Click fraud is a very complex issue. Unfortunately, much of the information available has been misinterpreted due to that complexity. In this video, Shuman Ghosemajumder of Google gives some information that will hopefully clear up that gray area.
For starters, click fraud is driven by 2 major incentives:
1. Attacking advertisers – when an advertiser tries to hide one of his competitors
2. Inflating affiliates – when an AdSense publisher tries to fraudulently increase his own revenue by generating false clicks
Methods of click fraud stem from simplistic techniques to sophisticated techniques. Simplistic methods include manual clicks such as when a fraudster personally clicks ads on his own computer. It gets a little more sophisticated when a fraudster organizes botnets and uses them for click fraud. Click farms, which consist of individuals or organizations that try to hire people to click on ads continuously, fall somewhere in the middle of the simple and sophisticated click fraud techniques.
Shuman does point out that: “the impact of invalid clicks at Google is minimal.”
Google detects click fraud through simple rules and statistical anomaly detection. Google’s simple rules consist of certain rules they have already defined as what they classify as an invalid click. Since some simple rules can be easily broken, statistical anomaly detection is more effective because it looks at specific activities on websites and compares the expected behavior to the observed behavior. This data gives Google a better understanding of how to detect invalid clicks.
For advertisers wanting to take proactive measures to prevent click fraud, Shuman advises keeping the return on investment (ROI) as the central focus. Research and gather as much data as possible, test everything, and track all results. If you apply these actions and your ROI drops for no reason, you have a good reason to suspect undetected click fraud and should file a claim.