obr3588x_ch10_319-367 10/18/04 15:38 Page 344

EQA

344 ● Module III / Business Applications

REAL WORLD Wal-Mart, BankFinancial, and HP:

CASE 2 The Business Value of AI

S ome managers still think that artificial intelligence—

the decades-long effort to create computer systems
with humanlike smarts—has been a big flop. But exec-
utives at many companies know better. Artificial intelligence
(Al) is often a crucial ingredient in their stellar performance.
set up predictive models. PredictiveMarketing will reduce
the time it takes the bank to develop a model by 50 percent
to 75 percent, says William Connerty, assistant vice presi-
dent of market research. The first major application is a
model to predict customer “churn,” the rate at which cus-
In fact, AI is now a part of the technology of many industries. tomers come and go. It will be used to identify the customers
AI software helps engineers create better jet engines. In fac- most likely to leave the bank during the coming month. The
tories, it boosts productivity by monitoring equipment and problem is, the model has access only to account information
signaling when preventive maintenance is needed. And in prepared from weekly and monthly summaries, not to the
the pharmaceutical sector, it is used to gain new insights into daily customer activity that would make it more timely.
the tremendous amount of data on the human genome. “The biggest obstacle is getting transaction data and
The finance industry is a real veteran in such technology. dealing with disparate data sources.” Connerty says. The
Banks, brokerages, and insurance companies have been rely- data that BankFinancial needs in order to assess customer
ing on various AI tools for two decades. One variety, called a loyalty comes from several bank systems and unintegrated
neural network, has become the standard for detecting customer survey databases. A lot of systems integration and
credit-card fraud. Since 1992, neural nets have slashed such interface work needs to be done before the bank will see the
incidents by 70 percent or more for the likes of U.S. Bancorp full fruits of its modeling tools, Connerty says.
and Wachovia Bank. Now, even small credit unions are re- HP Enterprise Systems. Hewlett-Packard (www.hp.com)
quired to use the software in order to qualify for debit-card has an Enterprise Systems Group that pulls together people
insurance from Credit Union National Association. with diverse backgrounds and strong analytical skills for its
Wal-Mart. Like banks, retailers collect huge amounts of group that does predictive modeling of customer behavior.
data. Wal-Mart Stores Inc. (www.walmart.com), for in- The group is part of “CRM operations” under a vice presi-
stance, harnesses AI to transform that raw data into useful dent for sales, says Randy Collica, a senior business/data
information. Wal-Mart consolidates point-of-sale details mining analyst.
from its 3,000 stores. Data-mining systems sift instantly HP uses software from SAS Institute Inc. in Cary, North
through the deluge to uncover patterns and relationships Carolina, to mine its database of customers and prospects,
that would elude an army of human searchers. Data-mining using AI techniques to predict customer churn, loyalty, and
software typically includes neural nets, statistical analysis, where to target promotions. HP also mines its huge stores of
and so-called expert systems with if-then rules that mimic unformatted text data from its call centers, including e-mails
the logic of human experts. The results enable Wal-Mart from customers and prospects and text typed in during voice
to predict sales of every product at each store with uncanny calls with SAS predictive analytics for text. The techniques
accuracy, translating into huge savings in inventories and use “lead ratings” of call center personnel’s assessments of a
maximum payoff from promotional spending. caller’s readiness to buy—coded as “hot,” “warm,” or “sus-
BankFinancial. The potential for mining cost-saving and pect”—to predict the customer-led rating of noncoded text
revenue-boosting ideas from data is increasing as companies sources with 85 percent accuracy, Collica says.
build bigger data warehouses, computers grow more power-
ful, and vendors of analytic software introduce products that Case Study Questions
are easier to use. But while many of the products that can an-
1. What is the business value of AI technologies in busi-
swer those questions use esoteric techniques such as neural
ness today? Use several examples from the case to
networks, logistic regression, and support-vector machines,
illustrate your answer.
they don’t require a PhD in math, users say. Indeed, the
biggest stumbling block to using “predictive analytics” is 2. What are some of the benefits and limitations of data
getting the data, not analyzing it, they say. mining for business intelligence? Use BankFinancial’s
That has been the case so far at BankFinancial Corp. experience to illustrate your answer.
(www.bankfinancial.com) in Chicago. It uses the Clementine 3. Why have banks and other financial institutions been
data-mining “workbench” from SPSS Inc., to develop mod- leading users of AI technologies like neural networks?
els that predict customer behavior so the bank can, for ex- What are the benefits and limitations of this technology?
ample, more accurately target promotions to customers and
prospects. The bank uses Clementine’s neural network and
Sources: Adapted from Otis Port, Michael Arndt, and John Carey,
regression routines for these models. “Smart Tools,” BusinessWeek, The BusinessWeek 50, Spring 2003;
It’s also beginning to use PredictiveMarketing, SPSS’s and Gary Anthes, “The Forecast Is Clear,” Computerworld, April 14,
new package of “best-practice templates” for helping users 2003, pp. 31–32.