Initially, it was an application in search of a name. Now, four years after launching what it now calls its "data warehouse" project, First Interstate Bancorp almost inadvertently finds itself in the midst of the latest information technology movement.
"This was very early to start such a project in the industry," said Tim Sullivan, senior vice president of technology at the $56 billion-asset holding company, based in Los Angeles. "Two years ago, very few people even knew the term data warehouse."
First Interstate has been joined by a host of other large banks that are also constructing data warehouses. These banks "are driven by their need for better access to the data that is captive on legacy applications," said Bill Bradway, a consulting analyst with The Tower Group, Wellesley, Mass.
"Banks need to have a better view of the dimensionality of their customers, business processes, and risks. This view allows them to better understand risk exposure, profitability, and who their most valuable customers are."
That need translates into large expense: The Tower Group estimates annual data warehouse expenditures among the top 500 U.S. banks at $450 million, and predicts that figure will grow 30% per year.
Though First Interstate's technologists may have struggled to find a name for their undertaking, they did have a clear concept of its mission: Create a single repository containing data culled from all the bank's applications, to be accessed by employees in marketing, finance, human resources, cash management, and other areas for use in department-specific applications.
Applications using the warehouse include risk management, credit analysis, balancing, cross-selling, and a wide array of reporting.
There are now more than 100 projects involving the data warehouse - half of them entailing applications for various lines of business, half addressing warehouse infrastructure issues.
First Interstate's data warehouse project began as part of its efforts to consolidate data centers and get its 13-bank, 11-state network operating on common third-party applications.
"At that time (1991), we had a fair amount of skill in data modeling and design," said Mr. Sullivan. "So we decided to develop an integrated end- user computing environment."
According to John Chatfield, vice president and division manager for data services, the benefits of First Interstate's data warehouse are twofold.
First, "Packaged software does not facilitate cross-bank reporting," he said. "What we're able to do is give people that kind of reporting more cost effectively, certainly more cost effectively than changing the underlying applications."
The other major benefit is the ability to see across applications. "For example, we can combine lending information with information from the deposits area and tie that to a customer, then balance it against the general ledger. This (the data warehouse) is the only place where you can have that kind of integration," Mr. Chatfield noted.
First Interstate's project is substantial in scope. Information from 20 major applications, including commercial and consumer lending, deposits, customer information files, and accounting, is fed into the warehouse.
"What is unusual about First Interstate is that when we say an application is on the warehouse, that includes every one of the banks, and all of their data," Mr. Chatfield said with a note of pride.
Although the common systems that feed the data warehouse are mainframe- based, data from nonmainframe systems eventually will be added.
For example, plans call for an IBM OS/2-based teller system to provide data to the warehouse. The warehouse, which resides on an IBM mainframe, stores about 500 gigabytes of data currently. Mr. Sullivan expects the warehouse size will top a terabyte once transaction information is added later this year.
The automatic extraction of application data that is performed involves more than 1,200 mainframe tasks.
"It's a very complex process," said Mr. Sullivan, adding that besides data extraction, quite a bit of data scrubbing and reformatting must be done.
The warehouse feeds smaller departmental servers, or "data marts." The bank is adding massively parallel processors running in the Unix environment.
About 1,000 employees, using both mainframe-based tools and PCs, currently have direct access to the warehouse. But the information gathered and analyzed is disseminated to a much larger population throughout the entire organization.
The group of direct users spans a wide spectrum throughout the bank, including business executives, mid-level staff and high-end, technically sophisticated users in areas such as cash management and finance.
That number is expected to grow, said Mr. Sullivan. For example, as a result of recently delivered point-and-click tools for accessing the warehouse, the Human Resources area alone will add 400 users.
Constructing a data base of the size and complexity of First Interstate's is not without pitfalls. For starters, success tends to give rise to demand for more information.
"There's an insatiable desire for information on the part of the businesses," Mr. Sullivan said. "As you open up information, more is wanted. Dealing with that demand can be a problem."
During the ongoing construction of the warehouse, First Interstate has brought its efficiency ratio to below 60%, in line with industry leaders.
The reporting capabilities provided by the data warehouse contributed to that improvement. With the major components of the warehouse now in place, management believes it can be used not only to deliver productivity improvements and cost savings, but to boost profitability as well.
Instead of using the warehouse simply to aggregate and report data, end users currently can, for example, access the warehouse on an ad hoc basis, collecting information and merging it with data from other sources, such as marketing customer information files, to develop targeted direct mail campaigns. But First Interstate wants to go beyond this standard type of reporting process in which users already know what they are looking for. It is moving toward data mining, which, Mr. Sullivan said, involves discovery processes and the detection of patterns and associations.
Along with these efforts, the bank will integrate currently separate information sources - such as the marketing files - into the warehouse. As a result, it will have created a much more broadly based, enterprise-level warehouse, said Mr. Chatfield.
The bank's data mining strategy will be aided by the planned capture and analysis of customer transaction data, which will provide a wealth of information on customer behavior.
"When you get into transactional analysis, you're coming up with new information based on customers' banking channel preferences and buying patterns," said Mr. Chatfield. "More discovery is possible."
Although the bank will be able to accommodate the huge volume of data associated with capturing customer transaction data, it is still figuring out the most effective way of analyzing the information and putting it to work in marketing efforts.
"What you do with the data is the trickier part," said Mr. Chatfield, adding that the bank is exploring advanced technology such as neural networks to assist in analysis.
The bank plans to use its expanding body of information, which will include customer segmentation analysis, to generate leads for more proactive sales efforts next year.
In addition to its brick and mortar branches, the results of the bank's data mining efforts will be provided to staff at its 24-hour telephone service centers, which are being consolidated from the current nine into two - in California and Nevada.
"When John Smith calls the phone center, we'll be able to notify the customer service representative that he is a prime candidate for a home equity loan, based on our analysis of his behavior and relationship with us," explained Mr. Chatfield.
And, conversely, "We're hoping to come up with the perfect list of people to call that will be highly likely to buy a profitable product or service," he added.
The ability to capture and analyze customer transaction data will help the bank enhance its credit card portfolio as well, bank officials believe.
Currently, Mr. Sullivan acknowledged, the bank does not have sufficient useful information from its portfolio. That is expected to change as the bank's credit card systems are added to the data warehouse and card transaction information is captured.
"We'll learn things like where people shop and what they buy, then be able to figure out what they will likely do next," explained Mr. Sullivan.
Data mining can also be used to assess customer profitability. According to Mr. Bradway, this ability is key to making a data warehouse truly valuable.
"The question you have to ask is, if we measure profitability down to the customer level - millions of customers - what are we going to do with the information? Do you give it to the branch people, the phone center people?" asked Mr. Bradway.
Making the best use of this information - that is, turning unprofitable customers into profitable ones - requires creativity on the part of banks, he said.
First Interstate had profitability systems in place on its previous applications. Working with software vendor Treasury Services, the bank plans to implement an integrated profitability system that will initially focus on business unit and product profitability. Customer profitability will be incorporated "down the road," according to Mr. Sullivan.
In the meantime, the bank is conducting limited analysis of customer profitability on the retail and wholesale sides.
"If you're going to get very good at data mining, from a marketing standpoint, you must know who your most profitable customers are to begin with," he added. "That knowledge helps in selecting the prospects that have the highest probability of the next profitable sale."
First Interstate has enjoyed a head start in the data warehouse arena over those banks that run disparate software applications across multistate networks and must struggle to integrate information, according to Mr. Sullivan.
He hastened to add, however, that "this advantage we have will last only if we can make money from it in the long term." v
Ms. O'Heney is a freelance writer based in New York.