Is greener always better? The nonlinear relationship between hotel green certification ratings and consumer sentiment

Purpose This study examines the relationship between green certification ratings and consumer sentiment toward green hotels and investigates whether this relationship is nonlinear. Design/methodology/approach Drawing on sustainable consumption theory (SCT), the triple bottom line (TBL) framework and the too-much-of-a-good-thing (TMGT) perspective, a two-stage approach was adopted. TF-IDF-enhanced LDA2Vec topic modeling and sentiment analysis were applied to 30,247 reviews from 12 green-certified hotels in Hong Kong to extract sustainability-related service themes and measure consumer sentiment. Review-level multilevel modeling, supplemented by theme-specific nonlinear analyses and robustness checks, was used to examine the certification–sentiment relationship while accounting for the nesting of reviews within hotels. Findings Results reveal a non-monotonic association between green certification ratings and consumer sentiment. Consumer sentiment is more favorable at intermediate certification ratings but declines at the highest certification rating, producing a pattern directionally consistent with an inverted-U relationship. Theme-specific analyses further reveal significant inverted-U associations for Service, Facilities and Hygiene, while formal cross-theme interaction tests indicate substantial heterogeneity across experiential dimensions. Practical implications Hotel managers should balance sustainability initiatives with guest experience rather than pursuing higher certification ratings indiscriminately. Sustainability initiatives may be more favorably evaluated when they are translated into tangible and experience-relevant service attributes. Originality/value This study extends sustainability signaling research by identifying an important nonlinear boundary condition in the association between green certification ratings and consumer sentiment. It further advances TBL research by demonstrating that certification–sentiment relationships vary across sustainability-related experiential dimensions. Methodologically, the study illustrates the value of integrating topic modeling, sentiment analysis, multilevel analysis and nonlinear modeling to investigate sustainability-related consumer evaluations using large-scale user-generated content.