The segmentation method taking production attribute utility as segmentation variable is based on the causality variable rather than description variable. Based on Hainan Mazada case 2004 of the new products' market segmentation, this paper conducts the market segmentation with product's attribute utility as segmentation variable first, and with product attribute's level utility further by K-Means cluster method. The paper gains some clusters with different attribute preferences and different attribute's level preferences, and finally describes the attribute characters of all the clusters.