The retail industry is continuously transforming due to digitalization, globalization, urbanization, and automatization. These factors contribute to new and more transparent retail with new customer behaviors, business models, and international competition. Literature mentions that the customers have more power to influence the retail industry today. Retailers must listen carefully to what their customers say to stay relevant and adopt data-driven strategies to meet evolving customer values. Therefore, it is essential to know customer behavior and value so that the company can target the groups of customers and thus lead their measures toward selected segments. There is a diversified growth in the tastes and likes of customers. To survive in competitive industries, a company needs to be more concerned with Customer Relationship Management. A central problem in Customer Relationship Management (CRM) is clustering customers into meaningful segments. This challenge is typical in retail, where various products are available. When it comes to segmenting customers, segmentation models provide purchasing patterns of customers, whereas clustering is the segmentation of data in several applications by grouping extensive data into groups with similar patterns. This experimental study focuses on customer segmentation, a typical customer analytics technique, and a traditional concept in marketing. The thesis aimed to perform an experiment by efficiently segmenting Swedish pharmacy retail dataset by combining segmentation models and clustering algorithms. Finally, the determined customer segments were profiled and named using a customer profiling technique. To support this aim, the research questions are “Which combination of mentioned segmentation models and clustering algorithms performs best for segmentation of a Swedish pharmacy retail company’s customers?” and “How can extracted customer segments be profiled?” This study was conducted with a Swedish retail software providing company, Extenda Retail, to gain better insight into customer behavior by analyzing experimental dataset of customer transactions of a Swedish pharmacy chain. K-means, Agglomerative and Mean-Shift clustering algorithms were paired with the segmentation models RFM, LRFM and LRFMP to generate the customer segments. This experiment showed that both K-means and Agglomerative clustering algorithms with the LRFMP model are the most suitable solution for customer segmentation for the dataset used in this study. Both combinations: the LRFMP with K-means and the LRFMP with Agglomerative clustering generated two customer segments which are then profiled as: "low-contribution customers" and "high-contributing loyal customers."
Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:su-219632 |
Date | January 2023 |
Creators | Memon, Nagma Athar |
Publisher | Stockholms universitet, Institutionen för data- och systemvetenskap |
Source Sets | DiVA Archive at Upsalla University |
Language | English |
Detected Language | English |
Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
Format | application/pdf |
Rights | info:eu-repo/semantics/openAccess |
Page generated in 0.0018 seconds