Bankers are forging new ways to differentiate profitable customers from those who offer only the potential for profit. Quantifying the divergence between them takes sophisticated methods of synthesizing data that use the latest banking technology. The next step may be the creation of "smart" systems that learn from shifts in customers' behavior over time, and can predict their changing profitability through different life stages.
To be sure, consultants and marketers have been exhorting banks for years to consider customer profitability in their marketing. Executives want to segment consumers by their behavior, rather than simply by geography or demographics. But the technology to do so accurately has lagged.
Only recently has system integration technology advanced to the point where customer scoring has become an integral part of many banks' operating procedures--at a typical cost of $5 million to $10 million for new systems, although they can run much higher. That investment generally takes a year or two to pay for itself in new revenue.
Given the expense, it's not surprising, perhaps, that profitability scoring systems are most prevalent in larger banks. According to a study last year by the GartnerGroup, the Stamford, CT-based information-technology consulting firm, three-quarters of banks with more than $4 billion in deposits were calculating current customer profitability at year-end 1999, and almost all of them planned to be doing so by the end of 2000. In contrast, fewer than half of banks with between $250 million and $4 billion in deposits scored their customers on profitability, although another 36% planned to by the end of this year.
The investment may be worth it to a bank that wants to make the most of its relationships with customers. "We have to look beyond the customer's ability to pay for products and services," says Jim Neckopulos, managing partner in charge of the Pacific region of Andersen Consulting's financial services division. "Bankers need to know degrees of willingness and desire to pay for a bank's offerings."
The Gartner study found that banks use their analysis of customer profitability to develop new products, determine prices and identify customers who should (or could) be moved to other products and ways of receiving services.
But the Gartner study indicates that many banks have been less than pleased with the results of their efforts. When asked to rate the applications of customer profitability that they supported on a seven-point scale, with a score of 1 meaning "very ineffective" and 7 "very effective," fewer than half the banks called their applications effective (a 6 or 7 rating).
Still, the winners in this arena are clear. Chris Formant, a consultant in PriceWaterhouseCoopers' global banking practices division, cites Citigroup, Chase Manhattan Corp., Providian Financial Corp., Capital One Financial Corp. and American Express Co. as financial institutions that have successfully merged the latest technology with cutting-edge business practices. "They are dealing in a response and risk environment simultaneously," he says.
"They've been able to develop the most elegant models," Formant adds. "The abstraction of almost all data is achievable using today's best tools, although there are only a handful of companies that have predictive capability so far. There is still a big separation between the people who are really good and everyone else."
Even the rudimentary software of the 1980s revealed that "three to five percent of the customers were generating half the profits," says Robert Hall, CEO of Xchange (formerly Action Systems), which serves the enterprise customer relationship management sector. Not surprisingly, banks took notice. For years, they've been building data warehouses and using that data to determine customer behavior patterns and better allocate marketing resources.
But understanding the numbers isn't enough. Bank executives must place an "end-to-end emphasis not only on software but on its implementation to be incorporated into a set of business practices," Hall says. "Three things lead to useful outcomes: One is to more effectively target opportunities. Second is the value proposition, the products that fulfill the target opportunities. Third is to deliver those products effectively through different channels."
Hall says that life-cycle variables may be factored in as well. "How do major events--marriage, divorce, birth of a child, graduation--impact customer behavior?" he asks. "We know, for example, that half of college graduates buy a new car within two years of graduating."
According to Hall, that sort of information can be used to predict a customer's changing profitability over time. "Each transaction becomes a data point in the map of where that customer might go," he says.
But although the technology plays a crucial role, banks won't replace their marketing executives with computers any time soon. "Programs will become more automatic, but there is a portion that requires judgment," Hall says. "The trend is clearly toward more personalized campaigns. The last 5% of customers may be too elusive in terms of quantifying their habits."
Take $9 billion-asset Sanwa Bank, based in Los Angeles. Renette Shigenaga, executive vice president of emerging technology applications, and Mona Chui, vice president, say that executives review customer scores before starting marketing campaigns based on them. Given Sanwa's size, Chui says that most of the work in creating scoring systems is done in-house, but Shigenaga notes that "proprietary, outsourced software figures into our scoring system."
Customer scoring is just one key element in banking system evolution that will culminate with so-called market-of-one programs. Theoretically, banks are aiming toward a model that centralizes activity within regions, but that to customers "it appears as if a product offering has been uniquely tailored for the recipient," says Andersen Consulting's Neckopulos.
Mary Adam, senior vice president at Little Rock, AR-based ALLTEL Corp., a provider of core systems and services to banks nationwide, says that large institutions "may have as many as 100 different score cards to keep track of a wide array of product offerings and customer geographies."
She adds, "The future may lie in intuitive systems."
Such thinking points to the development of "smart" bank systems in the future, whereby self-monitoring software incrementally adjusts scoring systems in near real-time as the bank's knowledge base grows.
If a bank has quantified all the costs of its score-based marketing efforts beforehand and knows the key risks, it may make perfect sense to send mailings to all those customers with similar scores. And the cost-effective way to do that is to have the system trigger marketing efforts.
The future Adam speaks of isn't far off. Already, a number of technology companies offer software and services that automate a great deal of the marketing decision process, including some predictive capabilities. And the technology is becoming ever easier for marketing executives to use without calling in programmers every time they want to change variables in their analyses.
For instance, Fiserv Corp. lets customers implement the weight of scoring variables and the number of variables. Its Windows-based tools readily integrate with outside demographic data, so it's easy to alter screens and fields to reflect system upgrades.
Steve Sammond, a Fiserv Corp. product manager who developed the company's credit scoring software engine, says his design goal is to make a "system flexible enough to handle any scenario, yet easy enough so that the end-user does not need extensive programming background to operate it."
Installing a good scoring system may be part of an overall upgrade program. Companies such as Electronic Data Systems Corp. often step in to fix upgrade-related problems and to thwart expensive failures. Accurate scoring systems result from "the acquisition of data to get good metrics," says Mike Schweitzer, a director with the global banking division of EDS. "We start with data clean-up."
EDS exports the data out of its clients' existing systems into its own system, where the data is reformatted and managed. Because most banks have, over time, built data warehouses in phases, when Schweitzer initially appraises a job, he finds that "more often than not systems are non-centralized. Usually there are multiple legacy systems, often comprising three environments: lending customers, deposit customers and customer-information files," he says. That's a common remnant of 1980s management techniques, which relied on software and hardware that today verge on extinction.
Once a bank's legacy data has been exported to EDS, a team creates an analytic, flexible system. Data can be transmitted easily between different areas of a bank's network, and EDS augments scoring capabilities with outside demographic data. "Of course, banks establish the scoring criteria," Schweitzer notes.
Some of his clients program the system to trigger marketing efforts and customer perquisites. Then the system might be programmed to track the results of essentially its own recommendations.
Indeed, the EDS approach, in many ways, reflects the qualities of a truly "smart" system.
But fast-changing variables in a bank's developing CRM campaign may make such automatic triggering a bad idea. A more hands-on approach, like EDS customer Sanwa's, may yield more favorable response rates.
"It remains to be seen," says ALLTEL's Adam, speaking of smart systems, "if such systems perform any better than the progression, regression techniques we've been employing over the years."
Neckopulos of Andersen Consulting considers $5 million to $10 million a fair amount to invest in such systems, with payback averaging between 12 and 18 months. "These systems can help banks determine what their core competencies are," he says, "and there's a lot of value in that."
John C. Hallenborg is a freelance writer living in southern California. He is a regular contributor to Thompson Financial publications.