B2B企業之顧客採購行為分析 ─ 以醫藥公司為例

Abstract

隨著時代的變遷，企業之間的競爭日益激烈，從前的企業想著如何降低成本來增加獲利，但在行銷的知識傳遞下，不僅是B2C企業，B2B的企業也開始思考如何了解自己的顧客，進而提供更符合客戶需求的產品與服務，如此不但能增加獲利，確立長久的關係，更能建立一個可維持的獲利模式。 透過資料庫的建立，公司可以進行成本、存貨的控管；但如何從資料庫中獲得有效的資訊，幫助企業了解市場需求與客戶潛在的行為準則，才真正考驗著企業的行銷思維與經營策略是否與時俱進。本研究針對此問題，利用層級貝氏統計模式（Hierarchical Bayesian），針對B2B企業之交易資料庫，分析現有顧客的採購行為，作為一對一的顧客關係管理基礎。 本研究方法以層級貝氏理論 建立模型，使用馬可夫鏈-蒙地卡羅方法進行參數估計，探討採購間隔（Inter-purchase Time）與產品售價、採購量的關聯性，並導入客戶區隔項目，分析區隔變數對總體行為的關聯性。 針對S醫藥公司資料庫，將實際交易紀錄分成四大品類進行實驗，結果發現在口服與外用品項中，部分客戶顯著服從於假設的採購行為，且根據總體層次的分析得知，信用等級B級的客戶，較其他客戶更具有顯著的行為表現，未來若能根據此結果制訂關係維繫與行銷的策略，相信會有相對應的顯著回應。

More and more B2B companies have begun thinking about how to understand their customers deeply, and provide more products and services to meet their needs. By doing this, they can not only increase profits, but also build long-term relationship with customers. Through the creation of database, companies can control cost structure and the inventory; but how to get effective information from the database to help understanding the demands of market and customers’ potential behavior are really testing the companies’ marketing knowledge. This research focuses on the issue, using the Hierarchical Bayesian Statistical Model to analyze the database and find out the customers’ purchasing behavior. In order to explore the relevance of inter-purchase time, product price and purchase quantity, the HB method is used to build the model and parameters are estimated by Markov Chain - Monte Carlo method. Besides, the actual transaction records are divided into four major categories to analyze and the customer segmentations are also analyzed in the overall layer. As the result, some customers have significant behaviors consistent with the assumptions in the oral and external category, and the customers with B class rating in credit are more significant than others. In the future, I believe that the marketing strategy and relationship based on the results would be more useful and effective.