Software testing is a core activity in quality assurance. To improve the test process, we can use best practice models which describe in detail what to do in organizational test processes. The improvement activities using best practice models are performed as follows: checking the current status of test processes, suggesting and planning new actions, and implementing the actions. However, it is difficult to apply all of these actions to the organization due to the limitation of resources. In this paper, we suggest a strategy for optimizing test process action plans. The background of this research is TMM (Testing Maturity Model), which is the most representative test process model. By applying design of experiments to the TMM assessment procedure, we can accept the actions selectively by statistical significance and find the best solution.