「顧客關係管理」(Customer Relationship Management)觀念在零售業行之有年,透過數位技術的發展、資料庫的建立,企業蒐集顧客個人資料與觀察其交易行為已非難事,但如何妥善地運用統計工具分析手中資料,藉以適切配置行銷資源,進而吸引不同屬性、行為的顧客群,才是發展顧客關係管理、資料庫行銷的核心目標。 由於企業資源有限,且各自擁有不同的核心能力,顧客價值(Customer Value)的衡量成為行銷研究領域不可或缺的重要項目。透過RFM指標與資料庫中各統計變數的關聯性分析,可帶給行銷人員科學化的策略建議。然而,目前行銷研究中最常使用的OLS迴歸分析法,其假設樣本母體呈現常態分配,而以平均數或中位數的概念設法找出變數間的相互關係,是否真能符合真實世界的樣貌?因此,本研究採用Koenker & Bassett (1978)提出的分量迴歸方式(Quantile Regression) ,以國內3C連鎖零售賣場為研究標的,進行顧客價值分析。希望能藉此發掘不同的統計方式帶來的不同分析結果,也藉此驗證分量迴歸分析在行銷上的應用是否合宜。 本研究分析區分成兩個部分:第一部分利用RFM指標來界定顧客價值,並透過資料庫所提供之人口統計變數、顧客消費行為變數與顧客價值作交叉分析。並同時進行OLS一般迴歸的分析,藉以比較兩種迴歸模型的異同。第二部分則選擇三項產品線作為研究主體,分析購買其高低價位商品的顧客之人口統計變數與交易行為,藉此可探討通路導入新產品時,其定位在不同價位可採取的不同行銷策略。 研究結果可明顯發現分量迴歸在分析具有極端值與偏態較大樣本時,相對於OLS一般迴歸,擁有較佳的母體分辨能力。同時也呼應了行銷學上對於顧客「異質性」的強調,使得行銷人員得以更了解位於不同分量的客群,不同的行為特質。
“Customer Relationship Management” now is a common concept for retail business. It has become much easier for enterprise to collect customer information and observe their behavior with the development of database technology. However, choosing the correct statistics tool to analyze information, to allocate resource and attract target customer is always a big issue for the marketers. Because of the limited resource and different core competence, “customer value” has become an essential part in marketing research. By analyzing the correlation of RFM indicator and information collected by customer database, the marketers can get scientific cues for marketing strategy decision. However, Ordinary Least Squares (OLS) regression model, which is commonly adopted in recent marketing research, describes sample as a normal distribution. Therefore, OLS method analyzes the correlation within factors by their average. To reflect the real distribution of the sample and get more precise result, this research adopt Quantile Regression(QR) model, which was brought up by Koenker & Bassett(1978), to analyze customer value of 3C retail chain store in Taiwan. Hoping to discover different result leads to different statistics tool and verify QR`s application in marketing.