[experiments] Re-running reset experiments generates invalid results

Result screen after re-running an experiment that was reset: