I suppose I'd like to see what threat models actually apply to companies, and what vectors are most often used to get pwned (outside of social engineering). I understand what encryption prevents, but I don't understand how helpful it is in practice.