A mechanism is posterior efficient if the agents' expected utilities, given the outcomes of this mechanism, can never be Pareto improved. A non-revealing (resp. completely revealing) mechanism is posterior efficient if and only if it is incentive interim (resp. ex post) efficient.
Forges, F. (1994). A Note On Pareto-optimality in Differential Information Economies. Economics Letters, 46(1), 27-31. https://doi.org/10.1016/0165-1765(94)90073-6 (Original work published 1994)