Evaluating the effectiveness of different communication channels for market basket analysis

Abstract

Nowadays, most supermarkets are collecting more data from all operations in their businesses, enabling them to open up the possibilities of using it for data mining to extract trends and patterns thereby improving the decision making process. One way how this can be achieved is through Market Basket Analysis to derive a list of products that are usually purchased together in order to use this information in marketing strategies, such as for cross-selling in order to increase the customers’ baskets. This dissertation addresses Market Basket Analysis and Clustering Techniques for data to be utilised for the supermarkets’ benefit when it comes to targeting customers as their marketing strategy. Through a controlled experiment, three communication channels (Push Notifications, SMS and Email) have been compared in relation to targeted marketing by determining whether there is a significant difference between the channels when measured in terms of usefulness, ease of use, ease of learning, satisfaction, and information fit-to-task. The Market Basket Analysis results were used to create offers with the help of domain experts, which were then tested in the controlled experiment. Reactions to these offers were subsequently evaluated using participants’ purchasing history and baskets created in the controlled experiment. These results were then mainly evaluated by clustering the participants to obtain the expected percentages of purchase of particular item categories and comparing them to the actual percentages of purchased items in the same categories.