When designing a project, do you always prioritize closing the feedback loop of ROI (or lack thereof) as early as possible?

Might you ever choose to defer closing that loop because you need something else sooner? If so, what kind of something could that be?