Most of our competitors and half the industry are putting their faith in certain black lists to keep fraudsters, cheaters and bots out. These lists are supposed to filter and let only the good actors inside, so the ads will fulfill their purpose and generate awareness/interaction/sales among their viewers.
We believe this practice treats fraudsters as little kids behind their computers who are unable to change or mask their identity. It is similar to putting up a shoplifter’s picture in a grocery store who wears a big red hat. Well, they are capable of much nastier things than to change their hat to green. We have been seeing a pattern lately of the same criminal groups getting blocked out and coming back again to the same place wearing different identities.
We are using behavioral analytics as their fake behavior is one key identifier they can change only with a large investment. The program that mimics human behavior is easy to catch with machine learning. We analyze interactions with the website & the creative and using over 50 heuristics we apply machine learning to build a criminal intellectual property database. Big Data analytics helps us to learn and understand the difference between how a human interacts with the site and how a computer program does.
As an added bonus this not just lets us differentiate between human and machine but also between humans and humans who are actually visiting a completely different site as it does happen during domain spoofing and arbitrage fraud. Here the underlying impressions and users are real, but the fraudster is taking an undervalued asset like a leaderboard on a torrent site, and masking it as a premium asset, such as that same leaderboard appearing on a first-tier news site. In these cases we will see the different user behavior and go after the fishy business.