Counting Dropout Rate of Consumers: A Case Study of Durable Goods.

Abstract

Customer buying behavior research has been a special research activity in industry due to its economic, social and developmental relevance. However, summary measures such as the total buying behavior per month provides little insight about individual level of shopping behavior. This research work   attempts to predict the percentage of customers’ churn of less than 10% using the data of some packaged goods that were collected within a specific period. The computational experience provides a string statistical evidence (p-value = 0.9928) for the use of proposed beta-binominal models as an effective statistical tools.

It was estimated that at the end of thirty sixth months, only 25% of the customers is expected to achieve the targeted drop-out rate of less than 10%. In view of the industrial and commercial goal of reducing the rate of customers churn to less than 5%, the predicted percentage seems to be high as regards the packaged goods that were examined.

Keywords:         Beta-Binominal Model, Maximum Likelihood Estimation, Dropout -Rate, Consumer Buying -Behavior.