Marketers have been using observed social-demographic variables to segment markets by tools like cluster analysis. In late years, along with the new statistical tools development, we have seen the tremendous potential of knowing customers much deeper by "seeing" the unobserved heterogeneity among customers. The trend is no doubt good news for organizations of both profit or non-profit. In non-profit marketing, uncovering the attributes of the people who have regular or one-time donate cash for charity purposes is vital to sustaining the resource for their mission by setting successful donor retention strategies. We believe that opening the new window to get a more profound knowledge of donor's unobserved heterogeneity, such as social-psychological attributes, is a watershed for nonprofit marketing. Based on the prior study of the model profiling by a Structural Equation Modeling (SEM) for who and how charitable donors' prosocial behavior experience sharing with what responding affections, Chang & Du, 2021. This research uses FIMIX-PLS as the primary tool to look into the new segments' unobserved heterogeneity and propose advice on donor retention and acquisition strategies. The analysis process has successfully partitioned the data into four segments with distinct segment features. We chose three significant segments which provide meaningful insights for strategy references. Besides, we have also illustrated our experience in the criteria of choosing the number of segments. Finally, we conducted segment comparisons using the PLS group comparison mechanism, which resulted in rich information that helped us understand the different donors distinguished by different segments.