Abstract
For many companies, the economic justification of a data warehouse investment is becoming an increasingly common prerequisite for project authorization. In this paper, a framework for data warehouse investment decision will be presented, the particulars of data warehouse cost/benefit analysis will be examined, and a model for external consultant involvement will be offered.
1. A Framework for Data Warehouse Investment Decision
In order to more accurately apply cost justification techniques for data warehousing, it is important to understand the context within which the decision to proceed with a warehouse is made. There are two important aspects of this decision to consider, namely what motivates people to build data warehouses, and how do they choose a data warehouse project over any number of competing IT projects.
1.1 Data Warehouse Motivators
The primary pressure leading to the development of executive information systems historically has been an increasingly competitive environment[1]. The same appears to be true for data warehousing. For example, the Telecommunications industry, which is in the midst of deregulation and experiencing intense competition, is a very hot market for data warehousing. Goodhue noted that competition creates uncertainty, and uncertainty creates a need for information that previously established rules and procedures may not be able to satisfy[2]. The detail and richness of the information in a data warehouse is one way to satisfy the information needs created by competition. In particular, companies are striving to know more about the customers and create products and services that meet their individual needs more fully. The reorientation of company from a manufacturing to customer service is a fundamental cultural change that demands data warehouse solutions.
1.2 Data Warehouse Decision Approval Model
Today's increasing economic and competitive pressures, decreasing margins, and scarce resources are compelling many companies to more carefully examine their IT investments. Global competition is moving the selection of the right IT projects from desirable to imperative. Data warehousing projects must complete against other IT projects for scarce IT investment dollars. At a very high level, IT investment decisions can be described by one of three models -- the mandated decision, the rational decision, and the non-decision decision.
The Mandated Decision Model
In the mandated model, the decision to proceed with a data warehouse project is based upon an executive mandate rather than a cost/benefit analysis showing an acceptable payback[3]. Whether the mandate is the result of some visionary planning or a "fire drill," the result is the same -- no justification is required.
The Rational Decision Model
In the rational model, the selected project must show an expected cash flow that produces an acceptable return. Bacon adds an additional criterion, "The ideal is for an IST project or investment to be undertaken in pursuit of both: (a) quantifiable net benefits and (b) explicitly planned business objectives[3]." Experience has shown that clear alignment with business objectives supported by credible numbers produces the most compelling argument.
The Non-Decision Decision Model
In cases where the technology is very new or the company is particularly risk averse, a decision to delay the decision is commonly made. A non-decision decision can take the form of pilot projects, limited roll-outs, or "try-and-buy" vendor agreements. (Note: the reference to pilot projects here should not be confused with the use of pilot projects to control project scope or to test infrastructure under controlled conditions). The non-decision decision is usually arrived at when the rational decision model fails to produce a compelling argument.
Once the decision to proceed with a data warehouse project is made, a secondary question arises, namely, "Which vendors should our company partner with in order to increase the probability of success?" Vendor selection here is essentially a matter of confidence in the vendor's company, their people, their track record, and their products. Some information sources for vendor selection include face-to-face meetings, reference sites, documentation, conferences and seminars, and previous experience.
Both the mandated decision and the non-decision decision are outside the scope of this paper.
2. The Particulars of Data Warehouse Cost/Benefit Analysis
2.1 General Characteristics of Data Warehouse Project Economics
Data warehousing has a number of interesting characteristics that make it different from typical IT investments. Three outstanding characteristics of a data warehouse project are the front-end loading of costs, event versus continuous benefits, and the risk of scheduling overruns.
Front-end Loading of Costs
More than most IT application projects, data warehouse projects incur the majority of their costs in beginning of the project (e.g. hardware, vendor software, consulting, and training) -- measurable benefits, if any, usually come much later. Many times data warehouses are sold like traditional application systems, promising productivity gains and improved decision-making. The problem with this approach is that data warehouses are burdened with a much larger infrastructure cost than the typical application system, making it an unfair comparison.
Event Versus Continuous Benefits
The continuous cost savings benefits usually associated with on-line transaction projects are generally not found in data warehouse projects. On-line transaction projects are expected to produce cost savings by reducing headcount, improving though-put, or avoiding expenditures like mainframe upgrades. Data warehouses are usually justified, after the fact, based upon "event" benefits, enabled by existence of the warehouse, data access tools, and trained users. For example, an insurance company may uncover systematic fraud -- saving the company millions of dollars. These savings are real, just hard to predict.
Risk of Scheduling Overruns
The chief risk that data warehousing projects face is premature project termination due to scheduling overruns. There are two aspects of the data warehouse project that are the culprits for these overruns -- data conversion and scope management. If either of these two aspects get out of control, the result is usually a dramatic lengthening of the project life cycle. Additional expenditures, such as adding contractor help, seldom makes up the difference in these cases -- the project has to run its course.
2.2 Cash Flow Analysis: The Basis of Financial Evaluation
The financial evaluation of proposed IT projects is based upon capital budgeting theory, in which all the cash flows associated with a project are combined to produce a pro forma statement. A key concept in cash flow analysis is the time-value of money. That is, a dollar today is worth more than a dollar tomorrow. Data warehouse projects are clearly (and negatively) impacted by the timing of cash flows (remember, the majority of expenses occur early in the project as compared to benefits that occur later in the project). So it is not only the volume, but the timing of costs and benefits that determine the project's financial performance. Please refer to the Suggested Reading section of this paper for references on capital budgeting theory, cash flow analysis, and financial indicators.
2.3 Popular Financial Indicators Used in Cash Flow Analysis
Once the expected cash flows have been collected for a project, any number of financial indicators can be applied to the stream of payments. Some commonly used financial indicators include Cash Payback, Net Present Value and Internal Rate of Return. In practice, a company may employ several financial indicators to an expected stream of payments. These financial indicators give the company a method of comparing various IT investment alternatives with unlike cash flows.
2.4 Other Selection Criteria
In addition to performing cash flow analysis, a company may also employ a number other selection criteria, including[3]:
Budgetary Constraint (is the project budgeted for)
Business Objective Alignment (how well does the project tie business objectives)
Probability of Benefits (the likelihood of actually seeing the promised benefits)
2.5 Benefit Tangibility and Estimation Accuracy
Expected benefits from an IT project can be grouped by their degree of tangibility and estimation accuracy. That is, some benefits, like headcount reduction, are easy to measure; while others, like improved management decision-making, are hard to estimate. Just like any good Gartner Group graph, it is desirable for a project to have its benefits in the upper right quadrant of the following graphic.
2.6 Cost Estimation Accuracy
Just as benefits vary in their estimation accuracy, so do costs. Some costs, like hardware purchases, have very firm costs; while others, like data conversion expenses, have a history of wild inaccuracy. Of particular concern to data warehousing is the difficulty in estimating data conversion costs. Data conversion costs represent a very wicked problem, in which very little may be actually known about true data quality until work begins, but a great deal assumed, believed, and relied upon during the planning stages. Careful and thorough data audits are required for accurate project planning. Witte categorized a number of IT project costs by their estimation accuracy[6].
| Good Estimates | Poor to Terrible Estimates |
| Hardware Purchase | Installation and Conversion |
| Vendor Software Licenses | Data Integration and Access |
| Networking | Maintenance |
| Internal Software Development | Ongoing Support |
3. A Suggested Consultancy Model
It has been previously mentioned that the most compelling argument for a data warehouse is a clear alignment with explicit business goals, supported by honest numbers. The outcome depends upon the credibility of the customer's internal sales skills. A data warehouse consultant can play an important role in this process. In the following section, the general nature of that role will be considered and some best practices will be presented.
3.1 The Consultant's Role In Internal Sales
The consultant's role in the internal sales process should be one of behind-the-scenes coaching and counseling. Experience has shown that vendor-supplied cost justification numbers will be given very little credence.
Several years ago, the author was involved in a data warehouse/decision support system project at one the largest beverage companies in the United States. The senior vice president, who as our corporate sponsor, was certain that the type of promotional analysis that the system promised could deliver millions of dollars benefit -- easily enough to justify the project. Nevertheless, our sponsor was very clear about one thing -- we, the vendor, were to be nowhere around when he presented the numbers to the management committee. It was his credibility alone that would make the case.
3.2 Best Practices for Data Warehouse Internal Sales
The following catalog of best practices has been collected from a number of case studies and from the author's experience. Like most consulting engagements, there is no mechanistic solution available. What is important is helping your customer be successful in their internal sales efforts.
Learning About the Company Culture Through Their IT Investment Evaluation Methods
Much can be learned about a company's culture and their attitudes towards IT investments by their methods of project evaluation. A hard line on cost containment and the denial of intangible benefits may reflect a narrow view of IT's role in company's core business, as this quote from the Harvard Business Review reflects, "Finally, many business -- though certainly not all -- have accepted that IT can play a strategic role rather than simply displace costs. Such basic attitudes toward technology tend to be deep and persistent. The company's technology principles should reflect these attitudes -- not try to change them...[5]." An emphasis on cash payback may reflect a short-term orientation; in which managers, fearing corporate raiding or unfriendly takeovers, will only authorize those projects that guarantee quick and reliable return on investment[3]. In either of these cases, cost justifying a data warehouse with its reliance on intangible benefits and heavy infrastructure costs, will be an uphill fight. Of course, external pressures, like competition, may come to the rescue, obviating the need to cost justify a data warehouse to a skeptical management committee.
Techniques for Handling Intangible Benefits
It is safe to assume that presenting intangible benefits as part of the cost justification for a data warehouse will meet with skepticism. Nevertheless, intangible benefits are real -- they're just hard to measure. The internal sales goal for intangible benefits is credibility. Therefore, the consultant should be very conservative, disclosing his or her methods, avoiding any appearance of self justification. Some alternatives for gaining credibility for intangible benefits are:
Claim only supportable intangible benefits. Consider
"pre-discounting" their worth to control the internal
valuation process that occurs during peer review[4].
Record intangible benefits along with all of the project's
other costs and benefits at three levels -- planned,
best, and worst case. Compare the three scenarios to get
a sense how the project's cash flows might vary. (Experience
has shown that the best case in not much better than planned,
but the worst case is much worse than planned)[6].
In a technique dubbed "Information Economics", Pastore suggests
quantifying the benefits and rank ordering them according to their
contribution toward accomplishing corporate goals and objectives[7].
The projects with the greatest contribution get selected.
Suggestions on Dealing with Taxation and Other Minutia
Dealing with the intricacies of capital budgeting, depreciation schedules, optimal switching and taxation can be tricky business for the consultant, especially when the business is in a foreign country. Thankfully, many times the tax implications of a data warehouse investment can be overlooked for one of two reasons:
In companies which have very sophisticated capital budgeting
practices, it is generally the sole responsibility of the finance
department to evaluate the tax implications of IT investments.
The consultant's only job, then, is to assist the customer in gathering
the cost and benefit cash flows; or,
Given the softness of the numbers in a typical IT investment analysis,
many companies may consider the influence of tax implications on
selecting the right project to be negligible.
Be Aware of the Impact of Politics on Cost and Benefit Estimation Accuracy
The effect of politics on accuracy of cost/benefit estimation can be dramatic. Political struggles can lead to unwise projects being accepted or good projects being rejected based on overly optimistic or pessimistic net benefit projections. One study of IT cost estimation concluded, "The study's most important implication is the formal recognition of the existence and pervasiveness of the Political Model"[8].
Establish Methods and Measures for Capturing Benefits Data As Part of the Pilot Project
The preparation of a data warehouse benefits audit has two key inputs: business user interviews and utilization statistics. Establishing measurement systems for warehouse utilization early in project will make a system benefit audit a much more manageable process. Some middleware products, like HP's Intelligent Warehouse, provide some utilization data automatically.
Avoid Some Common Pitfalls in Cost Estimation
There are some tricky areas in cost estimation that the consultant should be aware of. One of most glaring is software licenses, which are frequently expressed as a percentage of current list price, and thus may be raised without purchaser authorization[6]. Another area of frequent cost over-run is desktop tool and middleware add-ons. For example, a set of business users may become very excited about a certain DSS tool and push its approval forward, only for IT to find out that it is dependent on middleware that the company has not licensed. This may significantly increase the "price per seat". Data warehousing solutions almost always mulit-vendor solutions, which may only be integrated on the marketing slides. The total installation time is rarely the sum of the individual installation times.
Properly Assign Benefits for Maximum Credibility
Benefits should be assigned to the level at which they occur. Conoco found that is was impossible to place a total value on their executive information system. Comparing the company's "bottom line" before and after did not create a meaningful picture because there were too many exogenous factors that affected the company (e.g. national economy, actions of competitors, etc.)[4].
Summary
Data warehousing is limited in its ability to produce traditional OLTP type, tangible benefits. Further, data warehousing carries a significant infrastructure burden. Nevertheless, cost justification is possible. It is suggested that the internal sales effort adopt the following strategy:
lead with a strong business case clearly tied to explicit business objectives
support it with analogous success stories
back it up with reasonable numbers -- even if those numbers show a loss.
Finally, the consultant's role in the internal sales process is to act as a behind-the-scenes coach and counselor.
References
[1] H. Watson, K. Rainer, C. Koh, "Executive Information Systems: A Framework for Development and a Survey of Current Practices," MIS Quarterly, March 1991
[2] D. Goodhue, M. Wybo, L. Kirsch, "The Impact of Data Integration on the Costs and Benefits of Information Systems," MIS Quarterly, September, 1992
[3] C. Bacon, "The Use of Decision Criteria in Selecting Information Systems/Technology Investments," MIS Quarterly, September 1992
[4] L. Belcher, H. Watson, "Assessing the Value of Conoco's EIS," MIS Quarterly, September 1993
[5] T. Davenport, M. Hammer, T. Metsisto, "How Executives Can Shape Their Company's Information Systems," Harvard Business Review, Vol. 67, No. 2, 1989
[6] D. Witte, "Full-Life-Cycle Economics: An Evaluation Methodology for Information Technology Projects," Journal of Organizational Computing, 4(4), 1994
[7] R. Pastore, "Many Happy Returns," CIO Magazine, June 15, 1992
[8] A. Lederer, R. Mirami, B. Neo, C. Pollard, J. Prasad, K. Ramamurthy, "Information System Cost Estimating: A Management Perspective," MIS Quarterly, June 1990
Suggested Reading
T. Davenport, "The Case of the Soft Software Proposal," Harvard Business Review, May-June 1989 -- A good short case study involving intangible benefits and internal sales of IT projects.
C. Bacon, "The Use of Decision Criteria in Selecting Information Systems/Technology Investments," MIS Quarterly, September 1992
-- An excellent review article, examining the criteria used by 80 organizations in choosing IT investments.L. Belcher, H. Watson, "Assessing the Value of Conoco's EIS," MIS Quarterly, September 1993 -- A good case study of IT project auditing.
Brealey, R. and Myers, E. Principles of Corporate Finance, 3rd Edition, McGraw Hill Book Co., New York, NY, 1988 -- The standard college finance text, providing an excellent background for capital budgeting theory.