以商品屬性建立顧客分類之新商品預測模型

Abstract

當企業在進行新商品行銷決策時，常經由分析同類型商品的過往銷售資料，推敲消費者可能購買的狀況，但由於各項商品間皆具有差異性，往往預測結果與實際狀況間可能出現很大的落差。除此之外，如何針對不同消費者族群的需求進行合宜的行銷訴求，亦為企業最關心的課題之一。 本研究採用國內著名電視購物頻道E公司的顧客資料做為分析樣本，打破企業以商品功能及相關資訊進行商品屬性分類的方式，改採用消費者購買商品的動機因素，跨商品別進行商品屬性分類，並重新將消費者分群，推測其可能的購買預測。 本研究首先利用內容分析法將168項商品進行商品屬性分析，分別將168項商品貼上商品屬性標籤，再利用集群分析法將商品屬性分為六大構面:「效用導向構面」、「價格導向構面」、「高價導向構面」、「流行導向構面」、「知性導向構面」、「休閒導向構面」。根據E公司2004年之常購顧客的購買資料及六大商品屬性構面，利用邏吉斯迴歸模型進行顧客分群，將顧客分為以下五群:「工程師型」，重視效用導向、知性導向；「休閒玩家型」，重視休閒導向，效用導向；「會計師型」，重視價格導向、流行導向；「品味設計師型」，重視流行導向、設計導向；「消費高手型」，重視價格導向、知性導向、設計導向。經由驗證得知，商品屬性與顧客分群間存在高度相關性。 本研究主要成果在於建構一套新商品預測模型(New Product Forecasting Model NPFM)，根據此新商品預測模型的實證結果提供企業在推出新商品進行銷售前，先針對核心顧客進行購買預測的重要參考，利用此預測結果將商品推出給主要客戶族群，達到最大效益。

When companies deal with new product marketing decisions, they usually analyze the historical market data of similar products in order to conduct sales forecast. Because each product is different from one another, there would be an obvious gap between forecast and reality. Furthermore, how to access different customer clusters through various marketing implications is one of the main concerns of companies. This research regards the customer database of TV-Commerce Company E as the analyzed subject. In place of product categories and functions, we conduct product classification by product attributes on which customers regard when making purchasing decisions. This research also clusters customers into several groups according to their emphasis on different product attributes. This research attaches the label of product attributes to 168 products by Content Analysis , and then generalized six categories as the purchasing factors by Logistic Regression. The core customers (top 20%) in TV-Commerce Company E`s database can be classified into 5 clusters by Binary Logistic Regression. It reveals high relevance between product attributes and customer clusters. This research succeeds in building up a New Product Forecasting Model (NPFM). This model could help companies to find out proper match between products and customers, rise up the marketing efficiency.