Hidden service climate archetypes in hotels: A text-analytic framework using employee narratives

Abstract

This study investigates how hidden backstage organizational climates in hospitality settings shape latent service performance risks by leveraging digital employee voice data that are typically inaccessible through traditional survey methods. Drawing on 11,336 Glassdoor reviews from 378 hotels, we apply an integrative text-analytic framework combining structural topic modelling, sentiment analysis, and burnout lexicon matching to extract climate-related themes, emotional tone, and psychological strain embedded in employee narratives. These signals are integrated into a composite backstage failure index and used to identify distinct service climate archetypes. Importantly, the objective is not to assign individual hotels to fixed categories, but to uncover the latent structure through which hotels can be meaningfully grouped based on employee-generated narratives. The analysis reveals four empirically grounded archetypes—Burnout/Overload, Toxic Leadership, Career-Development, and Bureaucratic/Communication-Impaired Hotels—that differ systematically in their latent failure burden. By conceptualizing service climate as an emergent narrative regime rather than a purely perceptual survey construct, this study demonstrates how unsolicited employee-generated text can function as a diagnostic sensor of organizational climate. The findings extend hospitality climate research by offering a scalable, data-driven approach for identifying hidden internal risks and provide actionable insights for managers seeking to monitor, diagnose, and intervene in backstage service system failures.