Skip to main content
Getting the Elephant to Dance: Better Modeling for Big Data

Banks and other financial institutions (FIs) have always had Big Data. But now, with improved data modeling and closed-loop analytics, FIs can finally get closer to the real-time decision support models that other types of industries already enjoy. Business analysts can create and deploy new decision models to manage portfolios of credit card and loan products in just weeks, not months or a year.

In other words, this powerful, but cautious, elephant is starting to dance.

Kudos for the elephant

For decades, FICO has helped banks build and implement decision models to manage their businesses. A recent article in McKinsey Quarterly, “The benefits—and limits—of decision models,” highlights what we have seen for years:

Examples of successful decision models are numerous and growing… Banks approve loans and insurance companies extend coverage, basing their decisions on models that are continually updated, factoring in the most information to make the best decisions.

Some recent applications are truly dazzling. Certain companies analyze masses of financial transactions in real time to detect fraudulent credit-card use….

All of these activities are things that we in the industry do every day. But the article also talks about the importance of human intervention in the models, to ensure that the models are updated appropriately to take into account any changes in the environment.

This is where things get exciting for FIs.

Learning new dance moves

We’re used to hearing about how companies like Netflix and Amazon, and even less-known ones like easyJet or Loblaws, are constantly adjusting their personalization algorithms to present the right movie, item, holiday package or reward offer in front of the right person, at the right time.

These companies use closed-loop analytics to feed new learnings directly back into their personalization models, and inject this iterative process with a heavy dose of creative intervention. Marketers are able to test their “what if” hypotheses quickly, gauging their success on sales results.

It’s not that easy in financial services. Beyond the obvious—that we work in a heavily regulated industry—deploying champion-challenger and other decision models has traditionally been a long, labor-intensive process requiring significant IT resources.

It can take several months to build a new model, and several more months to deploy it on the mammoth systems that are part of our world. By the time a new model is in production, a year may have passed. The conditions the model was created to reflect may easily have changed, completely.

Speeding time to value

Today’s new decisioning platforms can allow FIs to significantly improve the way their models reflect current conditions, and the speed with which they do, through the process illustrated below.

This figure shows:

  • A “feedback” loop (cyclical process) involving the FI’s client data and their decisioning application, where the client data is regularly fed into an analytic data mart. Once in the data mart, the data can be stored here and extracted by business analysts.
  • With the extracted data, analytic tools can be used to develop scorecards, and to develop and optimize decision trees and business rule flows.
  • Once the analytics are developed, the business analysts can deploy them back into the decisioning application. Here, the new additions can be included in simulations to determine if the analytics will perform as expected in a production environment. 
  • Finally, if the results of the simulations indicate that the newly developed analytics will perform as desired, these analytics can be introduced by the business analysts into the production environment, where they can be monitored for actual results—and a new cycle of analytic development and deployment can begin.

Production-ready in eight weeks

The elapsed time? New models can be developed, tested and in production in as little as 8 to 12 weeks, all without IT intervention. These kinds of breakthroughs are significant for our industry, presenting a considerable improvement over more cumbersome, IT-reliant methods.

To learn more about how your bank or financial institution can speed its time to value with new and improved decisioning models, check out the FICO white paper, “Closing the Loop and Eliminating Costly Delays in Analytic Development and Deployment” (Mar 2014).

Right now, I’m going to watch this already-classic Jimmy Fallon video clip to get a few dance moves.

related posts