Why Big Data Will Not Solve Your Problems
Having more data is unquestionably valuable, but data has been “Big” for a long time now and simply collecting huge volumes of data does not give you a competitive advantage. Where you gain an edge is being smart in how you use the data. Here are two areas where retail banks could be doing better:
More Branch-Focused Analysis. Banks have dozens or hundreds of intelligent analysts performing in-depth analysis of Big Data to inform customer-focused strategies. This includes propensity models for which product should be cross sold next, models to predict attrition likelihood, optimization of direct mail offers to customers, analysis to inform which promotions should be displayed when a customer banks online and identifying who should be targeted in call center campaigns.
Curiously, this rigor and volume of analytics is not applied to branch-based decisions. The same bank that extensively evaluates whether to mail a blue or green envelope or offer a 3.99% or 4.99% promotional credit card rate will spend $50 million per year on branch renovations with minimal analysis to inform that decision.
Major branch decisions are expensive, such as operating hours, staffing levels, remodels/renovations and adding new technology. Understanding which branches deserve reinvestment and how to roll out a program for Saturday hours, for example, can make the difference between adding tens of millions to the bottom line and losing tens of millions.
The same rich data sets banks extensively analyze at the customer level are available on the branch level. These include demographics of the surrounding area, competitor locations relative to each branch, the profile of the customers at that branch and macroeconomic factors like unemployment and home values in the area. The same focus applied to customer analytics should be applied to branch analytics to use this information and measure how branch initiatives are working and determine how to roll them out to maximize return on investment (ROI).
You Have to Test! The next opportunity is in how data is leveraged. Banks extensively leverage correlation analysis, regression modeling and other related advanced techniques. The problem is that these approaches can result in recommendations that are expensive, risky and ultimately do not work. For example, regression analysis of branches with longer vs. shorter hours might indicate that longer hours are correlated with higher new account generation. But branches with longer hours could perform better for a range of reasons from quality of branch staff to competitive density in the area and adding hours may actually have no impact and just add cost.
Top retailers across the globe are disciplined about testing new ideas before rolling them out and the best will run hundreds or even thousands of tests each year. This avoids costly missteps and identifies which new ideas actually will work well. Banks, unfortunately, often lack this disciplined culture of testing. Big Data is most valuable when combined with in-market testing. Rather than correlate the wide range of data available with customer or branch performance, implement a test, measure the impact, and then leverage the data sets to figure out where the program worked well.
To demonstrate the importance of testing, consider this highly simplified example: regression analysis shows suburban branches generate more new accounts with longer hours while additional hours do not produce the same magic in rural and urban branches. Too often, banks will take this information and extend hours in suburban branches, which is a risky and expensive move. The right approach is to test longer hours at a handful of branches that span urban, suburban and rural locations to prove out this idea. You might find that adding hours actually has no impact on suburban branches. The suburban branches with longer hours that seemed to be doing better in the regression analysis are actually doing better because they are in higher market share areas and have more tenured branch managers.
Big Data is an overused buzzword that receives too much attention. Having more data is incredibly valuable and banks have the luxury of having more data than most. They just need to utilize it properly.
Mr. Weidman is senior vice president at Washington, D.C.-based Applied Predictive Technologies, a data analytics firm that helps retail banks test how various business changes alter customer behavior. He can be reached at [email protected].