February 7, 2019 | Written by: Dave Hemingway
Categorized: AI/Watson | Financial Markets
Share this post:
The New Payments Platform (NPP) opened up an opportunity for Indue to reinvent our financial crimes service due to the potential increased risk with NPP over traditional channels. Our relationship with the IBM Safer Payments solution has resulted in the following benefits:
- Lower false positive rates by 20% improvement
- ability to make rule changes 90% faster
- manage all payment channels in one system
At Indue, we strive to offer innovative payment and fraud containment solutions to our customers, which are generally credit unions, building societies and tier II financial institutions. Our shared-services solutions help them compete effectively against bigger institutions, which can afford to maintain standalone, in-house services. We can continually increase the scale and scope of our services while keeping costs down for our customers.
When Australia launched its New Payments Platform (NPP) real-time payment platform in early 2018, we saw an opportunity to reinvent our financial crimes service. Real-time payments mean greater risk of fraud and cyber crime, as fraudsters can move money extremely fast. We needed to create a real-time, full-service offering to help our customers meet the challenge presented by the NPP.
Real-time payments require real-time fraud prevention
With the launch of NPP, money can move in less than 15 seconds. Our traditional near real-time fraud monitoring solution was no longer suitable to effectively mitigate fraud with NPP. We realized that a real-time solution encompassing machine learning, or AI, was the most appropriate solution to control fraud risk while being operationally efficient to operate for analysts and fraud officers.
With the launch of NPP, money can move in less than 15 seconds. Our traditional rule-based fraud monitoring system wasn’t fast enough for NPP. We realized that a real-time solution would require machine learning, or AI, to deliver a robust response and bring fraud risk down without interrupting payments or compromising the overall customer experience.
We identified IBM Safer Payments as the ideal solution to bring AI and real-time detection capability to fraud detection. Our real “aha” moment, if you like, was the realization that we could use it to create a single cross-channel fraud detection offering rather than adding a separate real-time solution just for NPP. We determined this would be far more effective and cost efficient moving forward. Furthermore, unlike other solutions that require continuous vendor support, the IBM solution lets us control and manage critical elements such as rules changes.
Thanks to AI, the Safer Payments solution not only examines the payment transactions as they occur but also analyzes additional metadata information about the customer, for example previous transactional and demographic information. The solution takes in all this information in real time and responds with a risk score or decision that is more intelligent, accurate and efficient than a traditional rules-based system.
Driving efficiency through cost-effective fraud detection
We’ve seen a reduction of more than 20 percent in false positive results compared to our previous detection systems, for fraud across all channels, not just NPP. False positives are expensive because every positive flag needs to be reviewed by our team. Fewer false positives increases our operational efficiency, allowing us to deliver cost savings to our customers over time.
We’re improving efficiency even further because we’re using the solution to manage all payment channels with a single system rather than multiple systems. Safer Payments is, overall, less expensive to run than our multiple existing systems. And because it’s a cross-channel solution, we can bring in additional data sources to further improve the machine learning capability of the total system.
Another critical benefit and efficiency driver of Safer Payments is the ability to quickly adjust rules when we see fraud patterns changing in the industry. Our fraud analysts can write, test, and deploy rules extremely fast—in a matter of minutes—allowing us to effectively stop more fraud. With traditional fraud systems, rule changes can take hours or even days, depending on the complexity of the change. We’ve seen a 90 percent improvement in the time needed to make these critical rule changes, which means we can stay on top of fraudsters and fight fraud more effectively and efficiently.
AI – the future of fraud detection and beyond
Safer Payments is our first AI-powered platform at Indue, but it’s probably not our last. We see this implementation as the first step in a continuing journey to implement AI in other areas of our business.
For other companies looking to implement machine-learning capability into their systems, I would offer one key piece of advice. Prime the system correctly with the appropriate transactional data so that the engine can learn. Map your data appropriately, especially if you’re using proxy channels. Once you get those elements right, you’ll have an effective rollout and can start seeing the benefits of the AI solution.
Listen to Dave Hemingway talk about Indue using AI to support real-time fraud detection: