Being a pretty systematic and evidence based individual, I'm trying to come up with criteria for selecting a coach, and there is just no way to nail it down because there are so many exceptions to the rule. You do truly never know what you are going to get.
1. Proven at a high level: Saban was average at Michigan State. Pete Carrol was average. Both are examples of opposite. Jim Harbaugh is a curious failure of this where you get exactly what you want, but don't get the results you expect. Hugh Freeze would seem to fit the positive of this mold, but as others have noted, his overall win % was actual not great. A lot of nuance in that, but the point stands. Freeze could sort of be put in category 3a by that metric.
2. Success at a big school: What we typically want, but out of reach. The danger is they are playing with better players. Pruitt seems to fall into this category.
3. Did more with less:
a. Better for that school, but not stellar: Dan Mullen. Is he ever going to go to the next level or is he just going to be a really good coach?
b. Lesser program success: Justin Fuente, Dino Babers, Kevin Sumlin, Scott Frost. How do you know they can replicate it?
c. Up and Comer: Complete roll of the dice. Chad Morris, Tom Herman, Chadwell, Healy. Chad Morris and Tom Herman "should" have worked.
4. NFL Guy: Who knows? Lovie Smith just got fired. Pete Carrol won a championship. Charlie Weiss was a nightmare. Few NFL guys are "exciting" I'd love Joe Brady, but who knows?
5. Coaching Tree: Especially with Saban disciples. Dooley, Pruitt, Muschamp and others have failed. They know the recipe but it don't taste right.
These are just a few things I would look at, and there are frustrating exceptions to every category. I've got no clue how to choose a coach.