Customer Satisfaction Management(CSM) is becoming more common and the importance of measuring the CS Index(CSI) is getting higher and higher. Furthermore, the existing CSI have many problems in view of the link between enterprise result and those indecision, which could not explain the links clearly. Therefore, NPS(Net Promoter Score) index have been developed by Fred Reichheld and the NPS shows up rapidly and attract attention. In this paper, we provide conceptual frameworks on the links among the latent variables(total CSI, elementary CS and the referral) and propose more effective and practical index between CSI and the "would recommend" index through testing the Structural Equation Model(SEM) on "CSI${leftrightarrow}$would recommend${leftrightarrow}$referral".