Data Warehousing ROI: Justifying and Assessing a Data Warehouse

Data warehouses require a sizeable commitment of organizational resources. As a result, there is considerable interest in how they are initially justified and later assessed. Data warehousing costs
are relatively easy to estimate, but the benefits are more difficult to evaluate. Survey data shows that most companies quantify the costs of data warehousing but not the benefits. In order to
better understand how companies justify and assess data warehousing investments, eight case studies were conducted. Drawing on the data warehousing literature, survey data, theory, and the case
studies, seven propositions are presented.

Building a data warehouse is a complex, expensive, and time-consuming task. Depending on a warehouse’s scope, it may require a seven-digit expenditure and take months to initially develop and
years to become fully enterprisewide in its scope (Watson and Haley, 1997). There is no assurance that it will be successful, because many data warehousing projects are over budget, behind
schedule, fail to live up to expectations, or “belly up” completely (Watson, et al., 1999). Not surprisingly, then, there is considerable interest in justifying and assessing data warehousing
projects. This interest exists with companies considering getting into (or continuing to invest in) data warehousing as well as consultants and vendors who want to sell their products and services.

ROI analysis is a useful decision-making tool that helps ensure investments are aligned with business strategy and are an effective use of organizational resources. However, calculating ROI for a
data warehouse is a challenging task (Whittemore, 2003). Costs can usually be estimated, but some of the potential benefits are difficult to assess. For example, the cost of additional disk storage
is easy to estimate, but the returns from improvements in decision making may resist quantification. While net present value and internal rate of return are the measures usually associated with ROI
in academia, in practice, a broader set of metrics are included, including payback period and return on investment over some specific period of time. In some cases, financial measures are not used
with data warehouses (McWilliams, 1996).

This article explores the approval process and post-implementation review for data warehouses. First, we discuss the costs, benefits, and ROI measures associated with data warehousing. Next, we
present survey data that helps to understand industry practices. We conducted telephone interviews with eight data warehousing managers about their companies’ data warehouse to understand the
motivations for building it, what was developed, and how it was initially approved and later assessed. We conclude by integrating the case studies, the data warehousing literature, theory, and
survey data to provide insights in the form of propositions about the approval and assessment processes for data warehousing projects.

Hardware · Disk storage
· Processor
· Network
Software · Extraction, transformation, and loading software
· Database management software
· Metadata software
· End-user data access tools
Personnel · IT staff (e.g., database administrators, data modelers)
· Business and end-user personnel
· Consultants
Table 1. The Costs of Data Warehousing


Costs, Benefits, and ROI
The costs of data warehousing are relatively easy to estimate. Like most IT projects, they include the associated hardware, software, and personnel costs (both in-house and consultants). Table 1
provides a list of typical cost categories and cost examples in each category.

There are many possible benefits from data warehousing, such as:

  • Tangible or intangible
  • Revenue enhancing or cost saving
  • Operational, informational, or strategic
  • Data mart consolidation savings, time savings, more and better information, improved decision making, business process improvements, or support for strategic business objectives.

Examples in the last of these benefit categories are given in Table 2.

Data mart consolidation · Reduction of multiple decision support platforms
· Hardware and software cost savings
· Operational efficiencies
Time savings · Less time spent by IT personnel downloading data for users
· Less time spent by IT personnel writing queries for users
· Less time spent locating data
· Less time spent by analysts responding to requests for information
More and better information · Having information that did not previously exist
· Users’ ability to analyze data in new ways
· Ability to thinkofthe business in new ways
Personnel savings · Redeployment of IT personnel
· Faster company growth without adding personnel
· Redeployment of operational personnel to higher-value-producing activities
Improved decision making · Decisions based on facts rather than intuition
· Faster decision making
· Ability to analyzealternatives better
· Ability to identifyand act on problems better
Business process improvement · Redesign of jobs
· Procurement savings
· Shorter business cycles
· Ability to identifyand correct problems with business processes
Support for strategic business objectives · Faster response to changing market conditions
· Increased market share
· Improved speed to market with new products
· Supply chain integration
Table 2. The Benefits of Data Warehousing

It is not feasible to estimate all of the possible benefits from data warehousing. Instead, organizations focus their attention on the data warehousing “sweet spots,” where the greatest
business value and ROI can be found.

A common way to assess the tangible (e.g., quantifiable) benefits is to compare the current (i.e., baseline) state to future state (which assumes that the warehouse is in place) (Whittemore, 2003).
For example, a key warehouse application might help reduce customer attrition. To quantify the benefits, the current attrition percentage is compared to the reduced attrition percentage, and this
decrease in attrition is translated into profits.

In general, the greatest potential benefits come from increasing revenues rather than decreasing costs. In difficult economic times, however, companies often focus their attention on reducing
costs. Therefore, it is not surprising that vendors and companies have recently focused on data mart consolidation, cost-cutting measure.

A warehouse typically generates many intangible or nonquantifiable benefits (i.e., “soft” ROI). For example, the warehouse may provide better data integration, improve customer service,
or enhance a company’s image. The team working on the warehouse proposal may decide that it is better to list some benefits as intangible rather than to place a dollar value on them and have
the accuracy and credibility of all of the estimates brought into question.

There is no uniform measure or process for calculating data warehousing ROI. The most common measures are net present value, internal rate of return, average rate of return (over a specific time
horizon), and payback period (Whittemore, 2003).

International Data Corporation (IDC) conducted the first and most renowned study of data warehousing ROI. Based on 62 organizations, IDC found an average ROI of 401 percent over a three-year time
period (IDC Report, 1996). This average return excluded failed projects as well as exceptional performers (both good and bad).

The large number of vendors and consultants that offer ROI products and services reflects the importance of ROI to decisions about data warehousing (and other IT investments). In many cases, the
vendors and consultants want to help their prospective clients sell their warehousing projects to top management. For example, Teradata has industry-specific Business Impact Models that calculate
the ROI for proposed data warehouses.

A factor that complicates computing data warehousing ROI is that the warehouse is not usually directly linked to results. It is an investment in decision support infrastructure. Someone or some
process must take action based on the information obtained from the warehouse. As a result, warehousing has second- or third-tier effects—it leads to processes that, in turn, lead to
decreased costs or increased revenues (Whiting, 1999). It is difficult to assign value to the warehouse in isolation from the value created by other people and processes.

Using ROI has its critics. For example, Bill Inmon (2000) argues that, in many cases, the returns from data warehousing cannot possibly be known until the warehouse is completed and in use.
Illustrating this point, Inmon points out that “miners” (i.e., users who search through warehouse data looking for “golden nuggets”) may or may not find relationships and
insights that are valuable to the organization. The benefits resulting from their mining can only be known after the fact and can vary considerably.

Approval Process Percentage
Development of a prototype 42.9
Formal proposal with quantified (dollars) costsand discussions of how the warehouse wouldbe used 39.0
Mandate from senior management 37.1
Steering committee recommendation 21.0
Formal proposal with quantified (dollars) bene-fits and costs 20.0
Other 16.2
Table 3. Approval Processes for Data Warehouses (Watson, et al., 2001)

Survey Findings
One of the authors collected survey data from 106 organizations about how their data warehouses were initially approved and were assessed later after implementation (Watson, et al., 2001). The
survey findings are presented in Table 3 and provide insights about the approval process. The total of percentages does not equal 100 percent because the categories are not mutually exclusive.

Because of the cost, time, and risk involved, developing a prototype is common. The prototype is typically a data mart that maintains only a single or a few subject areas, supports a limited set of
applications, and has users from a single or a few departments (e.g., marketing). If the mart is successful, it may serve as a “proof of concept” for building an enterprise data

Before building a warehouse, most organizations require an approved proposal. The costs of building the warehouse —hardware, software, in-house personnel, and consultants —are
relatively easy to estimate. It is more difficult to quantify the future benefits. Consequently, more organizations have an understanding of the potential benefits rather than estimating the
specific dollar amounts that are required to calculate ROI. This finding is supported by a 1997 Forrester research study that found that only 16 of 50 companies calculated an ROI before building a
data warehouse (Whiting, 1999).

Senior management may also provide a mandate for building a data warehouse. In these cases, it may be that the warehouse is needed to support a specific business strategy (e.g., customer
relationship management) or to address general (e.g., improve information reporting) or specific (e.g., supply chain) information needs.

Type of Review Percentage
Agreement that the warehouse is meeting itsobjectives 65.5
Intuitive benefits and quantified (dollar) costs 48.3
Quantified (dollars) and costs 36.2
External assessment 22.4
Other 15.5
Table 4. Post-Implementation Review of Data Warehouses (Watson,et al., 2001)

Table 4 shows how data warehouses are evaluated after implementation. The most common assessment is to determine if the warehouse is meeting the objectives for which it was built. If it is, the
success of the warehouse may be so obvious that management decides that there is no need to spend the time and money on a formal review.

While formal reviews are common, it is still more common to intuitively assess the benefits than to place a dollar amount on them. The percentage of firms that quantify the benefits after the
warehouse is in place is greater than in the pre-development phase. This is not surprising since firms have actual (instead of projected) experiences on which to base their calculations.

Data Warehousing ROI Case Studies
In order to gain richer insights into how organizations use ROI in approving and assessing data warehouses, the authors conducted telephone interviews with data warehouse managers and professionals
in eight organizations.

In the following case studies, we briefly describe the companies and people that participated in the case studies. We also discuss the business drivers for the warehouse, the approval process, how
additions and enhancements to the warehouse are approved, and the ROI measures that are employed.

Boise Cascade Corporation
Boise Cascade is an international distributor of office supplies and paper, and an integrated manufacturer and distributor of paper, packaging, and building materials. It owns or manages 2.4
million acres of timberlands in the U.S. and its operations extend throughout North America, Australia, New Zealand, and Brazil.

Mike Lowe is a database administrator and has been involved with Boise’s data warehousing initiative since it began. Management in the business units sponsored the warehouse in order to
improve management reporting and data analysis. Existing reporting systems took too long and did not provide the big picture perspective that was needed. Also, users wanted to be able to do ad hoc
data analysis.

The warehouse was developed incrementally, one subject area at a time. The funding for the warehouse came from the business units and was justified by cost savings resulting from the elimination of
existing reporting and analysis systems. As the expectations for the warehouse were met, support grew, and new needs surfaced. Today, the custom-tailored reports and graphs generated from the
warehouse are a “must have” item for many managers.

The business units pay for enhancements to the warehouse. Rather than calculating an ROI for the warehouse, business units justify warehousing-related expenditures in their budgeting processes.
Because they pay for them, the business units do their own assessments of their value. A chargeback system is used to prorate the fixed costs of warehouse maintenance.

First American Corporation
First American Corporation (FAC) was a medium-sized bank headquartered in Nashville, TN until its acquisition by AmSouth in 1999. It served the southeast with operations in Tennessee, Kentucky,
Virginia, Mississippi, Arkansas, and Louisiana.

The roots for data warehousing at FAC go back to 1991, when the bank was in financial trouble and a new management team was brought in. The first step was to “stop the bleeding” by
cutting costs, but management knew that a new business strategy was required in order to survive in the increasingly competitive financial services industry. 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 large national and international banks dominated this market. Ultimately, FAC management
decided to compete on the basis of knowing its customers better than the competition did.

A data warehouse was needed to implement its customer intimacy (i.e., customer relationship management) strategy. Connie White was the project manager for the project and says ROI calculations were
not made for the warehouse. Because the warehouse was a requirement for executing the business strategy, the warehouse had to be built. Without it, FAC would not survive.

After FAC turned its fortunes around, more traditional financial measures were used to assess enhancements to the warehouse, but the focus remained on the use of the warehouse to support the
business strategy. By the time that AmSouth acquired FAC, the business strategy and warehouse were instrumental in transforming FAC into a highly successful financial services institution (Cooper,
et al., 2000).

Harlequin Enterprises Limited
Harlequin Enterprises is the world’s largest publisher of series romance fiction. From its headquarters in Don Mills (a suburb of Toronto, ON), Harlequin sells approximately 180 million
paperbacks a year in 23 languages in more than 100 countries. Its parent company is Torstar, publisher of Canada’s largest newspaper, The Toronto Star.

Gene Dankewich is Harlequin’s Director of Global Planning and Reporting Systems. He was at Harlequin in 1995 and was in the process of replacing existing financial systems when he realized
that he needed to broaden his scope to consider a more global solution for information management. This realization caused Dankewich to consider data warehousing. In 1996, Andersen Consulting did a
strategic assessment of the firm’s systems and recommended building a data warehouse.

Harlequin did not develop an enterprise data warehouse; rather, it built data marts for direct marketing and retail groups. The decision to build the marts allowed the project to “fly
under” the capital expenditure radar by doing the project in increments that cost less than $100,000. Dankewich also found support for building the marts in the recommendations made by
Andersen. His positive relationships with the CEO and other key executives were also helpful; by providing scenarios of how the warehouse would be used, he gained the support of key decision makers
in the company. In addition, the industry culture does not always require that ROI calculations drive investment decisions. A logical business argument is sometimes more persuasive than a financial

The data marts were very successful, especially for direct marketing. They are now ingrained in how Harlequin does business. The company will not conduct a postimplementation ROI evaluation because
the benefits from the marts are so obvious. Based on the initial success, there are plans to expand the marts into an enterprise data warehouse.

Harrah’s Entertainment
Harrah’s is a leader in the gaming industry. It operates 26 casinos in 13 states under the Harrah’s, Harveys, Rio, and Showboat brand names. It is widely recognized for its use of data
warehousing and customer relationship management (CRM) to build brand loyalty.

In 1993, legislation was passed that allowed gambling on Indian reservations and riverboats. Seizing the opportunity, Harrah’s quickly expanded into these new markets by building new
properties and acquiring casinos. As management thought about how it could create the greatest value for its shareholders, the company decided to take a brand approach—the various casinos
would operate in an integrated manner rather than as separate properties.

Critical to their strategy was the need to understand and manage relationships with their customers. In 1994, work began on Harrah’s Winners Information Network WINet). It collects customer
data from various source systems, integrates the data in a customer-centric data warehouse (which is used to identify market segments and customer profiles), creates offers for customers to visit
Harrah’s casinos, and makes the data available for operational and other analytical purposes.

At Harrah’s, all proposed projects go to the Corporate Finance Committee and must include net discounted present value and internal rate of return estimates—a standard business practice
at Harrah’s.

Harrah’s continues to expand and find new uses for its data warehouse (e.g., linking slot machine usage to customer segments). The business units are the primary drivers of new application
development and are responsible for preparing proposals for the Corporate Finance Committee. The required ROI calculations include the estimated incremental benefits (e.g., revenues and cost
savings) and costs associated with the project, including any incremental costs of expanding the data warehouse.

Large Retailer
Large retailer (the company requested anonymity) is a large North American department store retailer. The data warehousing manager was working at Large Retailer in the late 1990s when the data
warehousing project began. The CEO sponsoring the warehouse wanted to use the warehouse for inventory control and assortment planning i.e., which merchandise to place in which store). It was also
felt that an enterprise data warehouse would provide information to better support decision making throughout the organization.

Even though the CEO supported the warehouse, a detailed financial analysis was required. As the data warehousing manager said, “In the retail industry, everything has to be
quantifiable.” In order to quantify the benefits, attention turned to potential improvements in inventory levels and profitability. A two-year pilot study was conducted to prove that
improvements would be a reality. The pilot allowed the business users to demonstrate to the business executives that the benefits would exceed the investment.

The warehouse is subjected to ongoing post-implementation review. All data additions and environment expansions have to be self-funding. As a result, the warehouse team and its business users are
constantly looking for quantifiable examples of how the warehouse has benefited users and the company.

Any time a new subject area or application for the warehouse is proposed, it is carefully reviewed. The proposal must describe the information that will be created, who will use it, the decisions
and actions that will be taken, and the quantifiable benefits. All data warehouse enhancements must be business-unit driven. The data warehouse team typically leads the effort to formalize the
business cases.

While the approval and assessment processes used with the warehouse require quantifiable ROI analyses, they also consider how the warehouse supports the firm’s strategy and any intangible
benefits that might result.

Imation Corporation
Imation, headquartered in Oakdale, MN, is a leading developer and manufacturer of removable data storage products sold in more than 60 countries.
Imation’s offerings include magnetic and optical products that are a key ingredient in much of the world’s technology infrastructure.

Pat Redding is the manager of the data warehouse group and described the two major data warehousing initiatives at Imation. The first project, sponsored by the sales and marketing department, was
to eliminate reporting deficiencies caused by the migration from custom ERP and decisionsupport systems to Oracle applications. Because Imation had experienced previous decision support success,
the company understood the need for data warehousing. The decision to develop a sales and marketing data warehouse was considered to be a “no brainer” because the sales and marketing
information was so obviously needed. Consequently, no ROI calculations were made.

The second data warehousing initiative included the addition of almost all of the financial and logistical modules from Imation’s ERP system into the data warehouse. The scope of this
warehouse was large and by necessity was performed in phases. It eventually included more than 60 subject areas. The project went through Imation’s usual processes for large capital
expenditure approval, including ROI calculations, approval by a review board, and support from the business units. In order to estimate the benefits, Redding went to the various business units that
would be using the warehouse and asked them to identify and commit to the benefits that they would derive from the warehouse. These business units were the co-sponsors of the warehouse.

The cost of not building a warehouse was a major selling point. If the warehouse was not built, there would have been costs associated with storing data that needed to be purged from the ERP
system. In addition, the ERP infrastructure could not support the additional burden of ad hoc querying and reporting without significant upgrades.

Currently, there have been no calls for a post implementation analysis of the data warehouse. The feedback from the business units is that the benefits have exceeded expectations. For example,
there have been greater-thanexpected reductions in personnel in the business units. Also, as the data warehouse has been used over time, it is being used for applications that were not identified
at the time it was proposed, and these applications are delivering great value. Since the data warehouse has been in place, numerous projects have been initiated to better measure and manage key
performance indicators and improve business processes. These projects would be much more difficult and expensive to accomplish without the current data warehouse.

The University of Texas M. D. Anderson Cancer Center
The University of Texas M. D. Anderson Cancer Center in Houston, TX is widely recognized as one of the nation’s
leading comprehensive cancer care centers, with a worldwide reputation for cancer care, research, education, and prevention. As a component of The University of Texas System, M. D. Anderson is a
prominent teaching hospital and research center for the state of Texas.

Leslie Kian was hired in 1993 and is currently the Director of the Medical Informatics Department. He was internally recruited in 1997 to the department to advance M. D. Anderson’s data
warehousing efforts as well as to support the continued needs of collecting disease information.

As M. D. Anderson struggled with the changes in the healthcare industry in the early 1990s, information needs grew rapidly. While M. D. Anderson’s executive leadership required extensive
analysis and reporting of patient data, satisfying the ever-increasing demand for information was difficult and time-consuming due to data sources being scattered across many disparate systems.

In 1993, M. D. Anderson’s Chief Information Officer envisioned bringing all the data together in a single repository to meet the information needs of the leadership team. Implementation of
this initial data repository solution began with data from primary hospital information systems, and later integrated data from other legacy systems. This single information repository led to many
projects that supported both operational and research needs.

The decision to advance the repository to a patientcentric data warehouse was supported by several key executives. Growing information needs that were difficult to address with the existing
repository provided the justification for creating a more classically designed warehouse. Although a formal return on investment analysis was not performed, the financial benefits realized from the
older repository alone justified the advancement. (ROI calculations have since been performed as part of M. D. Anderson’s standard system development methodology.)

At the time of the new warehouse development, projects that cost $500,000 or more had to go through a formalized internal review and oversight process as well as project approval by state
government regulatory bodies. This requirement is still in place today, with the financial thresholds increasing over time. M. D. Anderson’s Information Systems (IS) Council reviews all major
IT projects; an executive level IS Steering Committee approves funding. Among several criteria for project review (both internally and at the state level), ROI determination is a requirement.

The initial development of the warehouse was guided by a highly engaged steering committee consisting of representatives from major stakeholders representing many organizations, including
functional patient care. After the addition of several data sources, executive management directly guided the data warehouse development efforts. The warehouse is now considered a key component of
M. D. Anderson’s information infrastructure, with maintenance costs included in the budget as an operational expense. Like most warehouses, M. D. Anderson’s patient-centric warehouse
continues to evolve.

To date, most of the customers of the warehouse are highly skilled data miners along with members of the executive management team. The data mining staff currently provides the primary delivery
vehicle of information to the organization. To expand the access delivery of information directly to operational and research staff, the implementation of business intelligence tools is underway.

A steering committee has been established to guide the future development of the warehouse, including the review of any proposals for major enhancements. Requests for enhancements originate from
warehouse end users, and the steering committee is highly representative of departments using the warehouse so the committee is familiar with the warehouse and enhancement requests. The committee
focuses its attention, as it did in the past, on the value to the institution and the timeline for the proposed work to balance added value with a quick return on benefits.

Owens & Minor
Owens & Minor (OM) is the nation’s leading distributor of name-brand medical and surgical supplies. It purchases products from more than 1,400
suppliers, stores them in 50 distribution centers, and sells them to more than 4,000 hospitals, integrated healthcare systems, and group purchasing organizations.

Don Stoller, Director of Decision Services, was hired in 1996 to direct OM’s decision support activities. This hiring was concurrent with OM’s realization that information was key to
the company’s survival. A key strategic goal is “Turning information into knowledge, then into profits using information technology to help suppliers and customers make well supported
business decisions.”

In order to provide the information that internal employees, suppliers, and customers need, the decision was made to build a data warehouse. This decision was made without a formal ROI analysis,
but with an initial development budget of $750,000. The warehousing effort has been highly successful in making information available to the people who need it. Over time, the warehouse has been
key to decreasing inventory costs and increasing sales, and has helped OM become an industry leader (Eckerson and Watson, 2000).

OM views the data warehouse as a part of the company’s computing infrastructure. There is no chargeback of costs to the various business units. New applications that use the warehouse,
however, are subject to ROI calculations. Some projects originate with the data warehousing team, and most are typically of a cost-savings nature. Most of the new project proposals come from the
business units. Project proposals go through steering committees and must include their cost, internal rate of return, net discounted present value, and payback period. The culture at OM is to be
conservative in making estimates. A post implementation assessment is used to determine if the anticipated benefits were realized.

The ROI practices at OM are typical of firms in the medical and surgical supply distribution industry. This may be due to the cross fertilization of people moving from one company to another and
bringing their experiences to the new organization.

The data warehousing literature, theory, survey data, and case studies, when considered together, provide insights into how organizations use ROI in justifying and
assessing data warehousing and why they employ the practices that they do. The insights are presented as propositions.

Proposition #1: A data mart strategy may be selected to keep the cost of the data warehouse below a financial threshold in order to avoid high-level corporate review. Many firms
build a data mart as their first data warehousing effort. Ralph Kimball (1992) recommends this approach. Because of a data mart’s limited scope (a single subject area, serving a limited set
of users, with a limited set of users), it can be developed quickly, at a relatively low cost, and provide a fast return on investment. When successful, a data mart provides a proof of concept for
further data warehousing efforts.

The survey data shows that a prototype, such as a data mart, is often used in gaining approval for data warehousing. The case studies reveal an additional important factor in why a data mart
strategy is popular; a factor in addition to the usual speed, cost, and fast return on investments arguments. The data mart strategy may allow a warehousing initiative to keep below the corporate
financial cost threshold level where a comprehensive, high-level review is required for project approval. This was the case at Harlequin. Though not as explicitly stated, it may have contributed to
the data mart strategy used at Boise Cascade.

Proposition #2: Data warehousing ROI may not be calculated when the warehouse is part of a larger, strategic corporate initiative. The building of a data warehouse may be an
integral part of a larger corporate initiative, such as implementing a comprehensive CRM strategy. Without a warehouse, the strategy cannot be effectively implemented. The warehouse is a required
enabler for the business strategy. In this situation, the warehouse may not be subjected to an ROI analysis. The analysis may take place at the business strategy level, and the cost of the
warehouse is just one of several cost components. Also, the benefits are associated with the strategy and not the warehouse per se. The experiences at First American Corporation and Harrah’s
Entertainment illustrate this situation.

Proposition #3: When corporate survival is at stake, strategic necessity may trump normal analysis processes. First American Corporation was operating under letters of agreement
with regulators and its survival was in serious doubt. Its CRM business strategy was literally a “bet the bank” strategy, and could not be implemented without a data warehouse (Cooper
et al., 2000). The need for a warehouse was so great that ROI calculations were deemed meaningless, even though financial analyses are the norm in the financial services industry.

Proposition #4: Institutional theory helps in understanding data warehousing ROI practices. According to institutional theory, an organization adopts similar structures and
practices as other firms in its industry to gain legitimacy and enhance the likelihood of its survival (DiMaggio and Powell, 1983; Teo, et al., 2003).

Organizational decision makers turn to norms, standards, and solutions that are institutionalized in their industry to fit with its external institutional environment (Lu, 2002). Knowledge about
industry structures and practices is gained through observation, the movement of personnel from one firm to another, consultants, conferences, and the like.

Firms implement a data warehouse because of business need. However, the real and perceived need for a warehouse is likely influenced by the actions of others in the industry. A failure to follow
the actions of other firms may cause customers, competitors, investors, business partners, and other stakeholders to question the legitimacy of the organization, since it does not adhere to
practices in its industry. Consequently, if performing ROI calculations for capital expenditures are standard industry practices, it is likely that individual firms will feel pressure to operate in
mimetic ways.

Elements of institutional theory are apparent in the case studies involving Harlequin, Large Retailer, and Owens & Minor. The culture and practices in each firm’s industry was mentioned
as a factor in influencing whether ROI calculations were made. At Harlequin, the industry culture did not dictate the use of ROI. Rather, the logic of the business argument was the important
consideration. By way of contrast, at Large Retailer the warehouse is always under close financial scrutiny. This practice is consistent with the retail industry, where “everything has to be
quantifiable.” OM’s warehouse was an important component of corporate strategy, and ROI was not calculated in the initial approval process. New applications for the warehouse are
subjected to ROI analyses. The practices at OM are consistent with those in its industry.

Proposition #5: Option theory pricing helps in understanding data warehousing ROI practices. Option pricing theory was developed to assess the value of call and put options within stock
market transactions (Cheung and Bagranoff, 1991).
The purpose of acquiring the option is to benefit from the upside potential of the underlying stock at a relatively small cost, which is
equal to the cost of the option. Option pricing theory allows one to view projects as analogous to put or call options on a stock and values them using the techniques traditionally employed to
value put and call options on stocks (Smith and McCardle, 1999).

The analogy between financial options and corporate investments that create future opportunities is both intuitively appealing and increasingly well accepted (Luehrman, 1998). The investment in an
IT project (such as a data warehouse) is a digital options generator that positions an organization to capture the inherent value of opportunities that arise from the initial investment. Executives
readily see why investing today in R&D, or in a new marketing program, or even in certain capital expenditures, can generate the possibility of new products or new markets tomorrow (Luehrman,

While a data warehouse is typically developed with specific applications in mind, it is also often viewed as an investment in decision-support infrastructure.

Companies recognize that the warehouse will support future applications, many of which are currently unknown. The best example of the applicability of options theory pricing is Owens & Minor.
The company’s business strategy was to turn information into knowledge into profits, and this required a data warehouse. A budget was given for the project, but ROI calculations were not
required. Once the warehouse was in place, additional applications surfaced that have helped OM become a leader in its industry.

Proposition #6: When the continuing need for a data warehouse is obvious, there may be no post-implementation ROI analysis. Survey data shows that the most common method of
assessing a data warehouse after it is implemented is agreement that the warehouse is meeting its objectives. This fact is also supported by several of the case studies. At FAC, there was agreement
that the warehouse was an integral part of the ongoing CRM business strategy. At Harlequin, “the benefits (of the warehouse) were obvious” and no ROI calculations were deemed necessary.
At Imation, there was no call for a post-implementation analysis. M. D. Anderson had a continuing understanding that a data warehouse was needed. Finally, at OM, no post-implementation ROI was
necessary because it was viewed as part of the organization’s computing infrastructure.

Proposition #7: Individual applications of the warehouse may be subjected to ROI analyses, even if the warehouse is not. Applications that use the warehouse are developed in
response to needs in the business units. The creation of these applications requires expenditures that are typical of any software development project. Unlike the warehouse, these applications are
not typically viewed as part of the organization’s computing infrastructure. Consequently, they are commonly subjected to the organization’s usual assessment processes for new projects.
For example, at Boise Cascade, the business units justify expenditures for applications and the warehouse in their budgeting processes, because they pay for them. At FAC, Harrah’s, Harlequin,
and OM, all proposed new applications of the warehouse are subjected to ROI analyses.

Our case studies show that data warehousing approval and assessment processes vary considerably among firms. This finding is consistent with the data warehousing
literature and academic research. There are many reasons why different approaches are used. For example, firms have different cultures and processes for approving all capital expenditures above
some threshold level. Some firms choose a lower cost data mart strategy in order to “fly below” the “ROI radar.” In some firms, the need for a warehouse is so obvious that
making ROI calculations is pointless.

The various factors that help to understand and predict companies’ data warehousing ROI practices interact in multiple ways. For example, if company practices require ROI calculations for all
capital expenditures over some amount, and an enterprise data warehouse (rather than a data mart) is proposed, ROI calculations will be made. On the other hand, if a warehouse is needed as part of
a business strategy to save the firm, this business need factor “trumps” all other factors and the warehouse will be quickly approved without ROI calculations.

Our research provides examples and insights into the justification and assessment of data warehousing projects. This information should help data warehousing managers, consultants, and vendors
better understand the drivers and processes used for data warehousing ROI, and add to the growing body of research and knowledge on the topic.


Cheung, J. K. and N. A. Bagranoff. “Assessing Intangible Costs and Benefits in the Systems Decision via Option Pricing Theory,” Journal of Information Systems, 5(1),
(1991), 36-48.

Cooper, B. L., H. J. Watson, B.H. Wixom, and D.L. Goodhue. “Data Warehousing Supports Corporate Strategy at First American Corporation,” MIS Quarterly, (December 2000),

DiMaggio, R. J. and W. W. Powell. “The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields.” American Sociological
, (1983), 147-160.

Eckerson, W. and H. J. Watson, Harnessing Customer Information for Strategic Advantage: Technical Challenges and Business Solutions, Seattle: TDWI, 2000.

IDC Report. The Financial Impact of Data Warehousing, Toronto, Canada: International Data Corporation, 1996.

Inmon, W. H. Building the Data Warehouse, New York: Wiley, 1992.

Kimball, R. The Data Warehouse Toolkit, New York: Wiley, 1992.

Luehrman, T. A. “Investment Opportunities as Real Options: Getting Started on the Numbers,” Harvard Business Review, 76(4), (1998), 51-61.

Lu, J. W. “Intra- and Inter-Organizational Imitative Behavior: Institutional Influences on Japanese Firms’ Entry Mode Choice,” Journal of International Business
, (2002), 19-38.

McWilliams, B. “Easy Money, Tough Decisions,” Computerworld, Vol. 30, Issue 26, (June 24, 1996), 2-4.

Smith, J. E. and K. F. McCardle. “Options in the Real World: Lessons Learned in Evaluating Oil and Gas Investments,” Operations Research, 47(1), (1999), 1-15.

Teo, H. H., K. K. Wei, and I. Benbasat. “Predicting Intention to Adopt Interorganizational Linkages: An Institutional Perspective,” MIS Quarterly, (March 2003), 19-50.

Watson, H. J. and B. J. Haley. “Data Warehousing: A Framework and a Survey of Current Practices,” Journal of Data Warehousing, V2N1, (January 1997), 10-17.

Watson, H. J., J. G. Gerard, L. E. Gonzalez, M. E. Haywood, and D. Fenton.“Data Warehousing Failures: Case Studies and Findings,” Journal of Data Warehousing, V4N2,
(Spring 1999), 44-55.

Watson, H. J., D. A. Annino, B. H. Wixom, K. L. Avery, and M. Rutherford. “Current Practices in Data Warehousing,” Information Systems Management, Vol. 18, No. 1,
(Winter 2001), 47-55.

Whittemore, B., “The Business Intelligence ROI Challenge: Putting It All Together,” Journal of Data Warehousing, V8N1, (2003), 4-10.

Whiting, R. “Warehouse ROI,” InformationWeek Online, (May 24, 1999). (

Share this post

scroll to top