All process improvements are not created equal. How should we select which processes (or process steps) to improve? How do we approach this for a really large migration project? Start with understanding the potential for improvement and then narrow it down from there.