台灣電子商務交易規模即將在2015年達到兆元產業，但至2013年為止，卻仍有60%的網路商店未獲利，為了改善現況，網路商家需要知道如何辨識哪些是可以帶來更多利潤的忠誠顧客。過去研究發現，商品屬性與買家忠誠度相關，本研究定義的忠誠度為：買家與特定賣家的交易次數與買家在平台上購買特定賣家有販售的商品類別的次數之比率，比率越大表示忠誠度越高。本研究的目的是分析買家所購買的商品屬性與買家忠誠度之間的關係，評估的商品屬性有兩個，分別為經驗品（相對於搜尋品）以及享樂品（相對於功利品），用以識別潛在的忠誠買家以提升網路商家的效率。 本研究以網頁內容探勘的方式蒐集2014 年“Yahoo!奇摩拍賣”公開的買賣雙方交易記錄，透過線性迴歸進行多種情境交叉分析。結果發現一段期間內偏好(購買)經驗品與享樂品較多的買家相較於同一期間偏好搜尋品與功利品的買家而言，在之後的一段期間成為忠誠度較高的顧客。此外，我們發現以不同長度觀察區間重複分析時，以半年觀察區間並加上限定之前期間及之後期間的交易次數需達2~4次以衡量其忠誠度時，上述顯著的關係最為穩定。本研究的管理意涵可以作為線上賣家非常務實的建議。 ;The sales of e-commerce in Taiwan will reach one trillion NT dollars in 2015, but there are 60% online stores made no profits by the end of 2013. To improve the bottom-line, online stores need to identify those loyal customers who can contribute more to their profits. Since the relationship between the product type and the buyer’s loyalty has been reported in literature, we define the loyalty by calculating the ratio of number of product types a customer purchased from a particular vendor to the number of product types purchased at the platform in a certain period. The larger the ratio the more loyal the customer is to the vendor. This study aims to analyze the relationships between the product types purchased by the customer and the measure of the customer’s loyalty. The product types are evaluated along two attributes, one is experience goods (versus search goods), and the other hedonic goods (versus utilitarian goods). The ability to identify the potential loyal customers will influence the effectiveness of their marketing effort. This study uses web content mining technique to collect the publicly available transactions data from Taiwan’s Yahoo! Auction website in 2014. We use the linear regression to cross-analyze many scenarios. The findings show that the buyers who in a certain period of time preferred (purchased) more the experience goods and hedonic goods before the cut-off time turn into customers with higher measure of loyalty after the cut-off time than those who preferred the search goods and utilitarian goods in the same period of time. Besides, we run the same analysis using time periods with different lengths, we found that when the length of time period is six-month and the required number of transactions for the measure of loyalty is 2 to 4 times, the above relationships with significance are more stable. The managerial implications of this study served as practical suggestions to online vendors.
