Advanced analytics has a home in banking and financial services, but it transcends the traditional use of Excel and OLAP to run trend-lines on revenue and cost as they affect the classic Profit and Loss statement. While macro trends are important to the financial services professional, two more important challenges involve:
- understanding the customer at the Point of Sale
- signing up profitable customers to avoid further attrition and/or mitigate losses.
Furthermore, by virtue of the fact that new products attract a customer base of unknown quality, many banks need the above type of information right at the product roll-out phase when the least info is available as the prospect turns into a new customer. With SPSS analytics an analyst can develop a holistic view of customers that purchase two types of products;
- Non credit based products like checking accounts and savings accounts
- Credit based products like credit cards and mortgages.
Key point, non credit based products like checking accounts behave more like “commodities” so the type of analytics most appropriate for this product set will resemble retail analytics (ie use of trend-lines, forecasting and customer profiling). For credit-based products like credit card, the analytics is more rigorous and takes the shape of a two step process data mining process;
- A ‘quasi-underwriting’ step in which you pre-qualify prospects who might be at future risk for default based on key criteria like home ownership, bankruptcy history etc.
- A direct response model to make sure the ‘prospect universe’ left over from step 1 has a decent tendency to respond to marketing offers.
In fact, most credit card and mortgage products suffer from the adverse selection law of attraction (ie the customer an institution least wants to attract is most likely to respond to an offer of credit due to a lack of solicitations from the competition). To combat this, analytics is necessary to micro-target good credit credit risks. Let’s say we have an existing portfolio of consumers with high usage of checking accounts from a recency, frequency and volume of usage perspective. Picture the customer on a 3-dimensional RFM grid – See diagram:
So basically, if we assume that all customers have the same default rate, we would market credit cards to these customers without doing additional advanced analytics to predict default rates. But knowing this is not the case, we would have to overlay RFM with an additional step to find out where within the business rules associated with high-default risk lurk within the customer base (which would be a subset of the RFM grid shown above)
Our advanced analytics practice can provide more insight into handling these types of situations in the banking and financial services domain space. For more information, contact:
Tony Firmani
Director Advanced Analytics
Tony.Firmani@perficient.com
