English to Greek Other IT (Information Technology) banking

Let me give you a couple of examples of real-world situations where I’ve seen this applied quite powerfully. There was one large bank in the US, which had not refreshed their small-business underwriting models in several years. Certainly not post the crisis. And they were getting worried about the discriminatory power of these models.

The Gini coefficient of their models—which is just a measure of how powerful a model is in terms of its ability to discriminate between good risks and bad risks—was down in the sort of 40- to 45-percent range. What these folks did was developed a 360-degree view of the customer across the entire relationship that that small business had with the bank, across all the silos. Not easy to do; easy to talk about, not easy to do.

And then what they did was selectively append third-party data from external sources, trying to figure out which of those third-party pieces of information would have the most discriminatory power. And they applied the analytical techniques to redo their models and essentially took the Gini coefficient of the models up into the 75-percent range from the 40- to 45-percent range, which is a huge improvement in the discriminatory power of those models.

Another bank that we were working with was in the developing markets, where data to begin with is pretty thin on consumers. And they decided that they would try to actually get data from the local telco. The paying behavior for the telco is actually a great predictive indicator for the ***credit behavior*** with the bank. And so they bought the data, appended that to the bank data, and again had a huge improvement in underwriting.

Another institution, a marketing example, where we ended up using, again, that 360-degree view of the consumer and then appending some external data around social media to figure out what’s the right next product to buy for that consumer and then equip the front line to make that offer to that consumer when they walk into the branch or when they call into the call center. And the efficacy of their predictor models on next-product-to-buy improved dramatically as well.

So these are examples of things that you can do. And part of the reason why this is so important is that in the banking world, of course, in the current regulatory and macroeconomic environment, growth is really, really, really hard to come by.