Three Steps to Start Fighting Click Fraud
Just how rampant click fraud really is might be up for argument, but there’s no denying that marketers must take steps to detect it so they can request refunds from the pay-per-click (PPC) provider. What’s more, says John Marshall, CEO of ClickTracks Analytics, a Web analytics software firm in Santa Cruz, Calif., marketers also might want to simply modify the ad to reduce the likelihood of fraudulent clicks in the first place or even pull the problem ad outright.
Regardless of what actions you take to combat click fraud, you first have to identify ads that are drawing dishonest clicks. Marshall offered the following tips to last week’s ad:tech conference attendees, during a panel session he took part in titled, Pay for Performance: Braving the World Beyond Pay-per-click.
• You will get better insights from studying your logfiles than your Java when analyzing click activity.
• Some metrics you should be capturing for each ad to compare against the performance of all campaigns:
1) number of sessions;
2) revenue conversion rate;
3) total cost of campaign;
4) number of single page sessions;
5) number of different IP addresses;
6) number of clicks from a particular country;
7) number of sessions with no referrer; and
8) number of sessions that reach any goal page.
• Look for commonalities across all your online ads’ performances—especially those ads that offer higher payment per click. Normal click activity tends to be fairly consistent over time when you run the same program elements, whereas click fraud usually surges up and then declines just as rapidly. In other words, you are searching for activity that doesn’t look like regular human behavior but rather the work of an automated computer program or some shady characters. Marshall’s No. 1 telltale sign: A very short time on site compared to other ads.