Posted by Alison Arthur and Bethany Frank on 01 Jul 2019
Machine learning is a subset of artificial intelligence that allows computers to “learn” from data without explicit programming from humans. In financial services, it can transform business processes related to customer service, personal finance, and fraud and risk management. Here are some real-life applications of machine learning in each of these areas.
Chatbots are being deployed in financial services to automate common customer service requests, answer frequently asked questions, and assist with transactions like making bill payments. They use natural language processing to understand what a person is saying and formulate an appropriate response in return. However, chatbots can sometimes struggle with interactions that go off-script from specific instructions they’ve been programmed to follow.
Machine learning helps customer service chatbots assimilate information from each interaction and “teach” them how to respond in the future. For example, machine learning can teach chatbots how to identify frustration and adapt subsequent responses. This might mean replying with calming language designed to diffuse the situation or perhaps redirecting the frustrated customer to a live agent.
Machine learning can also be used to improve the customer experience. For example, if a customer tends to pay the minimum amount due on their bill at the latest possible date, this behavior might indicate that the customer is facing a cash-flow crunch. Machine learning can identify these patterns and offer the customer a different due date, a payment plan, or even a personal loan to help improve their ability to make on-time payments.
Budget management apps powered by machine learning provide customers the benefit of highly targeted financial advice and guidance. These apps allow customers to track their spending on a daily basis using their mobile devices. Machine learning can then analyze this data to identify spending patterns that customers might not be aware of and identify areas where they can save.
Machine learning helped one international bank identify struggling customers and provide better options for serving their needs. When the system detected unusual spikes in late-night credit card usage among certain customer segments, combined with low savings rates, the bank was able to determine that these customers were facing some level of financial stress. To reduce the risk of defaulting, customers tagged by the system were automatically given credit limit increases and offered free financial advice.
Financial services organizations handle a staggering amount of customer and transaction data that must be scanned for fraud. Payments, in particular, are a hotbed for fraudulent activity and organizations are constantly looking for new ways to strengthen fraud prevention and risk management protocols to handle these demands.
Companies like PayPal are using machine learning to enhance its fraud detection and risk management capabilities. Through a combination of linear, neural networks and deep learning techniques, PayPal’s risk management engines can determine the level of risk associated with a customer within milliseconds. Similarly, financial services organizations can replace statistical risk management models with machine learning algorithms. Several organizations use programs that scan transactions for risk, move flagged transactions to a risk queue, and investigate suspicious activity automatically. Continuously expanding knowledge bases improve the ability of these programs to detect (and prevent) fraud over time.
Insight gathered by machine learning also provides financial services organizations with actionable intelligence that acts as a foundation for subsequent decisions. For example, a machine learning program could tap into various data sources to assign risk scores for loan applicants. Algorithms could then predict which customers are at risk for defaulting on their loans, allowing the bank to tailor its services or adjust terms for each customer.
Machine learning can also help lenders determine the creditworthiness of potential customers. By analyzing past spending behavior and patterns, a system could identify how much credit should be extended to a given customer. The technology would be especially useful for new customers that lack long credit histories. Automating credit and risk scoring processes on a mass scale can help these organizations enhance their models across the board.
The Bottom Line: The applications of machine learning in financial services extend far beyond these few examples. Machine learning shows promise in helping the overall financial system enhance security, deliver better service, and increase operational efficiency - and that’s just the beginning.
*This is an update on an original post published October 2016
13 Oct 2020 Blog Alacriti Insurance Series Part III: How Digital Disbursements Extend a Lifeline While Controlling Costs Insurance companies looking to deliver fast and efficient payouts to policyholders should consider digital disbursements. This blog makes the case for why.
22 Sep 2020 Blog Shortening the Distance Between You and Your Customers Customer-centricity is a key tenet to any digital transformation project. But what does that actually mean? For us, it means meeting your customer, where and when they want to be met, with how they want to pay.
09 Sep 2020 Blog The Importance of System Integration to Drive Core Banking Transformation. What do the words "fully integrated" mean from an Electronic Bill Presentment and Payment (EBPP) perspective? Here's a list of seven questions and answers to help you get started.