As business grows, so do data storage needs. For Kogarah, Australia-based St. George Bank ($74.3 billion in assets) -- which relies on customer data to develop marketing strategies, for example -- the need for additional data capacity became apparent in March 2004. At the time, Peter Karvinen, the bank's program manager, global information systems, projected that in a little more than a year, the financial institution's 2.1-terabyte data warehouse, NCR's (Dayton, Ohio) Teradata V2R5, would run out of room.
Simply adding capacity and creating siloed data marts was not an option, stresses Karvinen. St. George uses a centralized data warehouse to obtain a holistic view of its business and it costs 10 times more to generate reports for compliance, for example, from separate data marts than it does from the single data warehouse, he explains.
So the bank compared the newest version of Teradata's data warehouse software, V2R6, with a product from another vendor that Karvinen declines to name. Based on a recommendation from BearingPoint (McLean, Va.), which the bank hired to aid with the decision, as well as St. George's internal assessment, the bank decided to implement Teradata's V2R6.
According to Karvinen, the V2R6 data warehouse was implemented over one weekend in April 2005. The software arrived preloaded on a single server, replacing the old data warehouse on another unit, he relates. St. George Bank's data warehouse now comprises 4.7 terabytes and runs 30,000 batch jobs a week, Karvinen notes, adding that users can load and access information around the clock.
The bank uses the Teradata data warehouse to monitor accounts for everyday operations, such as collections actions, but also leverages customer behavior information to determine pre-approved offers, shadow limits, fast-track applications and other marketing strategies, Karvinen relates. Further, the data warehouse is the repository for all of the bank's credit risk data. St. George uses a BI tool from Business Objects to gather information and generate reports from the data warehouse.
The bank wants to save customers time and money, Karvinen says, so all communications need to be relevant to customers' personal circumstances. "The evidence shows that events-based marketing really does work," he points out.
Implementing a new data warehouse was expensive, but worthwhile, according to Karvinen, who declines to cite specific figures. The bank would not have been able to build it on its own, he adds. But installing a data warehouse is an ongoing process, not a project, Karvinen continues.
In order for executives to create reports, the tool will require usage simplification, Karvinen explains, noting that installation is only about 10 percent of the job. The rest comes from adaptations that users need to ensure the data from the warehouse meets their needs. "You cannot predict how [the process] will end up," Karvinen says. "You can, however, shape its destiny."