Not having a metric set out in advance may, in fact, be more rigging resistant! unfortunately.
But the real deal is that none of it matters because the program itself is the goat sacrifice. It's not there to work, it's there to signal the proponents values and ingratiate themselves with the goat suppliers. It's actually better for theses purposes if it DOESNT work, because then you can perpetually campaign on solving the same problem.
For example, if you implement a free lunch program for all public school students in a state, you actually get to subtract out the amount of money that each school district spends dealing with reduced and free lunch program qualifications for a subset of students, all of the paperwork involved in maintaining that list, and the system that manages their account balances at the cafeteria register.
Measuring success is good, but we also need to exercise the ability to make small updates and continue measuring, instead of just using that data to argue on the side of "throw it all out!" or "we can't touch it, it's essential!"