When are you supposed to build an economic value model to justify a proposed information technology investment? Whenever the CFO or CIO asks you to. But a more subtle answer rests in a decision framework that can enlighten measurement practices by the nature of the IT investment a company wants to make. This framework suggests that rigorous measurement isn't so important when investing in certain kinds of technology, which may be seen as heresy given today's urge to measure.
When seeking approval for an IT investment, technology professionals should be as interested in knowing when to measure as in how to measure. We know that the company's culture and financial condition, plus the CFO's predisposition, can determine measurement practices. So, too, should the nature and class of the technology.
Here's the simple argument: Some IT is needed to run the company, and it enables the deployment of more strategic kinds of technology. Is it really necessary to even attempt to model the internal rate of return, discounted cash flow or even the payback of, say, a WAN or storage investment? A rigorous financial model might determine when in a company's investment cycle such an investment can be made, but the model should have little bearing on if it's to be made. Clearly, the investment should be made since this kind of IT supports more strategic technologies.
Consider what I call the economic value depiction pyramid. It helps answer this: When is it OK to confine a business case to a one-page summary or defense of the investment, rather than a detailed measurement exercise in which each cost and benefit metric -- however arrived at -- decorates an ROI calculation? Consider network-attached storage. Let's say your company is adding three applications and a couple of new data stores, and e-commerce traffic is picking up. The sheer volume of information seems unrelenting. As costly as new storage technologies are, would a payback period calculation do any more to secure the investment than if you presented that list of company realities?
At the bottom of the pyramid are the infrastructure-support kinds of IT: networks, storage, operating systems, servers and databases. Moving up the pyramid, we find more support types of IT: systems-of-record applications -- accounting, budgeting, inventory -- and desktop and collaboration software. From this level, we move into the "magic kingdom" of strategic IT: customer relationship management, supply chain, field-force automation -- anything that can give a company a competitive advantage.
As we move further up the pyramid, two things happen: The kinds of metrics used to model the benefits change, and the pyramid narrows. The higher up the pyramid, the more the metrics focus on strategic issues, such as increased market share, reduced cycle times and increased revenue.
The narrowing of the pyramid symbolizes confinement; the more strategic the IT, the less room for measurement error. The higher you climb, the more rigor and accuracy are required. This doesn't mean that measurement of the support kinds of IT can be cursory or slipshod. Nor should we confuse strategic for complex. A storage-area network is as complex a proposition as an integrated marketing automation application. However, strategic IT introduces novelty -- new ways of organizing business processes and defining job roles.
Many will object to this framework of aligning measurement rigor against class of technology. Enterprise resource planning (ERP), for instance, can be viewed as an infrastructure or support type of software because the entire company depends on it, so ERP is both strategic and essential. And a company might invest in point-to-point T1 connectivity between dispersed facilities as a strategic weapon, since it might allow the company to more effectively collaborate in delivering a product or service, enhancing its competitive position.
These exceptions reinforce this general rule: Some IT is essential but not strategic, and never will be. Companies that accept the contours of this proposition will spend less time measuring what's immeasurable and more time measuring what's novel and, perhaps, able to generate revenue.