Given that predictive analytics software is increasingly easier to use, it’s no surprise the technology is being adopted more and more in the financial services industry. In general, it is applied there in two ways:
- Against customer data
- Against internal and market data for risk management
While both uses are predictive, there are large differences between the results. Using customer data, banks and other financial institutions are applying the technology to predict customers likely to churn and then taking action to prevent the churn from occurring. Predictive analytics identifies customers likely to churn, then segments those customers by profitability, volume, and length of engagement.
Once segmented, banking business analysts, often working in tandem with marketing and sales teams, again apply the technology to optimize marketing campaigns that ensure exactly the correct incentives are offered to each class of customers. This results in higher retention rates at lower costs, and can improve the customer experience by more precisely offering promotions that appeal to them.
Financial services institutions also use predictive analytics to segment customers and predict which ones will react well to cross-selling promotions. Since it is widely reported that it costs credit card issuers, for example, up to $200 to attract each new customer, banks are eager to recoup their costs early in customers’ lifecycles. Cross-selling is a popular strategy for doing so. Mortgage borrowers may be open to opening retirement accounts, while credit card users may be interested in mortgage offerings. Predictive analytics is ideal for classifying which customers are likely to respond to offers for additional products and services, allowing banks to achieve profitability in the near term, and add to the bottom line over time.
Altogether, predictive analytics of customer data, particularly including digital self-services, presents a 360-degree, forward-looking view of individual customers. In the increasingly competitive financial services industry, predicting how to retain good customers, what new services will appeal to them, and how best to manage these relationships can be critical competitive advantages.
Risk management, while not always associated with directly adding to the bottom line, can be supported by predictive analytics for activities such as analysis of internal emails -- in any language -- for anomalies that might indicate plans for rogue or non-compliant trading, creditworthiness, commercial property valuations, and market volatility. By applying text and data-mining to internal communications, financial services firms can look for particular words that might indicate plans for non-compliant or rogue trading and act to prevent those activities.
Predictive analytics can also be used to predict and segment quickly which customers are most creditworthy. Likewise it can forecast commercial property values for properties in investment portfolios, and those held as collateral. And, in the unfortunate event of default, the application can forecast the best methods of managing the process, potentially reducing losses.
Predictive analytics is also of value to traders managing stock options. One application uses the VIX measure of US stock volatility. With predictive analytics using VIX data relative to other data, traders can make better informed decisions to buy, sell, or hold Vicks options over 30-day periods.
No longer the sole domain of data scientists, predictive analytics is providing even better and faster ROI in financial services. Predictive analytics solutions have evolved to be easy enough to use by less technical business analysts, while still providing the deep feature set and power required by data scientists. In addition, the customer data and risk management applications are well understood by financial services executives tasked with purchasing innovative technology that doesn’t strain IT resources or exclusively require expensive data scientists.Ingo Mierswa is an industry-veteran data scientist, beginning with the development of the RapidMiner platform at the Artificial Intelligence Division of the University of Dortmund, Germany. He has authored numerous articles about predictive analytics and big data. ... View Full Bio