This study considers a problem of coordinating production, transportation and sales in a multi-echelon supply chain network. A simulation model is built to generate the random customer demands at different locations, which are affected by a marketing strategy. Customer demands need to be satisfied by the supply chain through production, transportation and distribution. The optimization problem for coordination of production, transportation and distribution is first formulated as a linear programming with demands as input parameters in the constraint. Our objective is to maximize the expectation of the optimal profit of the supply chain given random demands by selecting an optimal marketing strategy. A simulation optimization technique is proposed to control the generation of random demands and solve the linear programming for efficiently learning the optimal marketing strategy. Numerical results show that our method can significantly improve the expected profit of the supply chain and reduce the computational burden of solving linear programming for achieving a given level of probability of correct selection of the optimal marketing strategy. Furthermore, we extend the optimization problem to a mixed integer programming and also demonstrate the computational efficiency of our proposed method.