This article presents a method to optimize the formulation of ceramic masses for coatings using the statistical design of experiments. The project of simplex mixtures in pseudocomponents considered six control factors (wt. %): talc, quartz, limestone, phyllite, dolomite and clay and three response variables: linear retraction, bending resistance and water absorption. The proposed method was implemented in the stages: (1) identification of the problem, (2) planning and execution of the experiments and (3) optimization and discussion of the results. The optimization method is defined through the simulation, imposing the restrictions that each response demands.