To reduce the furniture distribution cost and improve the logistics efficiency, on the basis of the third-party logistics distribution method, we construct a mathematical model of the open vehicle routing problem (OVRP) with the shortest total driving distance and minimum number of vehicles as the objective functions and design an improved two-stage tabu search algorithm to solve the problem. The traveling salesman problem path of all clients solved in the first stage is used as the basis for dividing the OVRP path in the second stage. We design a stochastic dynamic tabu list. The "neighbor operator number" and "neighbor exchange point pair" are simultaneously used as taboo objects, thereby avoiding excessive taboos. In addition, we test five neighborhood operators, which show that the multi-neighbor structure consisting of four operators, namely, point exchange, sequence point insertion, point reverse order, and front point forward insertion, is the best. Through the case tests and literature comparisons, we verify the effectiveness of the improved tabu search algorithm. The use of third-party logistics distribution is more cost-effective than the use of self-operated logistics.