Reducing Fraud and Improving Customer Experience with Machine Learning
Julie Conroy is research director for Aite Group’s Retail Banking practice and covers fraud, data security, anti-money laundering, and compliance issues. Recently, Julie teamed up with ACI’s Marc Trepanier for a webinar, Key Trends in Payments Intelligence – Machine Learning for Fraud Prevention. I sat down with Julie to get her take on the topic.
Nidhi Alberti: It’s great to talk with you, Julie, thank you for taking the time. To set the stage, tell us a little about what you’re seeing and hearing about machine learning and its benefits from a fraud perspective.
Julie Conroy: Thanks! Glad to participate in ACI’s blog and webinar. Yes, depending on whom you ask, machine learning tends to have many different definitions. When we talk about machine learning at Aite Group, it comprises several different algorithmic techniques of data science to analyze data. The idea is to find the patterns within the data and then iteratively optimize over time. Banks can learn from that feedback loop that they get and continue to evolve their analytics. By understanding what ‘good consumer behavior’ is, banks can find those deviations that indicate fraud or financial crime activity.
When properly deployed and with the right data feed, banks can dramatically cut down on the false positives and improve operational efficiency with machine learning.
NA: In your conversations with banks around the world, what are some of the challenges you’re hearing from them around the implementation of machine learning for fraud prevention?
JC: When it comes to harnessing the data required to inform machine learning analytics, banks need to pull it in from disparate corners of the organization. Our research has shown that most banks struggle with it because it’s the most difficult and the most time-consuming part of the process. For example, to optimize the analytics, banks need to bring in insights from cross-channel, cross-product and various other parts of the business to provide a more holistic view of how the customer interacts with the bank. Not only so they can detect fraud, but so that they can also create a clearer picture of legitimate transactions versus fraudulent ones. That part of the data journey can take from six months to a year to implement in an effective way.
NA: So, how do real-time and faster payments fit into all this?
JC: We now see 42 countries around the globe that have faster payments deployed. For countries like Canada, where the national immediate payments platform is expected to go live through Payments Canada Modernization in the next 18 months, and the U.S., which has multiple schemes (Zelle, RTP, as well as push-to-card) it’s harder for banks to understand the points of vulnerability, or the points of data they need to bring in and implement so that they can provide the best experiences for both their retail and wholesale customers.
As we see faster payments coming into the market, we surveyed fraud professionals to see the trends associated with it. When comparing 2018 losses to losses two years ago, real-time payments fraud losses were the largest. There’s also a customer education issue, where they don’t realize that when they get duped into sending a large amount of money to people they don’t know (such as with social engineering scams), the liability is on them – they don’t have the zero-liability protection that they get from the card networks. When it comes to faster payments, banks need to ensure that security is in-step with innovation.
NA: So, what should banks look for in machine learning and fraud prevention?
JC: Banks need to understand that fraud prevention plays a key role in customer experience, and reducing friction with the customer is important. Most of the banks that I’ve spoken to are looking to take their current fraud alerts and, in real-time, interact with the customer. As good as your machine learning analytics are, you will have false positives, so if you have that real-time resolution capability, suddenly you turn the customer’s negative experience into a positive one. For example, if a bank detects that a customer has made an unusual purchase, based on their knowledge of the customer, they can decide to stop the purchase in real-time until they can resolve it with the customer and ensure it isn’t a fraudulent transaction. It improves the customer experience and gives consumers the feeling that their bank has their back in real-time. That communication channel with customers is important.
Want to improve customer service and reduce fraud? Download our guide: The Six-Step Guide to Leveraging Machine Learning for Payments Intelligence
Related Blog Posts
Turning British Players into Payers: Insights to Convert the Country’s 37.9 Million Gamers
As the sixth largest gaming market in the world, and with USD $4.5 billion in annual revenue, the United Kingdom presents an immense opportunity for game developers looking to drive profits. And with estimates of 40 million gamers pushing revenues above $5.5 billion by 2022, it’s imperative to understand this market and the payment preferences of its gamers.