PERANCANGAN MODEL SEGMENTASI PELANGGAN B2B PT XY DENGAN CUSTOMER LIFETIME VALUE MENGGUNAKAN METODE DATA MINING

• Main Author: Laras Hanifah, Puti
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/84146
• Institution: Institut Teknologi Bandung

PT XY is an information and communication technology (ICT) services company that implements Business to Business (B2B) model. PT XY's efforts to maintain relationships with their customers is by creating an account plan. Account plan is a strategy developed by PT XY’s account managers (AM) to ensure customer retention by encouraging them to continue using PT XY's services and products and keep renewing their contracts. The goal of creating an account plan to improve customer retention, keep existing customers, and expand sales potential. However, PT XY is experiencing sales issues that did not meet the desired target throughout 2023, along with a declining trend of enterprises customer amount from 2021 to 2023. This problem emerged because PT XY did not use their customers’ data to determine customer segmentation and the appropriate account plan strategies for each segment. As a result, the purpose of this research is to design a customer segmentation model for PT XY using data mining methods based on customer transaction data, with the hope of improving the effectiveness of PT XY's account plan strategies. The customer segmentation model in this research is made using the CRISP-DM methodology and clustering as one of data mining techniques, based on the Recency, Frequency, and Monetary (RFM) model, with segments that are categorized based on Customer Lifetime Value (CLV). The modeling process are carried out with cluster analysis based on RFM variables using algorithms such as k-means, CLARA, agglomerative, DBSCAN, OPTICS, spectral, and EM. Based on the performance evaluation of the clustering models, the CLARA algorithm produces the best clusters, and so the results of the CLARA clustering are used to calculate the CLV values. Customers are then categorized into segments based on their CLV values. The outcome of this research is a customer segmentation model prototype designed using a shiny web app on RStudio software.