Abstract: With the development of mobile internet, the emerging e-commerce O2O (Online to Offline) business strategy with the support of the widespread use of mobile apps and e-transactions has gained significant popularity in the retail industry. As a business strategy, O2O commerce draws potential customers from online channels to physical stores according to offline stores' service level,vice versa. Therefore, some of customers change their buying channel from one to the other when service level of online or offline changes, which is called demand shift. Thus, channel demand will be affected, as well as O2O system's profit.
In this paper, with the consideration of impact of the service levels to demand in both online channel and offline channel, the optimal service decisions in supply chain O2O system are studied. The two kinds of situations are discussed, one situation is that demand does not shift, although affected by the levels of service; the other is that demand shift from one channel to the other with the change of service levels. In each situation, there exist two different decision models, i.e., the centralized and decentralized service decision models of O2O system, respectively. The centralized decision model means the brand supplier provides products to self-own stores; while the decentralized decision model means the brand supplier provides products to its franchise stores. The optimal service strategy of centralized system and decentralized system are investigated, respectively. More specifically, a two-echelon supply chain O2O system with one brand supplier and offline stores is considered. The brand not only sells the product to the offline stores but also sells directly to the end customer through its online stores. For centralized system, the total profit model is established to discuss the optimal service levels to maximize the total profit of the system. Further, similar to the centralized system, for the decentralized system, the profit models of the brand supplier and the offline franchisee are established, respectively. Then, the optimal service levels that make the brand and the franchisee achieve Nash equilibrium are discussed. Accordingly, the condition for Nash equilibrium solution is obtained.
The research results show:1) for the centralized system, when one of online and offline provides services, the O2O system can achieve the optimal no matter demand no shift or not. If both online and offline services are provided, only in the case that the demand is not shift between online and offline, the O2O system can obtain optimal profits. 2) for the decentralized system, whether one of online and offline provides services or both provide services, only if some conditions are met, the O2O system always achieve Nash equilibrium both in demand shift and no shift situation. Finally, a typical clothing brand M in China is taken as an example for numerical analysis. Some parameters are assumed, and the impact of service levels on profits is discussed. Our research is helpful to companies who make optimal service decision as they practice the emerging O2O commerce.

Key words: online to offline, demand shift, optimal service decisions, Nash equilibrium