I agree with Michael on the need to focus on complete, end-to-end processes to get maximum benefit from efforts undertaken, but would like to add one more suggestion.
Do not -- I repeat -- DO NOT role out PI efforts widely across an organization. It will only consume resource time and attention and achieve minimal incremental improvement to bottom line results. The typical role-out of Six Sigma (and its predecessor TQM) goes out and deals with every link of the value chain. There is only one (or at most, very few) processes that are limiting an organization's ability to achieve more than its goal. Focus on that one constraint. All other efforts will result in either at worst ill-advised cost-cutting, reducing capabilities for future growth, or at best, be an exercise in managerial masturbation, feeling good, but achieving nothing. Work the weakest link. Strengthening other links will do nothing to strenghten the chain. For more on this, see a webpage on constraints and TQM. You might also want to check out an article on choosing a "strategic constraint."
Even in the choice of IT projects to spend your scarce resources on should be influenced by this thinking as well. There is a "whitepaper" on this subject found at www.focusedperformance.com/articles/firstcut.html.
