Fellow analyst and blogger Joe McKendrick has some interesting points on ROI for SOA. A number of analysts are chirping up on the subject, including respected colleague Neil Macehiter.
I think we need to get real and recognize that ROI is not going away. Yet I agree it's a hard test to make SOA pass. However, in a recent briefing with Jeff Stamen at Progress Software, he described working toward a best of ROI in SOA approach. And that is to define and prove ROI for SOA on a project basis, not on the holistic and longterm IT-strategy basis.
So in a pilot project, or for projects driven at the departmental level, SOA can and should show financial hard and soft benefits over traditional brittle approaches for applications that need integration and easy extensibility (and which don't these days?).
SOA can show ROI from the get-go on a project basis. Progress says they do it all the time. From that, more projects beget more SOA use and methodology adoption. Project by project, more work is therefore done to free up more data as a services layer. Same with more applications, components, and logic. Before you know it, SOA blossoms from tactical to strategic, perhaps paying for itself along the way -- and not as a "if we build it, they will come" bet.
Like a Jackson Pollock painting, a progression of drips and streaks of SOA use in the enterprise begins to cover the entire canvas. As the resuse benefits kick-in, and new infrastructure is designed with SOA in mind, the notion of applications shifts to a menagerie of WSDLs and available resources.
The key is to show ROI on a tactical basis on the project level. This is an investment by the vendors and SIs, as well as the users. Smart and aggressive vendors will seed the market with a lot of successful projects, application by application. Failure is not an option.