This paper is on assessing the quality of adaptation results by a novel confidence measure. The confidence is computed by finding evidence for partial solutions from introspection of a huge case base. We assume that an adaptation result can be decomposed into portions, that the provenance information for the portions is available. The adaptation result is reduced to such portions of the solution that have been affected by the change. Furthermore, we assume that a similarity measure for retrieving the portions from a case base can be specified and that a huge case base is available providing a solution space. The occurrence of each portion of the reduced solution in the case base is investigated during an additional retrieval phase after having adapted the case. Based on this idea of retrieving portions, we introduce a general confidence measure for adaptation results. It is implemented in the area of workflow adaptation. A graph-based representation of cases is used. The adapted workflow is reduced to a set of sub-graphs affected by the change. Similarity measures are specified for a graph matching method that implements the introspection of the case base. Experimental results on workflow adaptations from the cooking domain show the feasibility of the approach. The values of the confidence measure have been evaluated for three case bases with a size of 200, 2,000, and 20,000 cases each by comparing them with an expert assessment.