So I have a question for the audience: when you are measuring ROI on SOA projects, are you looking at just the SOA team production, or do you look at secondary effects?

That's assuming you look at ROI on your SOA projects at all; apparently 40% of companies engaging in SOA projects don't. I'm not entirely sure how I feel about that. I am generally pretty militant about ROI metrics, but on the other hand, I recognize that there are certain scenarios in which the very process of measuring complex second and third order investment benefits can be more complex and costly than those benefits are ultimately worth.

Since SOA is almost entirely about downstream benefits, I can imagine that for at least some of those 40%, this decision makes sense. But the inevitable question in tight economic times is, how can you prove the project is worthwhile?

This all comes up again because Joe McKendrick has hauled Anne Thomas Mane's "SOA is dead" mantra out and dusted it off again, provoking Manes into reiterating her stance (which is hardly "SOA is dead" but I recognize that sounds much flashier) and setting off industry executives who have a vested interest in selling the stuff.

For what it's worth, I think Manes' "SOA is dead as a term if not a concept" is premature. Her assertion that business people have concluded that SOA is expensive and doesn't deliver results presumes that non-IT executives have heard about it, understand it, and have established those metrics that I started out asking about, and I don't believe there is much evidence for that. Cut backs in this environment can hardly be pegged on a particular failed concept; it's not entirely clear to me that what SOA is, or isn't, or is labeled, or not, even belongs in conversation outside the IT department, anymore than IPX/SPX or TCP/IP do. My feeling is that IT departments which are being controlled at such a level that they are being precluded from making their own decisions as to the best architecture for service delivery have worse problems than justifying SOA budgets.

But whether you're talking to executives inside or outside the department, the original question remains: to what extent to do you try to look at your ROI on SOA projects? Do you need to? Or is the philosophy sufficiently convincing that it can be adopted without strict metrics, like, say, electricity and phone service?