You are 100% correct that data availability is always the first problem to solve. However, I think this is addressed indirectly in his thesis that advanced anal

You are 100% correct that data availability is always the first problem to solve. However, I think this is addressed indirectly in his thesis that advanced analytics are brittle in a rapidly changing business. Any change that breaks your data by definition breaks your models.