Bank of Montreal Aims To Score With Clients Using Data Mining Product

With customer relationship management the rage in financial services, the key to success isn't simply building a front-end application. A strong back-end data warehouse and data mining tools are also required.

For the past year, Bank of Montreal (BMO) has been trying to focus on customers rather than banking products. That means tearing down long-standing banking silos and sharing information across a broad range of divisions within the bank. To help in the task, US$155 billion BMO deployed IBM's DB2 Intelligent Miner Scoring (IM Scoring), which allows data mining to be conducted in a real-time environment.

IM Scoring allows the bank to apply an automated scoring process to its predictive models and segment, classify or rank a customer based on profitability, preference for buying specific products or even the likelihood of switching banks. Marketers can then use this information to respond better to a customer's unique demographics.

The scoring tool, which sits atop an eight-terabyte database, has enabled BMO to build a model that identifies which customers would be interested in buying a financial product, such as a mortgage. BMO is still testing the system but expects to roll it out next year.

BMO scores clients on a batch basis, but is building a real-time capability. With real-time scoring, "you can identify some relationships with clients that are in the dying stage without waiting till the end of the month," said Dr. Jan Mrazek, senior manager of business intelligence solutions at Bank of Montreal.

The bank concluded that this initiative, though costly, was necessary to stay abreast of the competition. "It's a big race. The key is who will get to the customer first and correctly recognize his or her needs," said Mrazek.

Still, deploying business intelligence solutions is no easy task. "You face lots of issues on the technical side," said Mrazek, noting the challenge of pulling information together from the bank's 18 million accounts across 32 different business lines.

Building a single view of client holdings can be tricky because clients' names could be spelled differently or addresses of accountholders could be out of date. "I have to be careful before I can say these two accounts go together," said Mrazek.

On the business side, deploying the solution requires a change in the way clients are viewed. If a client is looking for a break on mortgage rates, for instance, a bank employee might conclude that such a break would be unprofitable to his or her department, without realizing that the client has numerous other relationships with the institution that could be harmed by declining the request.

---

FAST FACTS

INSTITUTION: Bank of Montreal

ASSETS: $355 billion

BUSINESS CHALLENGE: Score customers based on profitability or propensity to buy.

SOLUTION: IBM's Intelligent Miner Scoring

KEY QUOTE: "You can identify some relationships with clients that are in the dying stage without waiting till the end of the month." - Dr. Jan Mrazek, Senior Manager of Business Intelligence Solutions