고객 정보의 종류와 양이 구매모형 예측력에 미치는 영향에 관한 연구

Abstract

The advancement of computer and telecommunicaton technology is providing numerous

ways for the marketing managers to collect and utilize consumer contact data. In other

words, marketers can now escape the limited boundaries of mass marketing channels and

begin to address individual needs of each customer, thereby realizing one-to-one marketing.

Those in academic fields have worked hard to develop models that predict the purchase

pattern and identify the causal factors. These efforts can be put into one word that has

become quite a sensation: the CRM(customer relationship marketing).

In this study, We have tried to quantitatively measure the value of customer information,

which has increasingly become important with the growing use of one-to-one marketing.

The customer information has been categorized into two, namely demographic information

and purchase history information. Different number and type covariates have been

incorporated into several models to see how they contribute to improving the goodness-of-fit of the models.

The results of the study are as follows.

First, even limited amount of demographic information about the customer can drastically

improve the goodness of fit of the models for target marketing.

Second, consumer purchase history data is more effective and better serve the purpose than

the demographic data when predicting purchase behavior.

Third, as more purchase history data are added, more accurate the model becomes.

Lastly, the marginal improvement of predictive capability of the model tends to decline as

more information is added. Therefore it can be deduced that there exists an ideal point in the

projectile of marketing budget that compromises cost and benefit.