現今的行銷模式必須考慮到消費者的「動態性」和「異質性」,本研究以國內某銀行的信用卡顧客為研究對象,嘗試以資料庫行銷的理論與統計方法達到此目的,並分析其顧客價值。首先,結合RFM模型和層級貝氏模型、最大概似估計法等估計方式,以購買區間、活躍性、穩定性等指標分析顧客價值。其次,利用馬可夫鏈理論建立移轉機率矩陣模型,以預測顧客的未來消費之狀態,達到平均54%的擊中率。並設定6條路徑,以逐步迴歸法分析價值遷徙路徑與顧客特質之間的關聯性。本研究的結果能幫助企業了解其顧客價值,並提供模型估計顧客未來之狀態,以利銀行規畫未來行銷策略,並建立長期顧客關係。
The main purpose of this considering consumer dynamic and consumer heterogeneity by theory of database marketing and statistical method, is to recognize the customer value which is based on a domestic bank’s database of credit card customers. In the first part of this research, we analyze customer value by indexes such as inter-purchase time, reliability index, activity index conducted by RFM model, Hierarchical Bayesian model and maximum likelihood estimates. Second, we build individual customer’s Transition Probability Matrix of Markov Chain to predict customer purchase state, and the average hit rate reaches 54%. Further, we set 6 meaningful paths and find out if there is any correlation within the transition paths and demographic variables. The contribution of this study could help enterprises verify customer value, identify customer’s transition paths, develop follow-up marketing strategies and build up long-term customer relationship.