DATA WAREHOUSING SUPPORTS CORPORATE STRATEGY AT

FIRST AMERICAN CORPORATION

Brian L. Cooper
President, Investment Services Group
First American Corporation
First American Center
Nashville, Tennessee 37237-0603
Phone: (615) 748-2364
Fax: (615) 748-1540
Email: brian.cooper@fanb.com
Hugh J. Watson*
Terry College of Business
University of Georgia
Athens, Georgia 30602
Phone: (706) 542-3744
Fax: (706) 542-3743
Email: hwatson@blaze.cba.uga.edu
Barbara J. Haley
McIntire School of Commerce
University of Virginia
Charlottesville, Virginia
Phone: (804) 924-8981
Fax: (804) 924-7074
Email: bhaley@mindspring.com
Dale L. Goodhue
Terry College of Business
University of Georgia
Athens, Georgia 30602
Phone: (706) 542-3746
Fax: (706) 542-3743
Email: dgoodhue@cbacc.cba.uga.edu
* Contact person

We would like to thank Carroll Kimball, Jay Phillips, Theresa Leahy, and Connie White from
First American Corporation for their helpful comments on this paper.

Data Warehousing Supports Corporate Strategy

at First American Corporation
Introduction
In 1990, First American Corporation (FAC) lost $60 million and was operating under
letters of agreement with regulators. Today, FAC is a profitable, innovative leader in the
financial services industry. This change in fortune is the result of an ambitious strategic vision,
and a major investment in data warehousing that made the vision possible.
FACs strategic vision is called Tailored Client Solutions (TCS), a customer relationshiporiented strategy that positions FAC clients at the center of all aspects of the companys
operations. Though many organizations espouse customer relationship management, FAC has
redesigned every aspect of its operations to meet its clients' needs as well as its own profitability
goals. Underlying these efforts is the recognition that, to succeed with this strategy, it must know
its customers exceptionally well, and leverage that knowledge in product design, in distribution
channel decisions, and in every interaction with its clients.
The execution of this strategy would be impossible without a data warehouse called
VISION that maintains client behaviors (e.g., products used, transactions), client buying
preferences (e.g., attitudes, expressed needs), and client value positions (profitability). Using
information from VISION, FAC has:

identified the top 20 percent of its customers who provide virtually all of the consumer
profits, and the 40 percent to 50 percent of those who are not profitable;

developed strategies to retain the top high- value customers;

developed strategies to move unprofitable customers to lower cost distribution channels,

different products, or pricing structures that boost profitability, while still focusing on
customer needs and preferences;

developed strategies to expand relationships with all customers;

redesigned products and distribution channels to increase profitability and better meet
customers' needs and preferences; and

redesigned information flows, work processes, and jobs in the bank's branches in order to
meet customers needs and increase their use of profitable products.
To implement TCS, FAC had to change the way its employees think about banking and

about their jobs, shifting from "banking by intuition" to "banking by information and analysis.
All of these actions combined have moved FAC from losses of $60 million in 1990 to profits of
over $211 million in 1998.
This paper describes FAC's transformation and emphasizes the information technology
that was essential to its success. The first two sections discuss First American Corporation and
the Tailored Client Solutions strategy. Then it presents the VISION data warehouse and the way
in which it was implemented. The final sections describe the uses and applications of VISION,
the impacts that result from it, and lessons that were learned from the initiative.

About First American Corporation

Founded in 1883, FAC is a comprehensive financial services holding company
headquartered in Nashville, Tennessee. Its holdings include:

First American National Bank;

First American Federal Savings Bank;

Deposit Guaranty (acquired in 1998 and now operating as First American National Bank);

First American Enterprises, Inc.;

IFC Holdings (formerly INVEST Financial Corporation, the nation's largest third-party
marketer of investment products (98.75 percent ownership), headquartered in Tampa,
Florida); and

The SSI Group (the largest processor of hospital healthcare claims in the U.S. (49 percent
ownership) headquartered in Mobile, Alabama).
With operations in Tennessee, Kentucky, Virginia, Mississippi, Arkansas, and Louisiana,

FAC had $20.7 billion in assets, 7,195 employees, 391 banking offices, and 650 ATMs in 1998.
It had the largest deposit market share in Tennessee, the second largest deposit market share in
Mississippi, and the largest small business and middle market share in Tennessee. 1
Conditions were not as good, however, in 1991 when Dennis Bottorff became Chairman
and Chief Executive Officer. Because of FAC's problems, there was concern that the bank would
be closed, and that the larger banks would come in and pick the pieces they wanted. Bottorff
and his management team started by fixing the broken things in the company; however, they
realized that FAC needed a long-term strategy if it were to survive in the increasingly
competitive banking industry. It needed a new way of thinking and a new business model. FAC
could not be the low cost provider because it lacked the economies of scale of larger financial
institutions. Product differentiation was not a feasible strategy because other banks could quickly
duplicate products that showed promise in the marketplace. It could not compete for large
accounts (e.g., FORTUNE 500 companies), because these are the province of large national and
internatio nal banks.

Tailored Client Solutions

Bottorff decided FAC should focus its attention on three market segments -- consumer,
small, and mid-size businesses -- in which the bank could compete effectively. It also needed to
shift from the traditional banking orientation on products to being much more customer-centric.
This new customer focus required a data warehouse to support it. Top management felt that the
effort should be driven from the marketing department, and after a year- long search, Brian
Cooper was hired as the Executive Director of Marketing.
The marketing department, senior management, and key members of the finance
department worked together to develop the Tailored Client Solutions strategy. Figure 1 shows
the four interlocking components of the strategy: excellent client information, a flexible product
line focused on mass customization, a consistent sales and service approach centered on meeting
client needs, and a distribution management approach focused on channels of choice. All this
was coordinated under a broad new brand -- About life. About you. -- that put the customer at
the center of everything the bank did.

At the end of May 1999, it was announced that FAC would be acquired by AmSouth.
5

About life. About you.

Flexible Product Line
provide what the
client needs

Client Information
know the client
better than anyone

The Client
Distribution
Management
offer the client
preferred channels

Consistent Service
help the client
achieve goals

Figure 1: Tailored Client Solutions

Client information is the first component of TCS. All banks possess a great deal of
information about their customers, but often the information is not easily accessed, and most of it
is not leveraged in the many interactions that the bank has with its clients. FAC wants to know
more about its customers, and to utilize that knowledge in every aspect of its business. In
addition to demographic information and information about transactions with the bank (e.g.,
ATM card usage), FAC includes information about how its customers prefer to do business and
how profitable each customer relationship is to FAC.
The second part of TCS is a flexible product line. Although all banks can offer similar
products, it is more important to offer products that meet client needs and are profitable. By
understanding its clients and having accurate profitability information, FAC phases out or
modifies unprofitable products, and designs lucrative alternatives that meet clients' needs.
The third component of TCS -- consistent service -- focuses on determining and meeting
customer needs. Banks traditionally have sold either what the customer asks for, or is the hottest
6

product on the market at the time, but neither approach explicitly considers what is in the client's
best interest. TCS calls for a major shift in sales culture because a FAC representative begins any
customer relationship by discussing that person's current financial situation and future goals.
Then, together, the customer and representative explore products that would help the client
accomplish personal objectives. Because FAC representatives have comprehensive information
about each customer's banking history, they are able to provide customers with tailor- made
solutions.
Distribution management is the final component of the strategy. Banking services are
delivered through various channels (e.g., branch offices, ATMs, PC banking), and banks must
make long-term decisions about the best way to make services available to customers. These
decisions require knowledge about how customers currently use the bank, how they would prefer
to use the bank under different circumstances (e.g., if there were a fee for using a teller), and the
costs of the alternative distribution channels. With this information, FAC is able to optimize the
design of its distribution channels to meet customer needs at the lowest cost.
All four of these strategic components fit under the umbrella of the new brand identity for
the bank: About Life, About You. Each component is conspicuously focused on meeting client
needs, and together they create a powerful synergy.
Interestingly, although TCS was sold to the board of directors as an integrated strategy, it
was introduced in stages to the rest of the organization. According to the marketing director, the
reason was that it would have been overwhelming, given where the company was at the time.
The champions of TCS communicated only parts of the vision to those groups involved in
developing the separate components. Five separate projects, each one representing a major piece
of the TCS strategy (e.g., client information, brand identity), were pursued by five separate

groups concurrently. Each project was designed to have significant positive financial lift (i.e.,
impact) and to embody some of the basic changes in the way the bank would operate.
On September 20, 1997 after each component was successful, the complete Tailored
Client Solutions strategy was formally announced at the 1997 FAC Leadership Conference. The
Leadership Conference was also the kickoff for the Destination 2000 challenge to all FAC
employees, with the goal that by the year 2000, FAC would join the Sweet 16, a group of 16
banks that led the industry in terms of financial performance and valuation measures. A video
tape that presented Destination 2000 was distributed to all bank employees, who were
encouraged to join the effort and to adapt creatively to the upcoming changes. The theme was
reiterated at company-wide meetings, in internal correspondence, and through visible changes in
work practices.

The VISION Data Warehouse

Because every piece of the Tailored Client Solutions strategy demanded better and more
accessible information about the client than FAC, or most banks, ever had, senior management
realized that it needed information technology to support the strategy. The critical piece of IT-- a
data warehouse called VISION -- would contain integrated customer information, product
profitability information, and distribution revenues and costs (see Figure 2). Unfortunately, at
that time FAC had no substantive in- house experience in data warehousing. Because of the
strategic importance, FAC couldnt wait for its own people to come up to speed. Management
decided to rely on outside help from vendors and consultants, who would transfer knowledge to
internal IT personnel as the project progressed.

About life. About you.

Flexible Product Line
product profitability

Client Information
integrated client
information

VISION
Data Warehouse
Distribution
Management
distribution
revenues and costs

Consistent Service
client contact
history

Figure 2: VISION Data Suppo rts TCS Initiatives

Phased Development Approach

Preparing VISION to be the foundation for TCSs core components took a phased
approach, clear goals, the right people, and a careful coordination between IS, marketing,
finance, and external consultants. Connie White, the project manager, had to establish
expectations, define responsibilities, and coordinate the many interdependencies among the
project teams. The development of VISION included a multi-phased effort, with each phase
calculated to produce tangible business benefits while moving the overall project forward with
technical deliverables. The timeline and goals for the VISION phases are presented in Table 1.
The project manager explains that the phased approach enabled FAC to get the end result and
bring in the revenues as we worked on the project, as opposed to many companies who go out
and want to do everything at one time -- and arent successful."

Phase 1

Phase 2

Phase 3

Phase 4

Phase 5

1-2Q 1996

2-3Q 1996

Identify the
top revenue
producers

Identify the
least profitable
customers

2Q 1997
1Q 1998
Understand all
aspects of client
and product
profitability

2-4Q 1998

Business
Goals

3Q 1996
1Q 1997
Include
actual
transaction
and product
data in
profitability
formulas

Technical
Goals

Enhance the
existing
customer
information
system with
retail revenue

Enhance the
existing
customer
information
system with
direct
contribution
view for
consumers

Enhance
existing
customer
information
system with
net income
after capital
charges
(NIACC) for
consumers

Deploy the
warehouse
proof of
concept
(consumer)

Complete
production
testing of the
warehouse

Incorporate
profitability
understandings
in business
processes

Commercial
profitability
integration

Table 1: VISION Phases

Phase 1 was impleme nted to help managers understand the overall revenue picture. The
First Manhattan Consulting Group, a recognized leader in the financial services industry at
understanding profitability, provided benchmark data to FAC. Together with First Manhattan,
Jay Phillips, Senior Vice President for Decision Support, and his team developed revenue, cost,
and profit formulas to use with VISION data. These formulas and actual transaction data were
added to an existing customer information system within three months, and the project team
began to understand exactly which data belonged in the warehouse. In the next two phases,
additional products and transaction data were added to VISION, and the profit formulas were
enhanced. The data warehouse team also developed extraction and transformation processes,
while modeling the data to be stored in VISION.
10

Implementing a complete data warehouse was the major target of the final phases. FAC
relied on external vendors such as NCR to provide hardware, software, and methodological
support through much of the initial data warehouse development because FAC could not afford
to wait for its own employees to master the learning curve. While the company welcomed
outside help, it was careful to appoint an FAC employee to manage the process. Emery Hill,
Executive Vice President of Operations and Technology, explains that we just jumped into the
pool because they [our external support] said that they would jump in with us and keep us
afloat. By 1997, the data warehouse proof of concept was delivered, and the data warehouse
team spent 1998 validating the data, rolling out the warehouse, and helping develop applications
for finance and marketing.

Warehouse Architecture
FACs production data warehouse platform is the NCR 5150M configured with five SMP
nodes running the Teradata Relational Data Base System. This configuration provides 1.5 TB of
storage to FAC, and at the end of 1998 it supported 200 GB of raw data, which continues to
grow at a rate of 10 GB per month. The database holds 2 million accounts and information about
1.2 million households, and FAC plans to store up to 37 rolling months of history for analysis.
Figure 3 shows the warehouse's architecture.

11

EXTRACT, CLEAN, AGGREGATE, CONSOLIDATE, AND LOAD

FAC TERADATA WAREHOUSE

Shared Data Tables by Account

Replication Services
Logical Dependent
Data Marts

Oracle

Cognos Cubes

PowerPlay Impromptu PAR

Applications
and Access Tools

Direct SQL
Access

Business Users

Consulting Services and Education / Training

SALLIE MAE Harte Hanks

Logical / Physical Tables & Views

FAC
Enterprise Warehouse

IBM
MAINFRAME

Teradata Metadata

Data Extraction
& Transformation

VSAM, IMS

Data Access Layer = PAR and Cognos User Tools

Operational Data Sources

(Internal & External)

Teradata Relational Database Management System on WorldMark 5100

IT DW Team

Figure 3: The VISION Data Warehouse Architecture

Data Sources
Currently, the warehouse utilizes over 100 source files that are extracted from 26 legacy
applications. From the mainframe environment, VSAM and IMS files are FTPd to a file server,
where Informaticas Powermart applies business rules to transfo rm the data (e.g., making
account numbers consistent across banks). The resulting files are loaded into a warehouse
staging area where business users validate the data (e.g., financial data is validated within five
percent of the General Ledger). After the users examine the files and approve the data quality,
the data moves into Teradata base tables that are organized by account and by activity.
Concurrently, files from external data sources, such as student loans from Sallie Mae, geographic
and financial data from Dun & Bradstreet, and psychographic and demographic appends to data
from Harte-Hanks, are incorporated into the warehouse. The process of populating the data
warehouse takes approximately ten business days, although the team plans to reduce that time by
half in 1999. Table 2 describes the various kinds of data that exist in VISION after the
warehouse is populated.

12

Data
Client Behaviors

Client Buying
Patterns

Client Value
Propositions

Description
Products
Delivery Channels
Transactions
Segments
Attitudes
Expressed Needs

Profitability

Source
IBM mainframe
Sallie Mae
IBM mainframe
Dun and Bradstreet geographic
and demographic data
Harte-Hanks household data
IBM mainframe
Profitability algorithms

Data can be analyzed at any level of aggregation, from bank-wide or line of business
down to individual account or client relationship.

Table 2: VISION Data

Customer Centric
All of the data are organized around the client to provide a comprehensive understanding
of the clients demographic characteristics, the products used, transaction activities, interactions
with the bank, measures of the clients relationship to the bank, and psychographic insights about
client preferences and propensities; see Figure 4. The data can be analyzed in multiple ways,
including slicing and dicing using time, products, geographical regions, and market segments
as dimensions; planning marketing campaigns for specific products and markets; and detecting
which clients are at risk of leaving the bank.

13

Teller Transactions
ATM Transactions
PC Banking

Loans
Deposits
Investments

Transactions
Client
Interactions

Products

Marketing Campaigns
Customer Calls
Mailer Responses

Transaction
Client
Demographics
Name
Occupation
Head of Household

Measures

Psychographics

Balances
Fees
Profitability

Buying Preferences
Financial Segments
Lifestyle Groups

Figure 4: The Cus tomer Centric Data Model

Data Access
Marketing and finance analysts have direct access to warehouse data. For the marketing
area, the data warehouse team uses Cognos Corporations tools to create multi-dimensional
cubes of data that are stored on a local file server. Users can access these data cubes on the server
or download the information to their desktops. They then use the Cognos tool suite, including
PowerPlay, Impromptu, and Scenario, to manipulate and analyze the data.
The finance users access warehouse data from both a dependent data mart created in an
Oracle relational database management system and Cognos cubes. The mart supports a
profitability analysis and reporting tool called PAR (Profitability Analysis Reporting), which
provides users with predefined reports and lets them run ad hoc SQL queries against the mart.
14

Data Warehouse Team

Eighteen full- time FAC employees comprise the team that supports all of the data
warehousing initiatives, and Figure 5 shows how they are organized. Warehouse Services
manages the extraction, transformation, and load processes; and the Warehouse Development
team -- the gatekeepers of the warehouse -- is responsible for project management of new and
enhanced data feeds. Business Access Tool Support implements and supports the data access
tools; the Analytics group performs analytical studies and ad hoc analysis of warehouse data; and
the Data Mart team extracts, transforms, and loads data into the PAR system, a VISION data
mart.
Early in the project (1996-1997), most positions were filled by internal IT employees
because there were few data warehousing professionals available on the market. Data
warehousing expertise was provided by consultants and vendors. However, with the arrival of
Lance Mattingly (Applications Development Manager) in 1998, most positions were filled with
external data warehousing professionals, such as Theresa Leahy, the data warehouse production
and development manager. In fact, the strategy that has worked best for FAC is to bring in
people with a contract to hire. This approach brings contractors into FAC and provides the
bank the opportunity to take a test drive before offering them a permanent position.

15

Data Warehouse
Manager

Warehouse Services
Leader
3 Programmer/Analysts
Warehouse Development
Leader
Data Modeler
Business Analyst
Programmer/Analyst
Project Coordinator
Business Access Tool Support
Leader
2 Programmer/Analysts
Analytics Team
2 Analysts
Data Mart Team
Leader
2 Programmer/Analysts

Figure 5: Data Warehouse Organization Chart

VISION Users and Applications

The Users
The VISION data warehouse has both direct and indirect users. The direct users are the
20 marketing and 30 finance analysts who use warehouse data directly. Based on their analyses,
recommendations are made to management. There are also hundreds of indirect users who

16

receive and use reports that are generated from VISION data. Table 3 shows some of these users
and a sample of the information that they receive.
User
Senior Management
Corporate Banking Managers
Small Business Banking Managers
Retail Sales Managers and Sales Force
Product Managers
Distribution Managers
Marketing Managers
Asset and Liability Managers
Credit Managers

Information Provided by VISION

Profitability by line of business
Profitability by industry segment
Sales performance by area
Retention of high-value households
Profitability of products within product groups
Channel migration opportunities
Segment profitability
Product profiles
Analysis of deciles and risk taking

Table 3: Indirect Users and the Information Provided by VISION

Applications
Many applications that support the attraction, enhancement, and retention of customers
rely on the VISION data warehouse. The following applications, organized by the TCS strategy,
illustrate some of them.

Client Information
Customer Preferences and Profiles
Customer preferences are important to many banking decisions, and FAC has created
preference information in VISION using a technique called conjoint analysis. The company
selected a sample of over 3,000 customers and asked each one what he or she would do under
different circumstances. For example, would you use an ATM to make a deposit if the
transaction were free and the same transaction performed by a teller cost $.50? $1.00? Based on
the answers, a number of different "types" of customer were identified, and by matching

17

transaction and demographic patterns for each type of customer with those of the remaining
customers, FAC was able to extend its preference information to the entire customer population.
FAC also generates and utilizes market segmentation information. Data are purchased
from Claritas, which are then appended to FAC household records by Harte-Hanks. Using
financial segmentation software from Claritas, demographic and financial transaction data are
used to place clients in one of ten financial categories (e.g., wealth market, wealth preservers). A
similar process is used to place customers in one of 62 lifestyle groups (i.e., young influentials,
pools and patios). The placements are based on the assumption that "yo u are like your
neighbors."
Preference and profile information is used in many ways, including targeting marketing
efforts and in designing the best mix of distribution channels.

Customer Retention
VISION calculates the profitability of every customer so that appropriate actions can be
taken. High value customers are identified and targeted for retention programs, primarily
because this group is responsible for nearly all of the banks consumer profits. The mid-value
customers receive targeted messages about bank products that should be attractive to them (and
move them into profitable relationships). Low- value customers are migrated to more profitable
products and lower cost distribution channels (e.g., PC banking rather than using tellers).
Regardless of their current profitability, FACs goal is to retain all of its clients.
The "Top 20 List" is one example of a retention program based on VISION data. Using
the results from customer profitability analysis, each branch receives a weekly list of its top 20
clients. Service representatives then discuss ways to retain these customers and to expand their
use of the bank. A required approach is to telephone the top customers to thank them for their
18

business, and then to discuss their financial goals and how the bank can help the client realize
them.
Retaining clients and expanding their use of the banks products are important
components of how employees are evaluated and compensated. On Mondays, branch managers
commit to performance objectives for the week, and on Fridays their performance is evaluated.
For example, the bank may have a campaign to increase the number of savings accounts and
each branch makes a commitment for their contribution to the campaign. Information from
VISION is used to profile customers who are most likely to open a savings account. Employee
compensation is tied to how well the commitments are met.

Select Rewards
Like airlines' frequent flyer programs, Select Rewards is designed to increase customer
loyalty and increase customers' use of the company's products. VISION data is used to analyze
the profitability of the bank's products, determine the points that are awarded, and establish the
point levels for different awards. With Select Rewards, a customer earns points for longevity
with the bank and the breadth (number of products) and depth (average account balances) of the
banking relationship. Each month a customer receives a statement that reports the points earned
and a redemption certificate (see Figure 6). The points can be redeemed in a variety of ways -discounted banking services, items from Service Merchandise, gift certificates for meals at Red
Lobster and other restaurants, and free airline tickets.

19

Figure 6: A Select Rewards Statement and Points Redemption Certificate

20

Flexible Product Line

Product Profitability
It is critical for FAC to know the profitability of its various products. This is a complex
calculation that requires cost accounting equations and data from over 20 joins in VISION data
warehouse tables. Provided with product profitability information, financial analysts look at
products in a variety of ways. To illustrate, an analyst might begin by examining the profitability
of deposits, loans, and fee-based services (e.g., managing trust accounts). Drilling down into
loans, the analyst sees that commercial loans are the least profitable when returns are adjusted for
risk and capital charges, a finding contrary to conventional banking wisdom, wherein
commercial loans are thought to be very profitable. The analyst sees that floor plan loans are the
most profitable of the commercial loans. The analyst then rank orders all of the commercial loans
to see which ones are the most and least profitable. For the least profitable loans, the analyst
examines all of the loans, by the relationship managers who are responsible for them. The
analyst learns that a particular relationship manager is consistently making loans at rates that do
not cover their full costs. This information is marked for managements attention.

Seniors Account
Product profitability information is also used to redesign products. Like many banks,
FAC offers a free checking account to customers who are 55 and older. Working with VISION,
financial analysts discovered that the bank was making money from a few senior accounts, but
was experiencing a loss with many others. Further analysis revealed that the average size of the
checking account balance was the key difference between profitable and unprofitable accounts.

21

Based on profitability information and input from clients, a redesigned product was
introduced that created value for both FAC and its customers. Next, FAC marketed the account
to retain profitable customers while improving the profitability of other accounts. The 20 percent
of the customers whose account balances were normally high enough to make them profitable
received a letter and a personal call saying that the new minimum balance requirement was
unlikely to affect them. The others received a letter notifying them of the change and
encouraging them either to increase their average balance or to switch to an another type of
account. The results of this change were impressive: there was a significant increase in average
account balances, net income after capital charges, and service fees. Less than two percent of the
clients closed their accounts and many of them ultimately returned as profitable customers.

Next Most Likely Purchase

FAC is applying data mining techniques to VISION data in order to better understand and
predict customer behaviors. Applications that are either in place or under development include
accurate identification of buying trends, precise definition of market segments, optimization of
promotional programs, customer retention and acquisition, customer cross selling and upgrading,
and fraud detection. To illustrate, one data mining application calculates the likelihood that a
customer will adopt a specific product. Table 4 shows the steps that are used to transition
VISION data into this useful preference information for marketing campaigns.

22

1. Analysts determine that the target is to determine the likelihood that a household would use
Product A.
2. Examples of households with and without Product A are extracted from VISION. 5,000 of
each are selected.
3. Data are pulled on the 10,000 households and a file is created using a SQL query tool.
4. The file is transformed into a format readable by the data mining application and loaded into
the product.
5. The data is analyzed using several statistical techniques. Each data variable is examined for
importance, relevance, coverage, etc.
6. A first pass is made to select variables for the initial modeling effort. Between 40 and 60
variables are usually selected.
7. The data is divided into training, test, and validation sets.
8. A neural net makes its first attempt at predicting the target variable (use of Product A).
Several hundred iterations may be made before a reasonable model is created.
9. The initial model is reviewed and tested. Adjustments are made. A new model is created. The
process continues until no other improvements are seen.
10. The neural net is released in the form of a scoring algorithm, which is usually a multivariate
non- linear regression calculation.
11. Using a reporting tool, every household meeting the correct profile is downloaded from
VISION.
12. The data file is formatted and then run through the data mining application. Each household
is scored for its propensity to use Product A.
13. A household score file is created.
14. The household scores are appended to VISION where they are used to create target
marketing lists.

Table 4: Steps to Determining the Next Most Likely Purchase

Consistent Service
Contact Management System
The contact management system helps FAC develop a one-to-one relationship with its
customers by giving service representatives a clear picture of the entire banking relationship.
Drawing from VISION and other data sources, it provides personal information about the
customer, how profitable the customer is to the bank, how long the customer has been with the
bank, buying preferences (e.g., serious saver, price shopper), the products used, transaction
history, and the customers financial goals. All of this information resides on the service
23

representatives desktop computer and supports better informed, more personal interactions with
the customer. Because it is a bank-wide system, it also creates more consistent, seamless service.
A customer can go into any FAC branch and the service representative has access to the same
information about the client.
The service representatives and the contact management system work together to (1)
build a more personal relationship between customers and FAC, (2) provide opportunities to
learn more about customers needs and goals, and (3) market additional products to customers.
When service representatives meet with clients, they follow a structured interview process that
includes meet and greet questions (e.g., How are you? Did your daughter enjoy her trip to
Europe?), questions about financial needs (e.g., Are you still planning to buy a cottage at the lake
and might you want to finance it with a home equity loan?), and questions about investment
goals (e.g., Have you thought about opening an IRA to help with retirement plans?). The answers
to these questions are entered into the contact management system by the service representative
either during the conversation or later on. When the client is interested, the service representative
prepares a detailed financial plan that ties the clients goals to appropriate bank products. The
information about clients is also used in identifying which customers to include in direct
marketing campaigns.

Redesigning Bank Operations

FAC has redesigned its network of branch banks, how processes are performed in each
branch, and the jobs that people perfo rm in each branch with the goal to increase the time spent
with clients in order to better understand their needs and to enhance the opportunities for
marketing products. For example, the jobs of the customer service representatives (i.e., tellers)
were expanded to handle routine tasks such as changes of address in order to allow personal
24

bankers to have more time for sales-related activities. The personal banker job itself was divided
and is handled by four different types of people. The personal banker still exists to handle walkin traffic and sell additional products as time permits. Then there are the consumer relationship
managers who perform an expanded personal banker role. Because of their abilities, they are
entrusted with the most valuable relationships that have been identified by VISION and who
have more time for marketing efforts. Small business relationship managers do the same thing
for the most valuable "mom and pop" businesses. The final role is handled by investment
specialists who sell an expanded set of investment products. With these changes, it is estimated
that 60 percent of the time of the business relationship managers and investment specialists is
spent better understanding customers' needs and marketing new and additional products.

Distribution Management
Channel Distribution Costs
In order to make good decisions about distribution channels, it is important to know the
costs associated with the various channels, and VISION data makes this possible. The cost
calculations require information about how frequently the channels are used, the nature of the
transactions, and the costs of operating the channels. An interesting example of how this
information can be used comes from ATM usage. In general, ATM transactions cost less than a
teller. However, a recent study found that the cost of accepting a deposit through a teller was
$.77 as opposed to $1.79 with an ATM. This surprising result was because there are fixed costs
associated with processing ATM deposits (the deposits must be collected from an ATM, posted
at the branch, mailed to a central location, and inspected at the central location), and these costs
were being distributed across a relatively small number of deposit transactions. In fact, only four
percent of FACs ATM transactions were for deposits, which was below the industry average.
25

FACs response was a program to educate customers on using ATMs for deposits and to extend
the hours whereby clients would receive "same day" credit for making their deposits through an
ATM. For a specified period of time, customers who were Select Rewards members also
received bonus points for making ATM deposits.

Distribution Management System

The Distribution Management System (DMS) is a highly analytical application that helps
FAC plan dis tribution channels for various market segments in a way that is profitable to the
bank, yet still meets the needs and preferences of customers. Using inputs from VISION, such as
household profitability, how customers currently use the bank's products, customer preferences,
segments and channel costs, and external data that match potential new customers to the
customer segments, DMS calculates the best way to distribute products to customers. Consider
an example of the use of DMS. In a particular marketing area, FAC was operating a main bank
(a hub), three branch banks (spokes), and various ATMs. Using DMS, it was decided that closing
one branch and increasing the number of ATMs would increase the bank's profits while better
meeting client's needs.

TCS and Data Warehousing Impacts

FAC finds that TCS, supported by the data warehouse, has changed the mindsets of
employees, which in turn has created tangible benefits at the application and organizational
levels. Further, FAC's senior management believes that the company is realizing strategic
benefits thanks to the organizational transformation and the way in which the bank is now
perceived within the industry.

26

A Change in Mindset
The organizational transformation at First American Corporation included major shifts in
the mindsets of its staff, primarily because of the information available through the data
warehouse. Jay Phillips, head of finance and marketing applications, says "We reengineered all
of finance, changed all of the systems, and made a huge shift in marketing. You see a real
change in how the lines of business do business; you see the branches really going through a
redesign process." Finance has moved from being bean counters to aggressively working to
find better ways of creating revenue. Good customers are now determined by the profitability
of their overall relationship with FAC. Marketing has moved from a suckers and balloons
mentality to predicting customer actions through careful analysis, and using this information to
promote profitability.
Across the bank, units that previously took a passive approach to business innovation
now see themselves as responsible for creatively improving the bottom line. For example,
accounts payable took an existing internally used purchasing card application and turned it into a
product that could be sold to smaller corporate customers. (A purchasing card is basically a
credit card used to allow particular projects or groups to conveniently purchase needed items
while automatically recording the expenses against their budgets). This was quite a shift in
mindset for accounts payable, which had never before been involved with revenue creation.
Brian Cooper adds: It is gratifying to see that VISION has transformed the organization. People
think differently now because of it, and new employees dont even know the way we used to
think.
This level of change has not been easy or comfortable. Throughout the organization,
those who could adapt to frequent changes and who could take the initiative to enhance

27

performance prospered, while those who could not, left. Some areas experienced 100 percent
turnover in one year, and many others experienced 25-30 percent turnover over three years.
Those who stayed, however, are part of exciting changes.

Application-Level Benefits
In order to assess the success and impact of TCS, senior management initiated project
tracking. Table 5 describes selected applications, the users of the applications, and the resulting
benefits. For example, the Seniors Account has seen an improvement in the risk-adjusted return
on equity from less than 25 percent to over 50 percent while maintaining deposit balances.

28

Application
Client Information
Customer Attrition and
Retention
Customer Complaints

Select Rewards

Flexible Product Line

Seniors Account

Repackaging of budget
checking, student
checking, 55 and better,
and small business
checking
Re-price certificates of
deposit, express
mortgages, NOW
accounts, FAIR
accounts, and savings
accounts
Next Most Likely
Purchase
Consistent Service
Contact Management

Users

Impact

Segment managers
Financial analysts
Marketing analysts
Service
representatives

Revenues grew by over 15% from high

value client.
Improvement in NSF fee collection efforts
increased fee income by over $1.3 million
in 1997 and 1998.
Over 43,000 customers are Select Rewards
members, accounting for over $1.4 billion
in loans, deposits, and investments.

Marketing analysts
Segment managers

Marketing analysts

Improved risk-adjusted returns on equity

(RAROC) from less than 25% to over 50%
while maintaining deposit balances.
Enhanced revenues in excess of $1.8
million in 1997 and $3.2 million in 1998.

Financial analysts

Financial analysts

$1.7 million additional revenue in 1997 and

1998.

Service
representatives

Improved sales effectiveness.

Sales representatives

Assisted in improving retention of high

value clients by 1%; worth $4 million in
revenue.
Assisted in improving retention of high
value clients by 1%

Redesigning Bank
Operations

Sales representatives
Service
representatives
Distribution Management
Distribution
Financial analysts
Management System
Distribution
managers

In 1998, 22 higher cost hub locations

were replaced by 30 lower cost spoke
sites. Over 20% return on investment.

Table 5: The Benefits from TCS Applications

29

Organization-Level Benefits
FAC's Tailored Client Solutions strategy, powered by the VISION data warehouse, has
fundamentally changed how the bank is managed and generated quantifiable financial returns. It
has allowed FAC to emerge as a profitable, innovative leader in the financial services industry,
as the highlights in Table 6 show.

Financial Indicators
Return on assets
Return on earnings
Earnings per share
Productivity ratio
Average assets (in billions)
Average core deposits (in billions)
Stock price (as of)

1996
1.33%
15.20%
$1.98
58.98%
$16.6
$11.6
$29.56
(1/17/97)

1997
1.40%
15.91%
$2.18
57.93%
$17.9
$12.4
$45.875
(1/15/98)

1998
1.55%
18.07%
$2.62
53.44%
$19.3
$12.6
$41.3125
(1/21/99)

Table 6: Financial Performance Measures

Comparison with the Sweet 16

FAC's goal is to join the Sweet 16 financial services organizations that are leading the
industry in financial performance and valuation measures. Table 7 shows the progress that has
been made since the beginning of 1996 until the end of 1998 on measures that are being tracked.
If these trends continue, FAC has a good chance of meeting its goal of achieving Sweet 16 status
by the end of 2000.

30

Performance
Measures

1996

Return on Assets
Return on Equity
Earnings per
Share Growth
Productivity
(lower is better)

1997

1998

FAC
1.33%
15.20%
11.9%

Sweet 16
1.55%
19.10%
12.1%

FAC
1.40%
15.91%
10.1%

Sweet 16
1.67%
20.30%
14.3%

FAC
1.55%
18.07%
18.07%

Sweet 16
1.66%
19.20%
15.1%

58.98%

53.2%

57.93%

52.2%

53.44%

55.8%

Table 7: Performance Comparisons with the Sweet 16

Industry Benefits
TCS also has affected favorably how other banks perceive FAC. The CEO of Deposit
Guaranty (which FAC acquired in 1998) said that FAC was attractive because he wanted to be
part of "a financial institution of the future and not a bank of the past." He had learned of FAC's
data warehousing initiatives and had not encountered banks of similar or even larger size with
the same capabilities.

Lessons Learned
The data warehousing experiences at FAC provide valuable lessons for organizations that
either have implemented or are planning a data warehouse.

1. Have a strategic vision for where you want the business to go and understand how data
warehousing can help you get there. At FAC, senior management knew that survival depended
on profound changes in how the bank was run. It had to pick its markets carefully, know the
profitability of its various products, understand the value of customers and their preferences, and

31

establish a plan to achieve this goal. A key component of this plan was to create a data
warehouse to generate the necessary information.

2. Both organizational and technological changes are necessary. Obviously, the introduction
of data warehousing involves technological change. But organizational changes must occur too,
and these can be even more difficult than technological change. Brian Cooper explains this
concept by referring to a "formula" that he learned in graduate school: OO + NT = COO. It says
that the old organization (in its thinking), plus new technology, equals a costly old organization.
It is important to change the organization and the technology at the same time; technological
change alone is not the solution.
At FAC, senior management recognized that the data warehouse was a key enabler of its
corporate strategy but also knew that the strategy called for a completely new approach to
banking. Information created from the warehouse helped FAC analyze their business and
recognize that certain long-standing practices and products were unprofitable, but managers had
to think about this before they could respond accordingly. To support this major shift, a
carefully conceived change management program was established before business processes and
accompanying jobs were redesigned.

3. Have the right sponsorship. It is well understood that a key to project success is good
sponsorship. For data warehousing, however, a more granular understanding of sponsorship is
needed. The "right" sponsorship for warehousing depends on the business drivers behind the
project, the cost and scope of the project, and the amount of organizational change required. FAC
illustrates a situation where the warehouse is a key enabler of corporate strategy, its cost and

32

scope are great, as is the amount of organizational change involved. For warehousing initiatives
of this importance and size, it is important to have sponsorship at the highest organizational
levels, from the business units, and from IS. This broad-based management support is critical. At
the other extreme, where the warehouse (quite likely a mart) is created to serve a limited, specific
need (e.g., production quality), sponsorship by a senior business unit manager may be sufficient.

4. Plan for "quick hit" success, and repeated successes. In today's business environment,
companies are very financial-results oriented. In order to maintain and enhance support for any
long- term endeavor, it is important to generate "bottom line" results as soon as possible. Each of
the initial projects under the Tailored Client Solutions strategy was designed to generate visible
financial results. The VISION data warehouse project was divided into five phases, each of
which provided immediate, tangible benefits. Phase I delivered information about the highest
revenue customers. Each additional phase delivered actionable information for the bank. In
addition, money- making programs were implemented in sales and marketing based on
information created through warehouse data analyses.

5. Acquire the necessary skills. When organizations change, it may result in a quantum leap in
the information skills and technology required. If these capabilities are unavailable in- house, the
organization may need to go elsewhere for the help it needs. A data warehousing project will not
succeed without people who are technically skilled and have suitable experience.
FAC brought in vendors, consultants, and new hires with data warehousing experience.
NCR provided hardware, software, and consultants for the project; First Manhattan Consulting
Group helped to analyze the profitability of the bank's different products; and Brian Cooper and

33

his team coordinated the work of the external partners and provided the vision and experience in
how to use the warehouse to support the customer relationship management strategy.

6. Prepare for a permanent production environment. Many companies' initial data

warehousing efforts are similar to a critical R&D project on a crash schedule. There is an
important business need for the project, which has to be satisfied immediately, so shortcuts may
be necessary. As a result, while the project may be completed on time, everything necessary for a
permanent production environment is seldom in place.
While such a crash program may be strategically necessary for many warehousing
efforts, it is critically important to prepare a stable production environment as soon as possible.
This includes having a scalable hardware and software solution, the requisite technical skills in
place, appropriate committees to provide direction, and formalized operating disciplines. This is
particularly critical if significant consultant expertise has been used in implementing the project.

7. Hire analysts who understand the business and IT. The use of a data warehouse usually
requires ana lysts with different, more advanced skill sets. It is obvious that they must know the
business, but the bigger challenge is finding analysts who are also comfortable with data models,
databases, SQL, managed query environments (e.g., Cognos), and data mining. These people are
business/IT hybrids. Some of a company's analysts may be retrained to work in the new
warehouse environment but others may not have the aptitude or inclination for the new kind of
analyst work. At FAC, many of the existing marketing and finance analysts either moved to
other positions or left the company, and new analysts were brought in to fill the vacancies.

34

8. Anticipate potential technical staff turnover. The demand for experienced data
warehousing personnel is high, and the data warehousing learning curve is steep. Expect that
your people will receive "off the scale" offers from other companies, including consulting firms.
The loss of a key person at a critical time can delay a data warehousing initiative for months.
Even consulting firms and vendors may have a difficult time responding to an immediate request
for help because they too may be short staffed. The best strategy is to have attractive retention
programs already in place (e.g., bonuses for completion of the project, commitments for
additional training) for key people.

9. The data warehousing skills that are needed change over time. There are three phases in
the warehouse life cycle: Design It, Build It, and Exploit It. Each of these phases requires a
different set of talents, perspectives, experiences, and skills. Those involved in the Design It
phase need a good data architecture background, good logical data modeling skills, and a strong
understanding of the business. Those involved in the Build It phase need strong project
management, process management, and technical (programming) skills. Those in Exploit It
phase need strong technical backgrounds, excellent analytical skills, and, most importantly, the
ability to translate the analysis into an actionable marketing plan.

Conclusion
FAC has undergone a significant organization transformation in a short time, including a
top-to-bottom shift in the ways of thinking about the banking business, a new customer-centric
focus, the redesign of its products and distribution channels, and changed business processes -all supported by a major investment in a data warehouse. Together, these changes have resulted
35

in a significant increase in performance measures, including profitability. To be sure, the

VISION data warehousing effort is not the only factor in this transformation. The transformation
effort originated in top managements crafting of a survival strategy, and the data warehousing
effort was one of many interlinked components necessary for success.
However, we can say that the transformation could not have happened without the
VISION data warehouse. At a very basic level, the FAC transformation is about making better
use of information to both meet customer needs and increase profitability. The many examples
and applications described in this paper demonstrate the value that can be gained from cleverly
exploiting the information available in the bank. Without the discipline imposed by the
warehousing effort and the technology to combine the data into a usable resource, many of the
benefits described would not have been possible.
Since the 1970s, proponents of information technology have promised that management
information systems would put all the right information in the right hands at the right time, and
radically change the way companies are managed. Unfortunately, for the most part, reality has
fallen far short of the grand promises. Today at FAC, however, the bank is realizing much of that
promised goal.
While not every business will aspire to the radical change seen at FAC, the lessons
learned at FAC are important for many companies. As IT capabilities increase and costs drop,
and as increased competition drives more firms to rethink their strategies, the radical change
picture experienced at FAC will become more common.
Even organizations that do not desire such dramatic transformations know that many IS
projects will involve considerable organization change in order to accommodate the new, highly
competitive marketplace. After all, the biggest benefits come not from the technology, but from

36

better and more effective ways of carrying out the mission of the organization, supported by the
technology. For many firms, data warehousing provides a quantum leap in the availability and
quality of information, and may be a critical component of important organizational changes that
are designed to improve bottom- line performance. For these firms as well, the lessons learned at
FAC should be very useful.

37