Identifying the direct mail-prone consumer

Abstract

Current modeling research in target marketing usually stresses the identification of profitable names for specific mailings. There is little recent research about the characteristics of typical direct mail (DM) customers. In this paper we determine the link between customers' socio-demographic characteristics and their propensity to purchase products through the mail. We hypothesize the existence of a latent variable, DM-proneness, which represents a consumer's tendency to shop via direct mail. Our model links the socio-demographics of customers to their self-reported purchase behavior through the latent variable of DM-proneness in a Multiple Indicators and Multiple Causes (MIMIC) model. We also introduce a second latent variable, DM information interest, which represents the desire to receive direct offers through the mail. The model allows for testing the influence of DM information interest on DM-proneness. We fit the model to actual consumer data using the LISREL program. The findings show that the characteristics of the DM-prone and the DM information-interested are similar, and that DM information interest appears to directly affect DM-proneness. The DM-prone consumer is identified as relatively younger, with somewhat lower education, rather well-off, from a relatively large family, and relatively more likely to have a female head of the household. We also show that consumers' tendency to shop via direct mail can be captured by a latent variable approach, and that DM-proneness can be quite well predicted by demographic consumer characteristics.

目标市场模型建造通常指的是对那些有利可图的特别邮件名称的识别。的确，每个直接市场统计人员通过对每个产品的测试分析，建立虚构的典型消费者模型。实际上，大型的数据财团如Epsilon 和Experian 提给企业邮件列表，可以让他们把大部分响应的顾客作为目标群体。但是如何使我们清楚地了解消费者，哪些消费者乐于接受广告信函消费，尤其是那些不关注卖什么的消费者。我们知道那些愿意接受信息的人并不总是能够转变为消费者。这种感兴趣和去消费是有区别的。本文通过辨别那些有倾向通过信函购物的人群的不同特征，来提升我们对典型广告信函（直销）消费者的社会背景的理解。本文的另一个目标是研究有兴趣接受广告信函的人群和有倾向通过信函购物的人群的联系。

在市场营销的文献中，倾向性被重复的应用到研究购物选择中去。倾向性简单的来说就是做某种事情的趋势。尽管直销倾向性是一个新的概念，但是许多市场营销人员已经在应用。例如，企业经常询问那些直接在线购物或者订阅博客的新顾客，他们在过去的“X”月内是否通过信函购买过任何东西。我们假设关于信函求购存在倾向性。直销倾向性就是通过信函购物的趋势。

了解哪些人是有直销倾向的在一些问题上是有优势的。设计人员可以更明确地针对直销倾向者改善直销，或者他们可以创造新内容来吸引一般来讲对直销不感兴趣的部分人群。可以基于直销倾向者的特征来设定选择的标准。市场人员拥有消费者对于企业的期望的人口统计数据，了解广告倾向性的人口统计从整体上来说可以帮助营销人员明确哪些人群可以被作为目标来增加消费者群体。

直销倾向性不能够被直接测量，并且是一个潜在的概念。我们采用Jöreskog & Goldberger (1975)的多指标多因素模型（MIMIC）来探究消费者统计结果和进行直销购物可能性的关系。

然后，本文介绍了另一个潜在概念，直销信息兴趣，是指消费者对在家收到目录和小册子的需求程度。我们假设直销信息兴趣是引起直销倾向的一个因素。我们采用MIMIC模型检验这个假设。可以通过MIMIC模型分析社会人口统计数据对直销倾向和直销信息兴趣的影响。可以比较统计数据和直销信息兴趣和直销倾向。应用LISREL项目，我们建立的模型适合真实的客户数据。

许多兴趣变量不正常。在本次研究中，我们同样介绍了一个修正异常观测变量的模型。我们同样探讨如何对待非线性变量。

根据我们的假设，我们发现直销信息兴趣对直销倾向性有影响。同时，我们发现直销倾向的顾客年轻，并且显现出较低学历和高收入。直销倾向的消费者相对更可能拥有较大的家庭人数和女性作为户主。那些可能提升直销信息兴趣的人口统计品质与那些直销倾向的没有太大差异。她们都趋向于年轻、富有来自于大家庭、相对更可能有女性户主。教育水平显现出对直销倾向性有负向影响，但是对于直销信息兴趣没有显著影响。

研究结果可以帮助市场营销人员将消费者的直销倾向进行排序。了解直销倾向消费者的统计结果可以提供简单设定目标消费者。广告部门可以根据社会人口统计标准选择地址。今后管理者和研究人员可以选择简单的方法去寻找直销倾向的目标消费者，这些标准可能是选择地址的有价值的指导。

Keywords:

• direct marketing
• consumer proneness
• structural equation modeling
• DM-information interest framework
• multiple indicators and multiple causes (MIMIC)

Keywords:

• 直销
• 消费倾向
• 结构方程模型
• 直销信息的利益框架
• 多指标多因素

Acknowledgements

The authors wish to thank Peter Verhoef, Erik Meijer and Paul Bekker for their helpful comments. They also thank the three anonymous reviewers and the editor for their helpful comments and suggestions for improvement. The authors thank Peter Verhoef, Umut Konus and Scott A. Neslin for providing access to an additional dataset to validate the results.

Notes

^1. See Bult & Wansbeek, 1995; Gönül & TerHofstede, 2006; and Simester, Sun, and Tsitsiklis, 2006.

^2. In the literature there are many studies that deal with the identification of consumers who, for example, are likely to purchase a new product (Kamakura, Kossar, & Wedel, 2004) or are potential candidates to adopt a loyalty card (Van Heerde & Bijmolt, 2005).

^3. Please note that we combine the binary indicators of DM-proneness in a simple summated score only for the descriptive statistics. When we estimate our LISREL model, we use each as a separate indicator of the latent construct of DM-proneness.