由於網際網路的基礎建設逐漸完善與教育普及下，國人對於網際網路的使用頻率與使用方式有大幅度改變，而本來在國外盛行的網路拍賣，也逐漸被國人接受，也有許多人在網路上再創事業高峰，但因虛擬通路存在匿名性，所以對於顧客忠誠度方面上，不像實體通路般能夠準確掌握，而本研究乃是先透過文獻探討實體通路上影響忠誠度之構面，並利用前測印證是否適用於虛擬通路，進而使用倒傳遞類神經網路來預測顧客在於忠誠度上的表現，以提供賣場經營者訊息，使其能針對其販賣產品特性選擇顧客群進行行銷，以達資源配置效率，在本研究發現到透過類神經網路進行之預測結果有相當高的正確度，足以給予賣場經營者在顧客關係管理上有效的訂定策略依據。

Under impacts of the maturity of Internet infrastructure and the increasing knowledge and understanding, the residents in Taiwan have largely changed the frequency and the ways of utilizing Internet. Moreover, online auctions have become widely accepted, and many of sellers on Internet auction begin to create considerable profits as well. Yet, to precisely assess customer loyalty online is much more difficult because of the anonymous nature on the Internet. Hence, we first discuss the dimensions which will influence customer’s loyalty over the physical distribution channels, and testify the suitability of these dimensions in the virtual distribution channels. Subsequently, Backpropagation Neural Network（BPN）will be applied to forecast customer’s loyalty and to provide useful information to the sellers. We expect the results can help the sellers better segment their customers and allocate the capital more efficiently. In this study, we conclude that BPN forecasting provides a tool with great precision in helping CRM and forming marketing strategy in customer’s loyalty forecasting through applying BPN.