A coarse-grained model is used to investigate how the added non-conductive nanospheres (NS) influence the network formation of nanorods (NR) in polymer matrix. First, an algorithm is developed to calculate the formation probability of the percolated network of NRs. Then based on it, conductive probability is obtained which presents a gradual increase with decreasing the NS size. It can be rationalized by analyzing the improved effective volume fraction and the optimized percolated network of NRs, which is induced by NS. Meanwhile, conductive probability shows an initial drop and then rises with enhancing the NR-NS interaction. This is because the connection behavior of NRs can be affected by tuning the NR-NS interaction, which can effectively influence the percolated network formation of NRs. It is noted that shear flow produces a significant influence on the percolated network, which leads to the similar connection behavior of NRs. Meanwhile, conductive probability initially rises and next goes down with increasing the NS volume fraction. NSs can be beneficent to rebuild the percolated network of NRs at the low volume fraction under the shear flow. Interestingly, the appeared phase separation between NRs and NSs breaks the percolated network down along the shear gradient direction at the high NS volume fraction. In addition, the NS volume fraction is similar at the maximum conductive probability for small NS sizes. Meanwhile, the maximum conductive probability initially rises and next declines with improving the NS size. This can be explained well by the balance between number and size of NSs.