Forensiq, which focuses on the detection and prevention of ad fraud, on Thursday released technology that finds and exposes the latest botnets creating invalid traffic to steal advertiser dollars. The latest bot hit sports sites in which the algorithm flagged up to 75% of pre-bid requests to NFL team domains.
The algorithms analyze browsing behaviors to identify patterns associated with automated browsing that simulates human traffic.
"Bots browse differently than humans," said David Sendroff, CEO of Forensiq.
Sendroff said the technology is self-learning, so it can pick up fraud easily. It analyzes impressions data.
Companies that provide ads to publishers, in particular, should benefit most from the improvement, which allows them to detect suspicious sources and keep their ad supply as clean as possible.
When the algorithms flagged 75% of pre-bid requests to NFL team domains running through the company's supply-side partners, it prompted further analysis.
It further prompted a study of team sports and sports media websites such as NFL, NBA, NHL, and MLB, leading to the discovery of Sports Bot. During the study more than 9.7 billion pre-bid requests analyzed translated into an estimated cost of $200 million to $250 million, estimates Forensiq.
The ads load on the sites without the sports team's awareness of involvement. Sendroff said there is much more video inventory because it tends to have higher CPMs.
Rebuilding the Forensiq platform has enabled the company to process more than 4 trillion requests per month -- about 1.5 billion per second at peak times.
The company also created Forensiq Fraud Labs after being acquired last year by Impact Radius, with the expertise of an expanded technology team.
Data scientists and ethical hackers sit in this business unit to drive faster development of new algorithms. The development of technology has been accelerated by the combined teams.