Mellon Bank Corp. thinks so much of Dreyfus Corp.'s data mining system that it plans to incorporate the model into its standard operating procedure.

Peter Johnson, vice president/strategic technology at Mellon Bank, said that at New York-based Dreyfus, Mellon's mutual fund subsidiary, millions of dollars have been invested in its Data Store, which contains investment and money information to serve customers better.

"All the technology investments have been guided toward customer service," Mr. Johnson said. "By having more customer information delivered to the point of customer contact, we can carry out direct sales over the Internet, telephones, and call centers. We can emphasize we're a retailer," he said.

The bank is trying to adopt the same approach, he said, but Dreyfus is further along in integrating different product relationships.

At the recent Data Warehouse World conference in New York, Mr. Johnson defined his six steps for data mining.

The first is to define the business problem, the second is to pull together the data for mining, and the third is to clean the data. "Steps one to three are what needs to go on in the data warehousing world. You can't do data mining if you haven't done those," said Mr. Johnson.

Step four is to pre-process the data, and step five is to model them.

"Have a consistent definition of what constitutes a product particularly after mergers and acquisitions," said Mr. Johnson. The final step is to deploy the model and make it work.

Most software vendors offer products that do modeling, Mr. Johnson said, but this is only "a relatively small part of the problem. Most of our time, about 50%, is spent on data acquiring and data cleansing. Data mining is 35% of the problem."

The bottlenecks, he said, are having the data to mine in a clean, consistent format. "In financial services, a lot of organizations have good data marts but when they want to cross-sell, they need an enterprise model."

That is one of the challenges the industry faces, he said, along with the large number of channel and product offerings.

"Theoretically all vendors agree to a common set of standards, except Microsoft, so their products can inter-operate," Mr. Johnson said. "But sometimes you're better off if you go to one vendor where the suite of products works together."

For three years, Mellon has partnered with International Business Machines Corp., using IBM's Intelligent Miner product for data mining. Although IBM may not be the cheapest, it "excels at customer intimacy and does well at product innovation," Mr. Johnson said.

"Mellon Bank forecasts a bright future for data mining," said Tom Kendra, vice president of data management marketing in the software solutions division at IBM. "It is part of its solution to increase cost efficiencies, improve service, and enhance cost management."

He said Mellon has taken a leadership role and is clearly an early adopter of advanced technology.

Last year, International Data Corp. ranked IBM as the No. 1 global data warehouse supplier, ahead of NCR Corp. and Oracle Corp. IBM's data base workstation license revenue has seen a 110% increase in the first quarter of 1998 over the first quarter of 1997.

Since May, IBM has begun to invest $47 million in business intelligence solutions known as Teraplex Integration Centers. Customers, such as Mellon's Mr. Johnson, can go to these centers to test the problems they are facing in the real business world.

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.