The key indicators of product correlation reflecting information friction and social network analysis method was used to discuss the effect of product correlation on the export of differentiated products in China. The results show that higher product correlation degree can weaken information friction and promote the export of differentiated products in China. Under the liberal taxonomy, GVC production activities provide an important foundation for improving the efficiency of technology and information transfer. Under the conservative taxonomy, the difference of classification methods will lead to a “product transfer effect” on the impact of information friction on Chinese exports. Betweenness centrality has a more significant promoting effect on the export of differentiated products in China. It reflects the higher control ability of network resources, the stronger control ability of information transmission, and the greater hub function of network operation efficiency, which is more conducive to the export of products. Geographical distance in traditional gravity model and product correlation reflecting information friction are both important factors affecting current international trade.