Abstract: Retailors of short lifecycle products, such as fashion goods, have a single replenishment opportunity during the sales season. The initial order quantity, replenishing time and quantity are three interrelated decisions. The improvement of forecast is time-dependent because of the observed demand information increasing with time. The time-dependent improvement of forecast is integrated into a two-stage inventory model by adjusting the range of estimated demand interval in the second stage. Four order policies (PMNI: the profit maximization assuming no inventory left at the end of the first stage, PMSI: the profit maximization assuming some inventory left at the end of the first stage, CM: the cost minimization of the first stage, and MD: meet the demand of the first stage in maximum degree) are proposed to determine the initial order quantity and illustrate the process to find the best one. PMNI is the best choice among the four policies when there is no inventory left over at the end of the first stage. As for the total expect profit of the two stage, if the whole sales price of the first stage is more than that of the second stage to a certain value, PMNI is dominant than PMSI, otherwise, vice versa. Handu, which is a typical fast fashion E-commerce enterprise in China, is selected to illustrate how to choose the best ordering policy based on the actual total profit. The empirical data analysis illustrates the profit in case of the two-stage inventory model considering forecast accuracy will improve 76% than that in case of single-stage inventory model, and 300% more than that in case of the two-stage inventory model for offline fast fashion products without considering the dynamic changes of forecast accuracy (Fisher et al, 2001); in general, MPSI will be dominant comparing with MPNI. If the range of forecast demand interval is large, the conservative ordering policy, like MPNI and MC will get more profit, especially the cost of overstock is more than the cost of lost-sale. The way of integrating the forecast accuracy into the inventory model to decide the replenishing time will inspire the research on time decision of other inventory model in different cases. The analytic findings and empirical observation will help the retailor to choose the best ordering policy to get more profit.

Key words: quick response; newsvendor model; demand forecasting; replenishing time; fast fashion