Models of success

AXA Financial finds success in SAS® data mining, analysis

At New York-based AXA Financial, customized "books of business" for financial professionals covering 2.4 million households, along with automated client reassignment and customer-satisfaction programs, have led to the success of its Mining Diamonds program.

AXA Financial is making use of SAS Analytic Technologies, says Melissa Corso, AXA Financial's assistant vice president of CRM analysis.

Her group supports Mining Diamonds by building cross-sell models with SAS Enterprise Miner. SAS Enterprise Miner allows Corso to pull 25 samples from the 2.4 million households and begin running models in minutes, while pulling in as many as 200 fields in the analysis with more than half a terabyte of data feeding SAS.

"SAS Enterprise Miner is so scalable that it allows us to build our models more efficiently," Corso says. "I can build models in just a few days, once the data is collected, as opposed to writing programs and doing analysis. And the models are more predictive than anything we had originally."

They're also more accurate. "With the assessment node in SAS Enterprise Miner, we can quickly see the lift we're getting from our models," Corso says. "You can have 10 models, for example, and see the effect and figure out which ones are most accurate. Then we can go back and apply that to recent data in our databases to see if that model is holding up on new information."

Building a holistic view
SAS was critical in helping AXA Financial build a holistic view of its customers for analysis and effective client management, says Raghib Muhammad, AXA Financial's director of data warehousing and business intelligence. Muhammad's group is the primary IT support group for the CRM analysis and marketing business groups.

Running an IBM DB2 platform with a data warehouse that includes DB2 client files, agent files and adviser files as well as an IBM CIIS platform, AXA Financial has a two-way flow of data with SAS. Imagine the product administration system, the general ledger and the human resources systems on the left; enterprise data in the middle; and the consumer system on the right. These data sources feed software applications, including those from SAS. In turn, results from SAS flow into enterprise data, customer and agent files and the data warehouse, which primarily includes product and financial data. Those data sources feed individual data marts, which provide the customization necessary for Mining Diamonds.

"Most consumer applications flow from left to right, which sets the stage for SAS analysis," Muhammad says. "The result of the analysis flows from right to left, which makes it available to a wider audience of people who can take action on it."

In AXA Financial's intelligence infrastructure, operational, enterprise and business-specific data stores rely on SAS Enterprise Miner, Base SAS and SAS/STAT to provide analysis for customer-value, retention and cross-sell analysis. Those results aid the planning process for campaign, content and channel management, which then enables event tracking, personalization and cross-product support. Feedback from the execution of those programs is added to the data stores in a continuous loop of information sharing.

And the two-way flow of intelligence through SAS allows AXA Financial to achieve goals without having to hire additional IT resources. "SAS gives us the ability to just go in and directly get the data we need," says Greg Winsper, AXA's senior vice president of marketing.

Automation the key to success
At the heart of the architecture, SAS/ACCESS to Oracle and DB2 feeds data from various customer-value and Web-based data marts – as well as from the IBM DB2 mainframe warehouse and Oracle open-systems warehouse – into SAS applications that run on a Sun 4500 server with four processors and a 280-gigabyte disk.

Mining Diamonds and other successful programs at AXA Financial are the culmination of Muhammad's ability to get intelligence out to a wide audience using SAS. "Three years ago, we weren't doing as good a job as we are now," he says. "Intelligence is only as good as your ability to use it. The right people must have access to it, and SAS allows us to make sure that they do."