With so many IT budgets currently stretched to their limits, many data warehousing project managers wind up in the hot seat, trying to justify their data warehousing efforts to upper management. However, determining return on investment (ROI) for a data warehousing project is no simple matter. One expert at The Data Warehousing Institute (TDWI) 2003 World Conference, conducted in Boston last week, offered tips on calculating ROI for data warehousing projects, a process that is new to most IT pros.
"A lot of data warehousing efforts today are under scrutiny," said William McKnight, of McKnight Associates, which has offices in Dallas and Boston. "It's not a bad thing; a lot of what we were doing over the past couple of years [was] speculative."
Company executives aren't prepared to sign off on data warehousing efforts so easily. They want to know how data warehousing will help get reports out, how customers are using those reports, and how it's driving revenue back into the enterprise, McKnight said.
William Lay, IT director at Los Angeles-based Public Communications Services, a national provider of pay phone management systems for prisons, said he has had difficulty justifying his data warehousing efforts. Lay said his "CEO realizes every one else has [a data warehouse], but he isn't really sponsoring it."
Lay used a data mining application to track the payment of collect phone calls from inmates. He then procured a data mart to identify possible scenarios of fraudulent use of the system customers, also known as churn management. Lay managed to map this information into a dollars-and-cents equation for his supervisors.
Instead of this piecemeal approach, Lay said he would prefer to consolidate targeted projects into a single data warehousing program.
Addressing the needs of various user needs presents a challenge for Jeff Blackburn, project manager of Largo, Fla.-based Cox Target Media Inc.
"The problem is that business customers are looking at data from different sources," Blackburn said. "Data integrity is a big issue."
Blackburn attended McKnight's session to glean tips on determining data warehousing ROI. "We're trying to figure out how to compute it. It still doesn't sound easy," Blackburn said.
McKnight told attendees that, depending on who you're justifying your data warehouse to, there are four different approaches to determining ROI:
ROI: (returns - investment) / investment
This calculation should be supported with a time frame (for example, three years), and it should be presented with assumptions and risks in an itemized format. "Don't be married to the way you do data warehousing," McKnight said.
When presenting ROI numbers, it's also important to remember that past successes don't matter much. "It's very much a what-have-you-done-for-me-lately mindset," McKnight said.
Net present value: Another popular ROI approach, net present value (NPV) represents the future value of the initial investment based on its cash flow minus the initial investment. McKnight said NPV estimates should stay within a three-year range.
Internal rate of returns:
The rate of return, or interest, that would make the present value of future cash flows plus the final market value of an investment equal the current market price of the investment or opportunity.
Payback period analysis:
This is the simplest ROI calculation. It's an estimate of when cash flows will turn positive and does not include discount rates.
"Our job is not to architect data warehouses only," McKnight said. "It's to tie it in with business returns. Data warehousing is such a fundamental part of so many positive programs that it's hard to believe it won't help solve a business problem.
For more information: