Detect Search Partner Fraud with FouAnalytics

While search is a powerful channel for connecting users with relevant information, search arbitrage can undermine its effectiveness. As I've previously detailed, search arbitrage sites deliver low-quality traffic from content recommendation platforms, which doesn't compare to the quality of users organically found on Google, Bing, Yahoo directly. Search partner sites also include "parked domains" those pages with no active website, just search ads disguised as content.

We've consistently found high delivery to humans on well managed search campaigns that exclude search partners, but when campaigns turn search partners on, they deliver large volumes of ads on search arbitrage sites like below, and click-throughs to the site (which is what we see below as the "referrer" (the page before the advertisers' landing pages).

You probably don't recognize those sites, but that's not the reason they are fraudulent. We have multiple ways of confirming the traffic coming from these sites is bots. If we look at one of the top referrers to the client's websites, we can see the traffic is almost all coming from iPhones and Mac Laptops. We would expect a normal website that humans visit would show a variety of platforms and devices. And by the way, these are faked iPhones and Macs (bad guys just declared that in the HTTP USER AGENT); that's why HTTP_USER_AGENT alone is not good enough for fraud and bot detection).

The big clue was in the click charts. There were lots of clicks on the site, but they were obviously automated. The charts below are broken out by screen resolution for easy visualization. Each red dot represents a click on a specific x,y coordinate (a specific pixel on the screen). The larger the dot means the more clicks on that specific pixel location. Even if there are CTAs or menu buttons we don't see so many humans clicking the exact same pixel. These clicks are neatly lined up all down the left-side of the screen. This is enough to fool the search network into thinking these are engaged users and then sending more of the same traffic.

The next image shows the mouse-movements for those same screen resolutions. We would not typically expect to see mouse-movements on the 2 mobile-screens to the left, because iPhones don't use a mouse. Meanwhile the desktop resolution 1680x1050 has no mouse-movements, but many clicks in the chart above.

This corroborates that these aren't real iPhones and Mac laptops, like we mentioned above. There are of course other secret methods we use to detect that these are not real iPhones and Mac laptops, but we can't share it all. We need to target real devices and real humans to drive business outcomes. The automated clicks seen here didn't drive any sales for the client or improve their bottom line in anyway. All this spend did was line the pockets for the search arbitragers.

If you want to check the quality of your search ads traffic, you can use FouAnalytics on-site measurement. The tag only takes a few minutes to install on your site and you'll be able to see in depth insights like this. If you'd like a trial to "see fou yourself", contact me and I'll get you setup.