PayPlug has developed innovative predictive algorithms in order to calculate in real time the risks associated to any payments. The goal is to feed our new fraud prevention solution, the Smart 3D-Secure.
Before PayPlug, fraud prevention tools were based on engine rules. It was necessary to manually configure scores of various rules and continuously update and tune them in order to maintain acceptable performances.
These engine rules are still widely used by some large e-commerce companies.
This type of solution is not convenient to handle, both for the fraud prevention tool provider and for the companies using them. In order to be maintained and optimized, it requires a dedicated team. This way to proceed showed its drawbacks.
PayPlug's new fraud prediction solution is a breakthrough compared to traditional solutions.
Instead of being backed by engine rules, PayPlug has developed a solution that is based on predictive analytics and can learn by itself (machine learning). Some complex mathematical models are supplied by some large historical data sets (« big data ») and learn how to detect fraud based on historical patterns and previous fraudsters' behavior in versatile situations.Our models noticeably analyze buyers' behavior on various websites, credit card information, localization, clients historical data. These large data sets are processed in real time and allow our mathematical models to be extremely accurate and improve over time. On the other hand, they are so sophisticated that it is very hard to draw conclusions a posteriori about some causes of risk scoring. This is the price to pay for precision.
The development of this groundbreaking solution is the outcome of huge research & development efforts. Our team is hard at work and thrives to continuously improve this solution, which benefits from European funding, thanks to the programme Horizon 2020.