Part of the issue is that in the college game, the margin for error is really, really low. For instance in the NFL, if you win 75% of your games, you're doing great. In college? You could actually lose your job depending on where you are.
So analytics does a fantastic job of showing you things to do provided you accept the downside. The problem is for instance you are winning a game, you take a risk because analytics says to, that goes wrong and you lose. Does it matter it was a high chance of success if it costs you the game?
In the NFL that's acceptable because it sorts itself out over time and you will succeed more than you fail, so on so forth. But to reiterate, you're not really allowed to fail in the college game. To put it in another context, heavy reliance on the pass also works better in the NFL than in college. I can rattle off a lot of successful college coaches, who won a lot of games airing things out but never won a title in part because high risk high reward works enough to be very successful, but it's another thing to work enough to win a title.
