If I had control of a team's draft, I would start with a pure analytics look such as Manny's, looking at their numbers starting at their Draft Eligibility minus 2 years in finding where to send my scouts. From there I would employ
Text Mining of Scouting Reports based on the completion of said reports from my scouts, but also from the other sources available on the web (but to a lesser confidence). Leading up to the actual draft I would have my final internal list drafted from both the pure numbers look and the pure text-mining look, and then sit in the room with my scouts for a week and go through it pick-by-pick in comparison to other compiled lists like Bob McKenzie's and such, identifying if there's any possible 'trade-downs' to make and grab the ones that my model and scouts really value.