You've deployed the technology. Now it's time to gauge the payoff. In many organizations, winning support for your initiatives from senior management means demonstrating the ROI of your IT expenditures. But is ROI a valid standard for proving the value of IT projects?

David Linthicum, CEO of Bridgewerx and a noted application integration and SOA expert, says yes. In a world where business and technology are tightly aligned, Linthicum feels it is only natural that shareholders and senior management should expect hard numbers to demonstrate that their money is being well spent. But Gene Rogers, chief technologist for transformational space systems at Boeing, sees it differently. According to Rogers, it's far too easy to read too much into numbers that are ultimately open to broad interpretation.

David Linthicum: While everyone is bound and determined to build IT architecture within their enterprise -- perhaps even between enterprises -- few have attempted to determine their return on investment to justify their approach and their technology choices. While it's appropriate to approach new technology with fervour and excitement, we also need to figure out the business case for this movement.

These days, IT architects and businesspeople must work hand-in-hand to determine if changes proposed are the proper course for a business. For example, the application of an SOA has different degrees of ROI, depending on the problem domain. The cost of implementing an SOA should be directly related to the benefits to the business, in both hard and soft dollars. Indeed, you can determine ROI, and should do so.

It's easy. The ROI of IT can be presented in a general formula. The first steps are determining the variables and making sure they are acceptable and reasonable in the context of the business. Some of them, such as "cost of inflexibility" and the "lack of agility," are judgment calls.

Gene Rogers: The problem with any metric is that you maximize what you ask for at the expense of everything else. Be very careful what you ask for; you just might get it. Generally, optimizing any single metric or small cluster of metrics results in a sub-optimal solution overall. Management by algorithm is the least effective way to run a company.

In David's example, the core issues of profit and loss and value to customers, investors, or shareholders is ignored. Ambiguous terms such as "cost of inflexibility" and "benefit of agility" are exactly the kind of indeterminate numbers that can be spun up or down by a clever manager who is inclined to play the system. In the meantime, while our clever manager is maximizing your metric-of-the-month, the core business is likely to be falling apart. When IT reduced the cost of manual processing, for example, what happened to the freed-up employees? Was the automated end product equally as effective as the earlier manual product? These are all questions that an experienced executive will want answered.

DL: I'm not sure that "management by algorithm" is a good way to put it. We use maths as a tool to determine productivity and results, and we need to ultimately determine the ROI from IT to justify its existence. The metrics are going to differ a bit from business to business, because what's important in each business is always something different. Moreover, you always have to consider factors such as the value IT delivers directly to the customer, competitive advantages, the dynamic nature of the marketplace, and other factors.

As I said, you can indeed model this. The efficiencies of a well-run IT shop go right to the bottom line; this is proven.

Any ROI calculation needs to consider "soft" issues such as the customer experience and effects on the marketplace. Thus, the value to the shareholder is also a result of having a good ROI from IT. Customer satisfaction goes up? That's all a part of the metric.

To your point, IT ROI should consider how IT affects the entire organization. Indeed, in some cases it should include the removal of humans who are doing slow, costly, and error-prone manual processing. If we can automate that, all the better. But you never achieve the ROI if that has a negative effect on the core business in any way.

GR: Certainly, the efficiencies (and inefficiencies) of IT go directly to the bottom line. But, just as you say, ROI calculations need to consider the "soft" issues, and few people can agree on what the key metrics are or how to effectively measure them.

I've managed technology development and implementation for a couple of decades, and from year to year I've been asked to demonstrate my program's contribution to ROI in terms of profit margins, price-to-earnings, return on net assets, profit before taxes, profit after taxes, reinvestment, product transitions, growth forecasts, or even inventory turnover. Boards of directors reflect and tier down the metrics that are being tracked by current or prospective investors and shareholders. Next year, they're guaranteed to focus on a new and different set of hot-button metrics.