The Knowledge Exchange / Risk Management / Big data improves speed, accuracy and automation for credit lenders

Big data improves speed, accuracy and automation for credit lenders

Data for risk management - 2012Credit lending is a fast-paced business that relies on large amounts of customer and transaction data. Premier Bankcard’s primary purpose has been to provide even customers with damaged credit histories an avenue to obtain credit and demonstrate positive financial patterns.

To stay competitive, the company uses analytics to predict and model customer data, as well as to detect fraud. Here, Rex Pruitt, Premier Bankcard’s Manager of MIS Profitability and Risk Requirements, discusses how high-performance analytics can help meet the challenges of today’s financial industry.

Waynette Tubbs: What does big data mean in the credit card business?

Rex Pruitt: Big data means information or transactions that hit capacity limits in our systems. Organizations have to deal with big data from a computing standpoint with thresholds and cutoffs because of storage limits, processing speeds and the like. Authorizations, for example, are a huge data set in the credit card business. Every time customers swipe their cards at a grocery store or a gas station, they require an authorization or pre-authorization. And those kinds of transactions have to happen instantly. To do that requires more information faster and it has to be accessible.

Authorization is only one challenge. There’s other data about our customers and people we want to be our customers. Because of server space limits – processor and memory – we have to have “purge rules” or retention strategies for data. If you’re purging on a 60-month rolling rule, you won’t have five contiguous calendar years of data to do your analytics, forecasting and modeling.

The law of large numbers is what prevails in statistics. The more data you can feed into your model, the more accurate it’s going to be in reality. But unfortunately we have to sample down, which incurs bias and can influence the outcome. It’s very possible you could end up with something that isn’t really what’s going on in your business. I’ve always said more is better.

Tubbs: How does high-performance analytics help foster big data?

Pruitt: Any business experiencing growth understands the need for more data. If high-performance analytics were incorporated, then you could process against very large databases. And the data would be turned into real information. If you’re looking for a needle in a haystack, high-performance analytics gets to it quicker. High-performance analytics could shrink a 96-hour process down to four hours, having a dramatic impact on output, which is really the ROI for the company.

Tubbs: So automation support is a determining factor?

Pruitt: Process automation is incredibly important. With the manual process, an underwriter would have to evaluate each application and try to spot potential fraud. Now high-performance analytics, coupled with the ability to score every record and feed it into the system electronically, can identify fraud faster and more accurately. That reduces loss and creates better customers when we get them in the door. Before we had to rely on probability based on a sample and that would be an expense to the company. The bottom line impact of high-performance analytics is revenue either through expense mitigation or the quality of business getting better.

Tubbs: How else does high-performance analytics improve your business?

Pruitt: It’s important to evaluate your current customer base on an ongoing basis. Most institutions do this monthly on a batch basis. To be able to look at that portfolio on a daily or even hourly basis improves scoring on the fly for authorizations. The relationship is truly around how much you can process. Speed is one thing, but you also need capacity.

Tags: , ,
  • Facebook
  • del.icio.us
  • Twitter
  • Digg
  • LinkedIn
  • email

Post a Comment

Your email is never published nor shared. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>