Decoding customer satisfaction in Michelin Green Star restaurants: BERTopic modelling and sentiment analysis

Purpose The purpose of this study is to explore how green practices and other key restaurant attributes in Michelin Green Star Restaurants affect customer satisfaction by analysing online reviews. The study provides insights into electronic word-of-mouth marketing for the restaurant industry. Design/methodology/approach Evaluation of 2,320 TripAdvisor.com reviews of Michelin Green Star restaurants were conducted using BERTopic modelling and transformer-based sentiment analysis. A multistage regression framework that progresses from topic only to sentiment weighted topic prevalence was applied. Findings Food-related attributes remain the strongest predictors of customer satisfaction, followed by service and atmosphere. Green practices, while statistically significant, had the weakest effect, indicating limited customer attention to sustainability efforts beyond food-related themes. Originality/value Previous studies have analysed sustainability initiatives in Michelin Green star restaurants via content analysis of corporate websites. This research adopts a customer perspective by applying a sentiment-weighted methodology to real customer experience data, evaluating the impact of sustainable practices on customer satisfaction. These findings contribute to the sustainability literature and restaurant management practices, while also supporting the applicability of Information Integration Theory in the hospitality context.