As large scale E-commerace platforms rise, the delivery demand peaks are becoming more remarkable. This is mainly caused by sale promotions and one-day delivery promises. In this field, there is an implicit assumption that re-allocating the deliveries to level delivery peak results in lowering customer utility and thus, the mainstream approach is to make efficient the work plan in given conditions. This paper analyzes real delivery data, and indicates the possibility of delivery re-allocation without lowering customer utility, if an transdisciplinary approach of customer analysis and vehicle routing is applied. A delivery allocation system based on this implication was proposed, and a simulation based case study was conducted. Through this, the effect on delivery demand peak leveling without lowering customer utility, was verified.