Predicting variation in repertoire size with the NBD model

• We use the NBD model to predict the variation in repertoire sizes.
• We validate the model with multiple datasets from UK consumer packaged goods.
• Repertoire size varies more among heavy than light category buyers.
• The variance of repertoire size increases as time increases.
• A simple method to predict brand loyal and multiple brands buyers is proposed.

Chinese abstract用NBD模型预测全部的尺寸变化本文提出并验证负二项分布，预测在一个类别内全部尺寸(在特定的时间内一名消费者购买产品的品牌数量)的变化。从管理者的角度来说，这些变化对品牌经理希望了解其品牌所面临挑战的本质和强度至关重要。根据针对来自英国消费性包装商品的多个数据集的研究结果表明，NBD模型在不同的时间段(1年、18个月和3年)以及不同的购买者类别(少量和大量)对全部尺寸变化的预测非常理想。然后本文提出一种简单的方法，只需少量的信息(例如产品类别及品牌渗透率)以及使用NBD模型的参数便可预测哪些人只忠诚于一个品牌，哪些人会忠诚于几个品牌，以及哪些人会选择多元化地购买多个品牌.

This paper proposes and validates the negative binomial distribution (NBD) to predict the variation in repertoire size (the number of brands purchased by a consumer in a specific time period) within a category. From a managerial perspective, the variation is crucial for brand managers who would like to know the nature and intensity of competition that is facing their brands. Empirical findings across multiple datasets from UK consumer packaged goods demonstrate that the NBD model predicts variations in repertoire size very well in different time periods (1 year, 18 months and 3 years), and different buyer groups (light and heavy category buyers). The paper then suggests a simple method to predict those who are brand exclusively loyal, those who are loyal to a few brands, and those who are diversified to buy multiple brands, based on very little information such as category and brand penetrations, using the parameters of the NBD model.