo! Click Protection has created a portal to monitor click quality called the Traffic Quality Center. (Features include IP and continent conclusion.)

All of the search engines are working on similar solutions and there are positive outcomes in the wind. However click fraud, while not a deal-killer, is still a very serious problem calling for vigilant advertisers, plugged in solutions, and help from the engines.

  • gabriel bear

    The rate of “bad clicks” –16%–seems remarkable constant with what it was 3 years ago.
    So, what has Google done, besudes increase cpc and get rid of hundred, if not thousands of google publisher accounts?
    And who got the replacement accounts? Websites more “friendly” to Google?
    The real issue, as always, is building community through providing value.
    Real SEO strategies are an aspect of getting value per click. Because the end customer is always in the value business.

  • CPCcurmudgeon

    The Internet Architecture Board (www.iab.org) is looking at click fraud, as part of a general investigation of the harm done from “unwanted traffic”. I have shared my concerns and experiences with them. Please do the same, if you want a ground-level understanding of how click fraud works and what measures for curtailing it are feasible.

  • Marty Weintraub

    gabriel bear and CPCcurmudgeon, thank you for the comments. You each raise good points. Thank you for visiting.
    Marty

  • Jon Myers

    Hi Marty

    Just a quick note to say thanks for the blog covering my session at SES last week. Hope you enjoyed the content and the new format.

  • Marty Weintraub

    You’re welcome Jon. I did enjoy it.
    Marty