New neural model builder & scoring engine creates powerful fraud models in under 12 hours, boosting fraud detection and cutting associated costs by 50%
17 November 2015 – Paris – Banks, merchants and payment providers can now cope with escalating payment data, while maximising fraud detection and prevention rates, using SmartScore, a new self-service, self-learning neural model builder and real time scoring engine from ai.
Set to transform payment fraud detection, SmartScore empowers users to build their own neural models, thereby shrinking model development and deployment timeframes from up to 6 months to less than a single day. Pilot tests have shown it can also slash associated operating costs by 50% and lead to a reduction in rule set size, complexity and false fraud alerts generated.
Providing easy-to-use, intuitive self-service tools, SmartScore enables in-house IT and fraud teams to respond quickly to threats by building multiple neural models and using them to protect millions of transactions, in real time, on the same day. It also ensures in-house data specialists spend more time on fraud prevention and less on data crunching, helping them deal with ‘threat information overload’ without having to outsource.
In addition, SmartScore simplifies workflow processes for security operation teams, reducing false positives and verifying real threats automatically. This minimises error and quickly focuses attention on the threats that matter most, dramatically improving the effectiveness of fraud prevention measures thereby reducing lost revenue and better safeguarding brand reputation.
Totally fraud platform agnostic, SmartScore can be used in any country to analyse data of any size or complexity and from any payment type or channel – be it traditional POS and ATM or mbanking – making it a true enterprise wide solution for accelerating decision making and enabling early fraud detection.
Reinforcing the potential for SmartScore, Kevin Sprake, Senior Vice President and General Manager of a leading Merchant Services provider, confirms, “From the evidence I have seen, SmartScore will allow me to produce a neural model in one day, when generally it could take over three to 6 months to deploy. Therefore, the speed of fraud detection and the action-window to stop the fraud are enormous. The other operational efficiencies we are likely to benefit from can bring savings of over 50 percent.”