The place to do that is to really push hard when a program comes in with its performance objectives and measures and ask, “How are you going to get this data? Can you get it?”
We have a very elaborate regime, in many cases, that has layers on it, but it turns out that in some cases, departments are not able to report in the way they intended, and you end up with an evaluation on program performance that is inconclusive. When you get one of those evaluations, it's a trigger to go back to the department and say, “Okay, rethink your performance metrics.”
I'm probably making my friend Antoine squirm here, but he knows how I feel.