A testing technique and apparatus are described for apply a test to a System Under Test (SUT) in one or more configurations of the SUT. The test can generate and store multiple output results that capture the behavior of the SUT in performing the test. Policy analysis logic applies a policy to the output results to generate an output verdict that reaches some conclusion regarding the outcome of the test. That is, the applied policy maps a subset of stored output results (and possibly input parameters) into the output verdict based on specified rules. A tester can apply different policies to investigate different aspects of the SUT's behavior.