ZIFFPAGE TITLEModel PiecesBy Mel Duvall | Posted 2004-11-01 Email Print
Bank of America spent $47 billion to acquire FleetBoston to get a stout presence in New England and New York. The bank will spend tens of millions to convert its 1,500 branches onto its Model Bank platform, so it can provide the same services nationwide.
The Model Bank was Lewis' baby, conceived in the early 1990s when he was president of NationsBank's retail division.
NationsBank's leader at the time was Hugh McColl, a former Marine who kept a grenade on his desk and was in the midst of a buying binge. Between 1988 and 1993, McColl acquired 25 institutions, transforming North Carolina National Bank into NationsBank from its headquarters in Charlotte.
But acquiring scores of banks meant acquiring scores of information systems that didn't necessarily work together. Branches in North Carolina operated on different systems and offered different products than those acquired in Texas, Georgia or Missouri.
The systems had difficulty even exchanging basic information, such as amounts held in a checking account; a bank in another state often would not recognize a person as a NationsBank customer. And, at the time, the bank appeared more interested in saving money from its acquisitions than spending millions to integrate the systems it had purchased.
Steve Romaine, a former senior vice president at NationsBank charged with managing the bank's relationships with technology consultants, points to his own experience. When he joined the company in 1996, he opened an account at the branch in the Charlotte headquarters. A few months later, he was transferred to the Atlanta office and walked into a downtown branch to cash a check.
He couldn't. The Atlanta branch was acquired from C&S/Sovran in 1991, and the two banks were operating on different computer systems.
"I was told, 'Sorry, you'll have to contact your bank in Charlotte,' as if it were a completely separate bank," he says. Romaine left the bank in 1997.
The inability to recognize Romaine or another out-of-stater as a customer could have been fixed by then.
In 1992, Lewis hired Systematics Financial Services, a Little Rock, Ark., computer services company, to create a single set of applications for the NationsBank branch network.
At the time, NationsBank was offering about 150 retail products, from general savings and checking accounts to more specialized accounts such as Christmas Savings, seniors savings accounts and specialized loans for students and small business. The accounts and products were largely supported by disparate banking applications that became part of the mix from acquisitions.
Lewis wanted to pare down to a standard set of 35 retail productsthe Model Bank. Systematics was awarded the job largely on the strength of a relationship with what was then Andersen Consulting, now known as Accenture.
Lewis dubbed the project Vision '95: To have the bank's 1,700 branches conduct business with common products, systems and operating procedures by the end of 1995.
It was a huge effort. The bank assigned an initial team of 250 programmers and budgeted the three-year development effort at $200 million. By 1997, when the 50 banks finally had been converted to the computing platform, about $1 billion had been spent.
Such a switchover is always fraught with problems, says Mike Blum, global leader of the banking practice at IBM, which supplied the mainframe computers and the operating system for Model Bank. For starters, you may have multiple instances of identical account numbers. There may also be more than one master file of customer information, leading to extensive scrubbing of names, addresses and other account information to make sure the new master is accurate and up-to-date. Plus, there are always new technical wrinkles that can redefine what data has to be kept and how it has to be dealt with, like the rise of online banking with the expanding popularity of the Internet in the mid-'90s.
At the time, Bank of America had more than 50 databases to work through. That meant manually checking the migrated files for errors, incomplete information and duplication, such as two customers with the same name or account number, or missing phone numbers or ZIP codes. There are tools to help automate this process, such as SAS Data Quality Solution, that use rules, algorithms and look-up tables to find errors. "But at the end of the day, you will want to have someone go through the files and look at them manually," says Jay Desai, a Chicago-based manager in the banking and financial services practice of Tata Consultancy Services. "You can't afford mistakes."
The biggest challenge though, says IBM's Blum, is converting systems while still keeping your doors open for business. "It's the equilibrium challenge," he says. "You've got to continue to service the customer and sell the products at the same time."
With Model Bank in place, the bank could know what accounts and services a customer already hadand could begin targeting individuals and businesses that it believed would buy additional products.
With the unified system, a teller could instantly know that a customer had made a large deposit and ask, for instance, if the customer would like to speak to an investment adviser. Similarly, it could prompt a teller to ask a customer who had recently applied for a mortgage whether the person would like to obtain insurance.
"It's a common practice now, but in the old system where the applications were separate, you couldn't do that kind of thing," says Alenka Grealish, chief analyst in the banking practice at Boston-based research firm Celent Communications.
Simplifying its systems would also save money. Instead of having to maintain dozens of different banking systems, there was now just one (or two, considering California). A bank as big as Bank of America spends about $3 billion a year on technology, including personnel costs. Gartner, the Stamford, Conn., research firm, estimates that as much as 80% of a typical information technology budget goes toward supporting and maintaining systems.
At Bank of America, Lewis believed fewer full-time employees would be needed to serve customers because so much information would be readily at hand, and there would be just one system to learn or deal with. As a result, more part-time employees could be hired for peak periods, such as lunchtime, Fridays and paydays.
"Also, the reduction in the number of products by itself made [bank operations] more streamlined," says Mark Salamasick, a former Bank of America executive who spent most of his 20 years with the company as director of internal audit for its information systems.
In all, the bank says it was able to spend 12% to 15% less on branch staffing, or the equivalent of about $120 million to $180 million in annual savings. After operating in a new territory for more than one year, the bank says employees were able to sell 35% more products, from credit cards to investments, using Model Bank.