大數據分析客戶行為樣貌之個案研究

在高度競爭金融環境上，如何維持高價值客戶的忠誠度，並開發潛藏在數百萬的潛力價值客戶是金控業者非常大的挑戰，尤其是提前掌握客戶偏好與需求，讓客戶體驗變得更加個人化，進而達到「圈客」與「集客」的效果。而金融業之數據分析，經過數年的發展，大多已能針對客戶基本特性、產品持有、交易行為及收益貢獻等，為客戶打上「事實標籤」。但是這只反應已發生的行為，缺乏客戶興趣及行為的「衍生標籤」，讓金融業對消費者行為的洞察，少了重要的一塊拼圖。因此，如何擴大可應用的數據範圍，並利用更即時且自動化的方式，為金融客戶貼上「事實、衍生、預測標籤」，以師法阿里巴巴，達到「千人千面、一客一策」，為本研究之目的。基於上述研究背景動機及目的，本研究將以國內某金控公司為例，分析其如何利用大數據來擴增數據廣度，以建置客戶單一視圖。探究個案公司如何進行數據洞察與數據處理，以提昇數據決策績效。最後分析其如何發展「一客一策」之精準行銷推薦機制，以提昇行銷精準度。

關鍵字

精準行銷；大數據分析；金融科技； AI動態貼標機器人；機器學習

並列摘要

In this highly competitive Banking industry, how Banks engage, retain, grow VIP customers and to keep them being loyal is an ongoing challenge. In order to win customers’ hearts and wallets, to identify potential customers and to predict their behaviors from millions of customer-base are the key successful criteria for Banks. This helps to create a hyper-personalized service for every customers. Therefore, the transactional data analysis is one of the solutions for financial institutions. It brings a powerful insight into customer’s product needs, preferences and behaviors. However, the analyze of the transactional data represents only one type of information assets that banks possess. This only reflects the behavior that has occurred, and lacks of the "derivative label" of customer’s interest and behavior, which makes the financial industry's insight into consumer behavior missing an important piece of puzzle. Hence, to expend the data source, and to use the real time data for revealing customer’s needs in advance will be our researching purpose. In this study, we will learn from Alibaba to label customer by multiple dimensions. These includes “Facts, derivations, prediction labels”. And take a Financial Holding Company as an example to show how them using those techniques to win the success.