AUTHOR=Chouaten Karim , Rodriguez Rivero Cristian , Nack Frank , Reckers Max 

TITLE=Unlocking high-value football fans: unsupervised machine learning for customer segmentation and lifetime value

JOURNAL=Frontiers in Sports and Active Living

VOLUME=Volume 6 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/sports-and-active-living/articles/10.3389/fspor.2024.1362489

DOI=10.3389/fspor.2024.1362489

ISSN=2624-9367

ABSTRACT=Implementing data-driven decision-making to strengthen a football club's commercial competitive position, especially against rival clubs, is increasingly important in today's competitive environment. Allocating resources to attract and retain profitable, loyal fans is key to advancing a club's marketing strategy. The commonly used Recency, Frequency, and Monetary (RFM) technique is applied across various industries to predict customer behavior, but its application in football is under-researched. This study addresses this gap by proposing a weighted RFM model, where the relative importance of RFM variables is determined using the Analytic Hierarchy Process (AHP) method. Unsupervised learning techniques are used to cluster fans based on their weighted RFM values, applying a simple weighted sum approach to estimate Customer Lifetime Value (CLV) and identify fan segments. The dataset includes 500,591 merchandising transactions, with RFM weights of 0.409 for Monetary, 0.343 for Frequency, and 0.248 for Recency. Six clusters were identified, sorted by rank, and labeled accordingly. Clusters 1 and 2, termed "Golden Fans," are characterized by high recency, frequency, and monetary value, making them crucial contributors to the club's profitability. Efforts should focus on maintaining their loyalty through special services or loyalty programs. Cluster 3, labeled "Promising," has the potential to become Golden Fans but needs to increase spending; targeted marketing and incentives can encourage this. Cluster 4, "Needs Attention," consists of former loyal fans who have decreased their engagement; strategies to re-engage them are necessary to prevent churn. Clusters 5 and 6 represent "New Fans" who show potential for growth with increased engagement and personalized offerings. Lastly, clusters 7 and 8, termed "Churned/Low Value," contribute minimally and require price incentives to re-engage, albeit with lower relative importance compared to other segments. The approach's usefulness is verified using four viewpoints and has been applied to Amsterdamsche Football Club (AFC Ajax) using Customer Relationship Management (CRM) data. This study provides actionable insights and a useful method for clustering analysis in the football industry. The findings may assist marketing practitioners in enhancing their marketing activities by implementing data-driven decision-making for effective and efficient market segmentation.