JUSTIFYING A DATA WAREHOUSE PROJECT: PART I
by D. J. Power
What is the return on investment for a proposed data warehouse project? What is the payback period? What is the opportunity cost? What are the anticipated benefits? What can we do with a data warehouse that we can't do with our current information systems? Do our competitors have a data warehouse? You may be asking these questions about a data warehouse DSS project. Are you receiving satisfactory answers? If you hesitated and said "maybe" that is not surprising because justifying a data warehouse project can be very difficult. Why?
For many years, business professors have been discussing the issues surrounding financial evaluation of capital expenditure projects. We don't all agree. Typical tools recommended are ROI, NPV, and discounted cash flow. These tools are closely tied to the capital budgeting process. Each is intended to provide a rational allocation of capital to projects.
Managers are asked to spend funds on a data warehouse project; the first reaction is to request that anticipated results and benefits be quantified so that the requested expenditure can be evaluated in comparable units. But for a data warehouse project it is difficult to quantify the results and benefits. We will need to make gross estimates and guesses. A financial analysis is especially difficult because the costs are uncertain and many of the benefits are qualitative and intangible.
A number of qualitative analysis tools can be used as alternative means for evaluating data warehouse projects. Peter Keen (1982) recommended using value analyis and staged prototype development for DSS projects. Parker, Trainor and Benson (1989) developed an Information Economics approach to IT project value analysis. Maglitta (1997) discusses "intangible value analysis". In this approach managers evaluate "soft" benefits such as improving staff productivity, improving the speed of strategic actions, enhancing a company's competitive advantage, or improving access to data. A related approach, the business value added approach "measures IT contribution not in dollars, but by its support of key goals and metrics of functional groups (check Maglitta, 1997)". A fifth alternative, the scenario approach attempts to project what decision making will be like when the data warehouse is in place and hence speculate on how the company will benefit. All of these qualitative approaches have pluses and minuses, but each can be improved by understanding the upside and downside of a data warehouse project. Let's briefly review what managers are reporting in companies that have implemented data warehouses.
Upside potential
The primary benefit of data warehouses should be improved decisions. This intangible benefit presumes that managers will change their decision processes and actually use a data warehouse. In a recent Sentry Market survey, 30% of respondents identified "access to data" as the biggest benefit of a data warehouse. Other important benefits included: improved data accuracy; gaining competitive advantage; better control of data; better data consistency; decentralization of data; cost savings; and less reliance on legacy systems. Few respondents thought that there would be cost savings.
According to Theresa Rigney, Sentry Market Research's 1996 Software Market Survey "shows that 55% of the respondents polled declared that data warehousing is vital to their organizations' business objectives." The following case examples identify reducing the number of programmers, reducing staff hours spent preparing reports, and improved sales staff productivity as tangible benefits of a specific data warehouse implementation. "Your results may vary!"
All of the data warehouse vendors and the Data Warehousing Institute (DWI) have glowing reports of data warehouse successes at their web sites. For example, to justify their data warehouse investment, UPS managers "looked at both tactical and strategic benefits. The biggest tactical benefit was a sharp decrease in the cost of programmers. Roughly half of the 130 programmers on staff were not required after the data warehouse was up and running. Users developed their own reports instead of ordering new features from the IT department. That savings alone justified the investment in the data warehouse."
In 1992-93, Ertl Co., Inc., Dyersville, Iowa, a manufacturer of replica toys, built a data warehouse that eliminated 18,000 hours a year in report creation work. This amounted to several hundreds of thousands of dollars a year in savings (see the DWI case study).
A Hewlett-Packard case study quotes Greg Friedrichs, H-E-B grocery company's manager of decision support systems. He states "There's no question that our new decision support system represents a significant investment, and a leap of faith. We can't point to specific sales results that we've achieved by having this system. But on the other hand, none of our users can imagine how they ever did their work without it."
In a mini-case study of MCI titled SOLD Delivers Quantifiable Results , Flanagan and Safdie quote an MCI manager. "The best measure of the quality of our calling lists, and of the service we provide when we deliver them, is salesman productivity--sales per individual... ."Since we've modernized our system, that figure has grown 200 to 300 percent. What's more, this has all happened when our sales force grew from 300 to 2,000 strong. Just keeping up with growth would have been impossible before this system, let alone improving the quality of our deliverable". So analysts need to ask if improved staff productivity can be expected.
These case study and survey results are positive, but we also need to examine the major analysis of the ROI for data warehouse projects that is being discussed by vendors and consultants. The next part of this article will review the International Data Corporation white paper titled "The Foundations of Wisdom: A Study of the Financial Impact of Data Warehousing" and summarize some downside issues.
Finally, in part 2 we'll see what can be concluded about justifying a data warehouse project.
References
Last updated November 14, 1997 by D. J. Power. Copyright (c) 1997, D. J. Power.
The concluding article in this series will appear in the next edition of D S * .