Fraud detection and prevention has always been a top priority for banks, especially as customers conduct more banking online across a wide range of devices, increasing the chances of fraudsters getting their hands on your customers hard earned cash.
Not only that, cyber criminals are becoming more advanced with their methodology all time, increasing the need for financial bodies to up their game when it comes to fraud fighting technology.
According to banking body, UK Finance, scammers stole £616m from UK bank customers during the first six months of 2019. That works out at over £1m a day, up 40% on the same period in the previous year.
To add to that, Barclays research has revealed that fraud costs UK businesses up to £35,000 per incident. Unfortunately, this meant that companies across the UK had to make 50,000 redundancies in total just to recover costs.
Traditionally, banks would use a human-written rule engine to identify fraud, but this only captures a small percentage of fraud cases and, counterproductively, churns out a high number of false positives. So how do you add an extra layer of security to your online banking processes without asking your customers to provide a blood sample just to prove it’s them?
Enter artificial intelligence and data analytics.
Nordic finance giant, Danske Bank, knows just how powerful these technologies can be when implemented effectively. The bank saw that it needed to modernize its fraud detection processes. At just a 40% fraud detection rate, managing up to 1,200 false positives per day the bank discovered 99.5 percent of all cases they were investigating were not related to fraud.
False alarms require a lot of time, money and people just to process them, these are valuable resources which could be put to much better use.
Danske Bank worked closely with a data analytics solution provider to modernize their fraud detection defence and reduce the high number of false positives they processed per day. The bank adopted a modern enterprise analytic solution with AI and Deep Learning capabilities. The model is able to identify potential fraud while smartly steering clear of false positives, shifting operational decisions from users to AI systems.
The deep learning systems compare models to determine which one is most effective, learning as it goes which traits are more likely to indicate fraud, all in real time. The model is constantly fed more data, such as geo-location or recent ATM transactions. When a system outperforms another, it provides a roadmap to successful fraud detection for the other models.
The transformational project certainly proved value for money. According to Danske Bank, they were able to:
- Realize a 60% reduction in false positives, with an expectation to reach as high as 80%
- Increase true positives by 50%
- Focus resources on actual cases of fraud.
The platform has even been made accessible to other team across the organisation including engineers, data scientists, and investigative officers, enabling them to collaborate on uncovering fraud, delivering even more business value to the rest of the bank.
Danske bank has certainly reaped the benefits of adopting a powerful and specialized platform. Armed with the power of AI and Deep Learning, the bank can now uncover fraud without being distracted by a huge number of false positives.
Do you want to uncover and act on hidden trends in your data to make smarter business decisions? Innovate with us.
Posted by Kate Auchterlonie