傳統行銷方式對於消費者行為的了解,通常採行市場調查的方式,然而其研究結果仍然是一種預測,而無法完全反應實際狀況。隨著電腦科技的發達,顧客資料的記錄與儲存效率提升且成本下降,使得「資料庫行銷」日漸受到重視。有別於傳統大眾行銷僅利用人口統計變數進行市場的區隔,資料庫行銷重視消費者間存在的異質性,以顧客購買行為為基礎,了解顧客的需求,提供更加客製化的產品與服務,與顧客建立良好的關係,並且讓行銷資源達到最有效的運用。 本研究以國內某線上購物網站為對象,藉由實際的交易資料進行資料庫行銷的實證分析,並根據歷史交易紀錄進行分析與預測,並提出適當的行銷策略。研究方法首先利用RFM模式為基礎進行顧客價值分析,接著利用最大概似估計法與加權最大概似估計法計算顧客平均購買期間,最後利用關聯法則分析進行顧客購買商品之相關性分析。 實證結果如下: 一、利用顧客價值分析可將顧客分成「高價值」、「高潛力」、「成長性」及「不穩定」等四群,並提出一套模型進行銷經費之分配。 二、利用平均購買期間分析可得知顧客平均消費時間的間隔,並可依此規劃促銷循環的週期。 三、利用產品關聯性,分析顧客購買商品間的交互關係,從而提出交叉銷售以及該線上購物網站商品推薦系統的建議。 根據上述實證結果,可幫助個案公司更加了解顧客,並發展能與顧客維持長期且緊密關係以提升顧客忠誠度的策略,進而提高企業獲利能力及行銷投資的效益。
In traditional marketing, the most frequently used methodology in understanding consumer behavior is the marketing research. However, the result is only the prediction that can’t reflect the real situation. The development of computer technology results in the efficiency improvement and cost reduction of the customer data recording and storage. Therefore, database marketing becomes more and more emphasized. Database marketing focuses on the heterogeneity of consumers to provide customized products or service according to the purchase behavior. Besides, it’s also helpful to build close relationship with customers and achieve the best allocation of the marketing resources. This research focuses on an on-line shopping retailer and uses the recorded transaction data to do the empirical analysis. The methodologies of the research includes the RFM model to segment by the customer value, the MLE and WMLE to calculate the average purchase period of the customer and the Association Rule to analyze the correlation of the products that customers purchase. The results of the analysis are as following: 1.Using customer valuation analysis to segment the customers into four groups which including “high value”, “high potential”, “high growth” and “uncertain”, then proposes the model to allocate the marketing budget. 2.Using average purchase period analysis to calculate the interval of each purchase by customers and develop the promotion schedule. 3.Using the Association Rule to analyze the correlation of what customers purchase and propose the cross-selling and the recommendation system suggestions of the website.