I would agree that there are many lines of work where I too would be astounded by a manager's inability to be able to point to what "better performance" looks like. But I speak as someone working within a central agency, where a great deal of what goes on is data analysis and policy development. I'm not anybody's manager (thank heaven), but for many of the people I work with, I'll be damned if anybody could point to what "better performance" is and how it could be validly measured. If I look at the managers around me, I have no idea what their performance could be indexed by. I do know that senior officials think they have a set of managerial performance measures for awarding performance pay which are effectively devoid of any validity. I also have access to government-wide data, and boy oh boy, never underestimate the oddness or idiosyncratic nature of government jobs, because somewhere out there is someone with just about the weirdest job you can imagine, such that feeding a set of performance indicators about that job back to senior levels will make them tilt their head sideways like the RCA Victor dog.
That's the long way of saying that an individual manager might have some sense of what better performance might look like for employee X, but not necessarily anything that could be applied beyond that particular employee such that the performance of work units Y and Z could be compared. It's that elusive goal of organizational performance, as opposed to employee performance, that is the problem/challenge.