What reengineering was to the early 1990s, data warehousing has become for the latter part of the decade.
Establishing huge, centralized data bases that can improve customer service, profitability measurement, and product sales has become an obsession. Nary a conference passes without mention of data warehouses, and few bankers will admit to not having at least looked into constructing one.
But like reengineering, data warehousing has been slow to pay dividends. Bankers are finding that building data bases is difficult and expensive, that maintaining them is harder still, and that the pursuit of analytical and marketing benefits resembles a quest for the Holy Grail.
Nonetheless, bankers and consultants say the cause is just and victory awaits in the form of better customer service and more capability to tackle nonbank competitors.
"Improving the way we, as an industry, use the information available to us is a must," said Michael R. Zucchini, vice chairman at Fleet Financial Group. "Data warehouses will help us to do that."
Historically, banks have stored data for each product line in distinct "silos." As a result, a bank customer looking for information on a mortgage payment, a credit card balance, and a checking account might have to call three separate places.
An effective data warehouse breaks down the silos and centralizes the customer information. The warehouse can let employees view entire customer relationships without jumping from system to system.
"What banks are struggling to get is a unified field for the customer," said Wayne Eckerson, senior consultant at the Patricia Seybold Group in Boston.
He noted that warehouses can support more than just customer service. Once data is centralized, profitability measurement becomes much easier.
"The real goal," said Mr. Eckerson, "is not to get rid of the unprofitable customers, but to either change the pricing or create products that enable them to be profitable."
Other applications include supporting tightly targeted marketing to small segments of customers.
Though all of these applications are attractive to bankers, many find it hard to get them off the ground. Constructing a warehouse involves almost every department and its technology.
As a result, organizing data in a useful way is taking many banks longer than they expected.
"It's easy enough to say, 'We've got all these clients that are going to want this service,' or, 'Let's give branch platform people incentives to do sales referrals'," said William M. Saubert, vice president in the financial institutions group of Electronic Data Systems Corp.'s A.T. Kearney consulting subsidiary.
"But when you look under the covers to see what it actually takes to make that happen, it's ugly," said Mr. Saubert.
He said data warehousing is the hardest type of technology project because it requires a fundamental change in a bank's computer infrastructure.
Before anything useful can come out of the warehouse, data silos must be rigged to feed the centralized data base with information presented in a consistent form.
"Because you're dealing with so many different kinds of feeder systems, one of the major problems is scrubbing the data so that the info that populates the warehouse is clean and accurate and usable," said Mr. Saubert.
The complexity of these and other tasks has caused many banks to reassess their approach.
A few years ago, most institutions embraced what Mr. Saubert calls the "Field of Dreams" approach-build an entire warehouse first and let the applications follow.
But senior bank managers have grown increasingly insistent on seeing returns from mega-investments. With many warehouse construction projects dragging on longer than expected-several large banks have been in the building process more than two years-the pressure mounts to see some payback on millions of invested dollars.
Action Systems Inc., a Dallas-based consulting firm, has formed an alliance with Hewlett-Packard Co. and other technology providers to smooth banks' path to "customer knowledge" and incremental sales. For banks that cannot or do not want to do everything at once, the group offers gradualism-but in the context of longer-term vision.
"We have seen lots of pieces of method, but there was no overall way of thinking about it," said Action Systems chief executive officer Robert Hall.
Many banks are creating mini-warehouses to support specific applications. These smaller systems, known as data marts, buy time to do everything right with the main warehouse.
"It's important to get something off the ground in three to six months," said Rick Millem, solutions marketing director for financial services at Hewlett-Packard.
For institutions in the early stages of constructing a warehouse, Mr. Millem and Mr. Saubert both advocate creating applications that can be delivered during the building process. Examples include targeted marketing campaigns and sales incentives for specific departments.
Also, Mr. Saubert recommends paying close attention to middleware, the software that sits between the data warehouse and other systems.
"The more broadly you apply middleware, the better off you are," he said. "You are going to have to connect things eventually, so that's a layer you're going to have to put in place-whether you're going to stitch together five or 20 data streams.
Some institutions will choose not to design applications until the general warehouse is built. Mr. Zucchini said Fleet would take this tack because he expects the enterprise-wide data base to suggest applications that could not have been fathomed in the early stages.
But consultants said only the largest banks were likely to have the resources and patience to wait a few years for payback.
Even in the absence of tangible benefits, banks are increasing their investments.
Mentis Corp., a bank technology research firm in Durham, N.C., said U.S. banks spent just over $1 billion last year on equipment, software, and external services related to data warehousing. Mentis projects $1.75 billion by 2000.
These numbers exclude staffing expenses, which could double the cost of building a data warehouse, observers said.
But bankers are optimistic that they will more than offset the costs. Many see great promise in new-product sales.
Mr. Saubert said A.T. Kearney was working with a client that wants to help branch staff refer potential brokerage customers to those who can sell securities.
"They want to track the referral and they want to track if a deposit product, like a CD, is maturing and is going to be put into a securities product," he said.
"Today, that is a gigantic fax and E-mail recordkeeping headache. One of the things that will let them grow is creating a data warehouse application that allows them to watch all those flows in a highly automated way."
Although many companies may be able to get complex data warehouse applications off the ground in the near term, observers said the banking industry is years away from realizing the full potential of their data bases.
"If we're a quarter of the way there, that's being generous," Mr. Saubert said. And the warehouses are likely to remain works in progress that change with each new data stream.
"Banks have to rethink information technology, not just in terms of cost reduction," said Steven Carpenter, president of Action Systems. "They have to change behavior-the way the organization uses information technology."
"Get familiar with what this technology enables you to do, get some real benefit from it, and leverage it," Mr. Saubert advised. "It can only become more powerful and cheaper over time."
