I agree with many of the sentiments described here.
I'd add that we need to be clear IF we intend to upgrade to SQL2016 at ANY point.
Someone has to go first otherwise there will never get to the luxury of choosing N-1. I'd be curious as to what approach the brave souls who went first took.
If I was planning on early adoption I would be looking at full parallel running. I would also look to use it as a lever to getting or enhancing TDD/BDD practises, monitoring and test tools. I suspect that the test community would be naming their children after me!
I'd be looking at stress testing at multiples of peak load and a full DR of my largest system.