Techtalk: Fair Isaac Corp.'s James Taylor on automating the decision process

By Jeffrey Rothfeder  |  Posted 05-15-2006
You know those pesky credit scores, the number that sums up a lifetime's worth of trying to keep out of debt? They were invented by Minneapolis-based Fair Isaac Corp., which recently translated its credit score concept into a model that can be applied to nonfinancial data as well.

The aim of their Enterprise Decision Management system is to forecast future customer transactions and design sales and marketing strategies based on those predictions. CIO Insight asked James Taylor, Fair Isaac's vice president of product marketing for enterprise decision management, to discuss how organizations can implement the technology to better control data.

CIO Insight: How might enterprise decision management affect information governance?
Taylor: Governance involves two things: making the right choices about how to use data to get the most value out of it, and having an audit trail. An EDM system lets companies make real-time automated decisions based on historical data, and corporate policies and regulations. That fits the first category of making the right data choices. Simultaneously, an EDM system provides a high level of data transparency, because it shows you which rules fired for each transaction. That's the audit trail.

How would such a system work in a real-world situation?
A typical call center will have a script that attempts to cross-sell a specific item to virtually every customer no matter who they are. An EDM system delivers a decision instantly, as the caller is talking, about the best cross-sell to make to this specific customer in these specific circumstances. It's a predictive model based on the past behavior of this and other customers.

It sounds like such a system will require a lot of private information. How do you address any security concerns?
Instead of doing live data mining on sensitive information, which is done with traditional business intelligence systems, with EDM the model and the outcome of the analysis is the only thing employees have to see—not the data behind it. The personal data can be kept in a separate network, and the model can be run throughout these networks without pulling up the confidential information.

Does EDM eliminate human decision-making?
To a degree. But with most high-volume, complex transactions, such as insurance underwriting and retail sales, there are more mediocre employees than superstars. Since you can embed any decision tree in an EDM system, these models can inject the best expertise in the company into everyone's sales process.