Automated Analysis of Causal Relationships in Customer Reviews

Abstract

Abstract As the volume of unstructured data on social media continues to grow, it's becoming increasingly important to have a proactive marketing strategy that can extract knowledge from this data. This study explores the use of ChatGPT for detecting causal relations and analyzing significant themes in order to build models for marketing analysis. Using 400 sample reviews and contemporary techniques, a causal graph was synthesized and tested, showing good model fit. All paths in the causal network were found to be significant except for the one from "Customer experience" to "Customer Advocacy." The system identified three serial mediators: "Exceptional hospitality" ➔ "Quality lodging" ➔ "Customer experience" ➔ "Enjoyable time" ➔ "Customer Advocacy," with an effect size of 0.0106. This research highlights the potential of linguistic data for developing mathematical models in marketing research and expands the scope of scientific inquiry in this field.