保險業顧客生命價值之實證研究

Customer Lifetime Value in Insurance Industry: an Empirical Study

汪芳國(Fang-Kuo Wang)

指導教授 ： 吳坤山

淡江大學/商管學院/管理科學學系博士班/博士(2013年)

https://doi.org/10.6846/TKU.2013.00932

摘要

國內保險市場的發展蓬勃，市場滲透度高居世界第一，公司家數眾多且競爭激烈，善用電腦科技挖掘有用訊息，以制定妥善的市場區隔策略，有助於穩固市場地位及提升佔有率。
本論文以國內大型壽險公司大台北地區保戶資料為基礎，計算保險業客戶價值，並提出顧客價值矩陣。再分別利用約略集理論來分析四個區隔之客戶特性，以FCA來分析各特性間之從屬關係，並提出行銷人員可利用之行銷重點。其次，再針對影響公司市場發展之鑽石級區隔及鈦級區隔，以決策樹方法深入探討不同目標下之重要影響因素，並以此二區隔客戶提出行銷策略及執行規劃。最後，針對個案公司整體市場策略提出建議。
透過研究，依顧客現在價值與潛在價值，建立2x2顧客價值區隔矩陣，分別命名為鑽石級(HH)、鈦級(LH)、銀級(LL)及黃金級(HL)等顧客群。以RST分析發現鑽石級顧客主要購買PL保單(v12)，主約保費低於$2,500(v71)；鈦級顧客群較常購買PWL保單(v15)，總保費為最高級(v95)，年收入介於58至76萬元間(v83)。以FCA執行結果，在鑽石級客群中，購買PL保單的客戶特徵為：保額介於20~50萬元間(v62)、保單繳費期間10-20年(v52)、總保費最高(v95)、但主約保費最低(v71)；鈦級顧客群中客戶總保費最高級(v95)會受下列因素影響：被保險人年齡低於24歲以下、保額介於50-100萬元間(v63)、主約保費介於5,000-7,000元間(v73)。
對公司整體市場區隔策略之建議為：1.建立更明確的品牌形象，深入都會區以外的市場；2.強化資源基礎理論；3. 進行差異化行銷策略；4. 強化客戶資料庫的功能，提升向上或交叉行銷的機會。

關鍵字

保險業 ； 顧客生命價值 ； 約略集理論 ； 形式概念分析 ； 決策樹

並列摘要

Abstract
Domestic insurance business has boomed for decades and the penetration rate is the top of world. There are many insurance companies and business is very competitive among them. It is possible to be stable or increase market share if insurers can propose marketing segregation strategies and use computer technologies to excavate available data.
This research takes customer profiles, provided by a domestic insurance company, in metro-Taipei area as the targets to discuss customer values. The themes of this research include: taking customer current and potential values as two dimensions to process customer segregation and propose customer value matrix, applying Rough Sets Theory to analyze customer characteristics in the matrix, applying FCA to analyze the subordination among customer characteristics, and proposing feasible marketing key-issues for salespersons. Using decision tree to discuss influential factors of diamond and titanium level customers under various goals, proposing marketing strategies and implementation processes are subsequent themes of this research. The last, comprehensive strategies and suggestion are proposed to the company based on depicted above.
A 2x2 customer value matrix is established based on current and potential customer value in this study and the customers in four quadrants of the matrix are segregated into Diamond (HH), Titanium (LH), Silver (LL) and Gold (HL). From the RST results, most of customers in Diamond (HH) quadrant bought PL (v12), premium of master contract is lower than 2,500 NTD (v71); customers in Titanium (LH) quadrant bought PWL, total premium (v95) is the highest among other quadrants and annual income is between 0.56 and 0.76 million NDT (v83). As the FAC results, the customer characteristics of Diamond quadrant who bought PL includes: amount of insured is between 0.2 and 0.5 million NTD (v62), the policy duration is about 10 to 20 years, the gross premium (v95) is the highest, premium of master contract is the lowest (v71); customers in Titanium (LH) quadrant who having high gross premium can be affected by following factors: the age of insured under 24 years old, amount of insured between 0.5 and 1 million NTD (v63), premium of master contract between 5,000 and 7,000 NTD (v73).
The suggested marketing segregation strategies are: 1. Build up a clear brand image to penetrate market in non-metro area, 2. Strengthen resource-based view, 3. Promote differentiated marketing strategies, 4. Enhance the customer database functions and enhance upstream/cross selling opportunities.

並列關鍵字

insurance industry ； customer lifetime value ； rough sets theory ； formal concept analysis ； decision tree

參考文獻

Wu, C.H. (2003). On the granulation simplicity for the decision rule discovery in databases: EWI vs. EFI, International Journal of Science and Technology, Vol. 14, pp. 28-36.

Beynon, M.J. & Peel, M.J. (2001). Variable precision rough set theory and data discrimination: An application to corporate failure prediction, MEGA: International Journal of Management Science, Vol. 29 No. 6, pp. 561 - 576.

Bose, I. & Chen, X. (2009). Hybrid models using unsupervised clustering for prediction of customer churn, Journal of Organizational Computing and Electronic Commerce, Vol. 19 No. 2, pp. 131 - 151.

Bruggemann, R., Voigt, K. & Steinberg, C.E.W. (1997). Application of formal concept analysis to evaluate environmental databases, Pergamum, Vol. 35, No. 3, pp. 479 - 486.

Chen, F.L. & Li, F.C. (2010). Combination of feature selection approaches with SVM in credit scoring, Expert Systems with Applications,Vol. 37, No. 7, pp. 4902 - 4909.