Abstract: Addressing the issue of logistics mode selection in e-commerce supply chains, we construct a three-tier supply chain model involving a single supplier, a logistics service provider, and an online retailer. By introducing the logistics service level sensitivity coefficient and the return rate, it systematically explores the decision-making logic of three logistics modes: direct mail, transit hub consolidation, and hybrid proportional direct mail. Combining the Stackelberg game model and backward induction, it derives the equilibrium solutions for each supply chain member and reveals the impact mechanisms of core variables on decision-making through sensitivity analysis and inter-model comparisons. The findings are as follows: Firstly, regarding the decision-making processes of supply chain members, transit hub consolidation emerges as the optimal logistics model, while direct mail is the least preferred option. All members tend to favor the transit hub consolidation model. Secondly, the logistics service level sensitivity coefficient positively impacts unit logistics service prices, logistics service levels, and retail prices (with no effect on wholesale prices) and drives profits for all members. Additionally, the effect of return rate on retail prices is moderated by the commission ratio threshold (showing negative correlation below the threshold and positive correlation above it), but exerts a negative suppression on other decision variables and profits. Finally, numerical examples are employed to visually illustrate the theoretical findings.