I don't think there is a "good" dropout rate - without knowing why students drop out, the number is meaningless. But my overall view is that education (on any level) should be measured by the improvement in knowledge/skills/competence/... it provides to students and not by its ability to classify students into those who pass and the rest.
In this view, ideal (but unachievable) drop out rate is zero as you make sure that only students that are a good fit for the program enlist and then you do your best to lead them to maximal achievements they are individually capable of.
In the real world, it is hard to know for both the student and the institution whether the program is a good fit. Also people and their priorities change over time. So I guess there are a few "valid" reasons for dropping out:
- The student's life situation unexpectedly changes (e.g. has a child)
- Although the studies proceed well, the student realizes that he/she wants to do something different with his/her life. This should not include the case when the student had incorrect information on what the study program is actually like.
- Early in the program the student realizes that he/she is unable to fulfill the expectations of the program (that were known to him/her before enlisting)
- The student behaved in some unacceptable manner and was asked to leave.
To me the goal is to have very few dropouts of this "valid" kind and exactly zero dropouts for other reasons. In particular I think that if students fail because they learn midway that your program is too hard for them or because the environment is too harsh or because their ideals of what the study program would look like were shattered, than it is a failure of the institution (although often the student is also at least partially responsible).