建立適合顧客關係管理之模糊分群模型－以汽車維修服務為例

Abstract

對大多數的企業來說，顧客關係管理以成為企業在市場上的競爭利器，若企業能夠對顧客的交易特性進行區隔，進而提供合適之服務與產品，將可提高資源配置的效率及行銷策略的有效性。但企業常用的RFM模型與其他硬式分群法如K-means對於多變性、不確定性，或是多重性顧客資料，皆無法顯現良好分群效果，因此本研究主要將RFM變數引入Fuzzy C-means分群演算法中，在資料分群過程中，保留資料與資料間的不確定性，最後利用分群測量性指標評估最佳分群數，建構出符合多變性、多重性顧客特性的分群模型，並將顧客歸納於隸屬群體中，提供完整分群分析之框架。 本研究利用國內一汽車總代理公司所提供的汽車零件服務作為個案研究，並分析各群體之交易特性，共可分為三個群體，參考三個群體的基本統計變數建立顧客分類規則，作為企業對新客戶快速分類之參考依據，最後依據各群交易特性，進行經營管理策略上之建議，可有效提升顧客滿意度與降低庫存之壓力，並使得總代理與經銷商服務廠均達到雙贏的地步。

A lot of companies have implemented Customer Relationship Management as a competitive weapon. It will raise the efficiency of internal resources allocation and the efficacy of marketing strategy, if companies can provide customerized service or products to customers who are clustered according to selling record. Common used clustering methods such as RFM model or other hard clustering methods ﹙eg. K-means﹚ produce unreliable result when processing data is uncertain or Multiple. Therefore, this study put RFM variables into Fuzzy C-means algorithm which will retain data uncertainty during clustering process. Finally, clustering number validity is combined to this clustering model and established a complete clustering framework. To validate the proposed model, field data from a local automobile general agent is collected. 136 customers are clustered into 3 clusters. The statistic description from three clusters is analyzed and referred to classification rules for new customers. Suggested managerial strategies for each cluster are effective to raise customer satisfaction and lower inventory.