Reading between the lines: analyzing online reviews by using a multi-method Web-analytics approach

Purpose The dynamic yet volatile nature of tourism and travel industry in a competitive environment calls for enhanced marketing intelligence and analytics, especially for those entities with limited marketing budgets. The past decade has witnessed an increased use of user-generated content (UGC) analysis as a marketing tool to make better informed decisions. Likewise, textual data analysis of UGC has gained much attention among tourism and hospitality scholars. Nonetheless, most of the scholarly works have focused on the singular application of an existing method or technique rather than using a multi-method approach. The purpose of this study is to propose a novel Web analytics methodology to examine online reviews posted by tourists in real time and assist decision-makers tasked with marketing strategy and intelligence. Design/methodology/approach For illustration, the case of tourism campaign in India was undertaken. A total of 305,298 reviews were collected, and after filtering, 276,154 reviews were qualified for analysis using a string of models. Descriptive charts, sentiment analysis, clustering, topic modeling and machine learning algorithms for real-time classification were applied. Findings Using big data from TripAdvisor, a total of 145 tourist destinations were clustered based on tourists’ perceptions. Further exploration of each cluster through topic modeling was conducted, which revealed interesting insights into satisfiers and dissatisfiers of different clusters of destinations. The results supported the use of the proposed multi-method Web-analytics approach. Practical implications The proposed machine learning model demonstrated that it could provide real-time information on the sentiments in each incoming review about a destination. This information might be useful for taking timely action for improvisation or controlling a service situation. Originality/value In terms of Web-analytics and UGC, a comprehensive analytical model to perform an end-to-end understanding of tourist behavior patterns and offer the potential for real-time interpretation is rarely proposed. The current study not only proposes such a model but also offers empirical evidence for a successful application. It contributes to the literature by providing scholars interested in textual analytics a step-by-step guide to implement a multi-method approach.