With this update, I am introducing a newer, easier-to-analyze format, the graphics provided by Survey Monkey. This first question constitutes the (really) short form:
For folks willing to fill out a bit more information, I broke the question out into not just a success metric, but also into costs:
This is turning into an interesting experiment in survey design, and an illustration as to how various surveys can come up with different results. Notice that, in the short form, the vast majority considers their project outcome to have been mostly successful. In the long form a different picture emerges: a considerable proportion of end users are disappointed with projects that also usually come in late and cost more than planned.
I am not sure how the picture of success on one form meshes with a decidedly more mixed one on another. Perhaps there is selection bias at work here - those who answer the short form tend to participate in more successful projects. Or perhaps developers consider their project successful even when end users do not. I tend to think that end user satisfaction is more important :)