For the past 2 days, I've participated in the PMI Kansas City Chapter's Professional Development Days. On Monday, I gave an overview of Scrum. Yesterday I gave a presentation on User Stories. One thing I mentioned in both presentations; while Scrum won't fix all your problems, it will make them more visible so that you have to deal with them.
After yesterday's presentation, one attendee came up to talk about a problem he was having with his projects. His end users were very slow at providing requirements information. Questions in emails would go unanswered for a week or more. If he got key stakeholders together on a conference call, they would debate among themselves and no decisions would get made. Being geographically dispersed added to the problem.
My advice to him was to capture as much detail as he could but don't hold up the iterations waiting on input. He should build what he could based on the limited input he got and push things to production. An application with limited functionality was better than no application at all, and this would help force the stakeholders to provide additional input.
I find the same advice applies when I'm implementing business processes. If a process is broken, taking any step to quickly apply a fix is better than doing months-long analysis on what the solution might be. Once the quick fix is applied, you can see how effective it is and move forward from there.
It's application of kaizen and lean. Find the biggest issue, take care of it, then go back and see what your next biggest issue is. I told my presentation attendees that they should look at their projects with a product view; they may not get everything right in the first release, so they should plan on going back and updating the application in the future.