Integration of Sentiment Analysis and RFM in Restaurant Customer Segmentation: A 7P-Based CRM Model with Clustering

Abstract

The increasing use of digital platforms like Tripadvisor has created opportunities to transform customer review data into strategic insights for Customer Relationship Management (CRM). This study proposes a novel CRM model by integrating the Recency, Frequency, Monetary (RFM) framework with the 7P marketing mix to segment restaurant customers more effectively. Using 3,716 Tripadvisor reviews, annotated based on 7P elements and clustered through unsupervised learning, three key customer segments were identified: acquisition, retention, and win-back. Evaluation metrics show strong clustering performance with a Silhouette Score of 0.73 and a Davies-Bouldin Score of 0.08. The acquisition cluster (Product) demonstrates the highest Frequency (37,664) and Monetary value (64.94), signifying high engagement and revenue potential. The retention cluster (Physical Evidence, Place, Process, Promotion, Traveler) shows stable interaction patterns with Recency values of 1261–1262 and moderate Frequency (378–2,079). The win-back cluster (Price, People) reflects lower Frequency (198–946) but equal Recency (1259), indicating recent but infrequent activity, which is ideal for reactivation strategies. By mapping customer reviews to 7P labels and analyzing them using RFM, the model uncovers specific behavioral patterns tied to service quality, pricing, and promotions. This integration allows restaurants to apply tailored strategies: offering loyalty rewards to high-frequency customers, promotional incentives for those with high Recency, and prioritizing high-monetary customers for exclusive programs. The novelty of this research lies in its combined use of sentiment-based review analysis and RFM–7P segmentation, offering a scalable, data-driven framework for enhancing customer satisfaction, loyalty, and long-term business growth in the restaurant industry.