StupidScript - 11:29 pm on Apr 5, 2005 (gmt 0)

np2003, how do you document your click fraud reporting to Google? How does it compare to the clicks you see in the reports Google generates? Are you absolutely certain that Google isn't just capitulating because they don't have the man-hours available to track down every click fraud report?

I mean, if you record 30 instances of clicks from the same IP address on the same term within X minutes, do you see exactly that in your regular Google reports? Or do you see 30 instances in your logs and Google reports 20, leading you to request refunds for the 10 missing fraud clicks? I bet you request a refund for all 30 clicks. ;)

I'm going to stand by my earlier statement that they are doing a good job of filtering fraudulent clicks out before the bill gets to you, AND to reiterate that no system could ever track 100% of all fraudulent clicks. np2003, your tracking may simply be supplementing the work Google is doing, filling in the gaps so you get refunds for most of the fraud.

If that is so, then I again stand by my statement that programmatic filtering PLUS human oversight is relatively reliable, and that you are adding the human element that Google simply cannot include, so you may be catching more fraud than Google's programmatic approach alone can possibly catch.

I wanted to add that, in the early days of click fraud before G and Y started trying to deal with it, our company used to submit click fraud reports and requests for refunds every day, as we had an employee whose only job it was to seek out click fraud evidence and package it for reporting. Our competition forces us into the $10-$80 CPC range. Our average refund in those days was around $400/day out of a $45,000 monthly spend. Now, we can barely find $20 worth of fraud each WEEK. That's a massive improvement, and the reason why I'm sticking on G's side in this discussion.