Omnichannel pricing and inventory strategies considering live streaming selling: A data-driven distributionally robust optimization approach

Recently, Live Streaming Selling (LSS) has become increasingly prevalent. Numerous omnichannel retailers are striving to introduce live streaming channel to absorb additional demand. However, it is challenging to investigate robust pricing and inventory strategies that consider the characteristics of omnichannel operations and LSS with uncertain demand. We consider a joint optimization of ordering, replenishment, order fulfillment, and pricing, where customers are sensitive to prices and delivery times. LSS can influence demand and benefit other channels to take free-riding. Furthermore, service level requirements are formulated as joint chance constraints to guarantee adequate performance. The Worst-case Mean Quantile-Deviation (WMQD) is employed to measure risks. The Wasserstein metric is adopted to design the data-driven ambiguity set. Accordingly, a data-driven Distributionally Robust Joint Chance Constrained Programming (DRJCCP) based on WMQD is constructed. Leveraging the dual theory, Conditional Value-at-Risk (CVaR) approximation, and linearization techniques, the developed model can be transformed into tractable formulations, which can be solved by commercial solvers. We further conduct numerical experiments to demonstrate the efficiency and practicality of our developed model. The comparative results reveal that the DRJCCP model based on WMQD has superior out-of-sample performance and is capable of effectively managing uncertainty, thereby ensuring more robust service levels. Furthermore, the sensitivity analyses are performed to verify the effects of some key parameters on the decision-making. The results indicate that introducing live streaming channel is not always profitable for the retailer and increasing the level of LSS effort can enhance free-riding effect without necessarily improving retailer's profits.