There can be a vast number of ROI interpretations. A project may be based on a perception - a project leader insisting that the four minutes saved on boot-up time translated into a $40M savings over the course of a year.
Or, a client being disinterested in any analysis as they are going ahead with their installation anyway. They just wanted to come up with some numbers detailing why they should.

As with all forms of communication and analysis, we first have to establish what we wish to accomplish. In an ideal world, our answer would always be "to find the correct answer." In the real world, ROI analysis encounters the business, personal, and political agendas of a large number of people in any given organisation. This encounter between abstract theory and the intense pressure of business should lead to predictable project goals.

There are three general reasons for ROI analysis:

• Justification of an existing project
• Rationalisation of a previous expenditure
• Persuasion to take a specific course of action

Roughly half of the ROI projects fell into the first category. In these cases, the CIO wished to loosen the budgetary or political constraints on an existing project. The stated purpose of the analysis was to identify whether to go forward with the project. The unstated assumption was that we would find for the invested political capital.
A quarter of the projects fell into the second category. In these cases, the CIO put himself in political hot water. He invoked the ROI analysis to justify his past decisions and to show his skill in planning future projects. Findings that didn't support these purposes came into hot dispute.
The final quarter of projects fall into the third category. In these cases, the CIO needed tools with which to argue for a favoured course of action.

Most of the literature talks about cost reduction: - how we generate ROI by cutting into downtime, removing performance barriers, and eliminating service calls. Fundamentally, cost reduction is only modestly persuasive. It suggests that the way we are currently doing business is wasteful and that another way generates budgetary results.
Cost avoidance is even worse. It assumes that we "make money" by somehow removing a risk or barrier. The consultant who insisted four minutes a day boot-up time translated into $40M a year represents a classic example of this kind of thinking. Although legitimate examples of cost avoidance exist (avoiding downtime during financially critical moments in time), these are few and far between. Even downtime avoidance, the holy grail of ROI, doesn't really pan out in many cases. 99.999 percent uptime doesn't mean that the uptime is productively used or that the downtime would have effected an actual revenue-generating transaction.

Of the three, only the reduction of cost of entry into a new profit is a persuasive support. This kind of ROI comes from technologies that allow companies to reduce the actual cost and risk of exploiting new opportunities. They include reducing time to configure for a new market, time to fill the order, the product design cycle, and the internal resistance to change. Too many executives put to much emphasis on a simple evaluation of ROI. A project taking 10 years to pay back is deemed a bad project... But they don't see that it is an entry into a new profit area or new business that has to be done to keep the competitive edge.