Cross-border e-commerce such as AliExpress provides great opportunities for small to medium sized enterprises (SMEs) to enter into the global market. Using scraping data, this paper investigates the impact of hyper-sales, in fact the platform subsidies, on these SMEs in the context of AliExpress. Noticing that there are many homogeneous products sold by different retailers, I carefully construct 70 groups of identical iPhone cases with 220 products in total and ensure that there are variation of hyper-sales within each group. I show that the variation of hyper-sales across time is high, and this further enables me to use panel data and group fixed effect models to quantitatively estimate the effect of hyper-sales. I found that hyper-sales are in fact effective ways to increase sales controlling the characteristics of the products and sellers - hyper-sales may lead to an increase of 10 units on average for every two days. These results are robust under a suite of robust tests, and there is no heterogeneity between old sellers and new sellers. I also found that consumers may not respond well when the hyper-sales switch too quickly. This indicates that the platform and policy makers may continue using this marketing strategy to help SMEs grow, but they may need to think carefully about who to sponsor and when to sponsor.