Integrating Explainable AI and the Kano Model to Derive Improvement Strategies for Essential Oils from Online Reviews

Growing consumer interest in natural wellness products, particularly essential oils, highlights the need to understand key quality product attributes affecting consumer satisfaction. In the digital era, customer reviews in marketplaces have become the main source of consumer-driven insights for improving production and service processes. However, conventional approaches often fail to systematically extract actionable insights from these unstructured data sources. This study proposes an integrated machine learning framework for three market on essential oils and their derivatives. This framework transforms thousands of online customer reviews into a structured analysis of satisfaction dimensions. The approach uniquely contributes by employing regression model combined with Explainable AI (SHAP) and KANO Classification to systematically applied based on SHAP insights to develop a marketing strategy based on three market segments for essential oil products and their derivatives. Eleven critical satisfaction dimensions were extracted, including aroma, price, packaging, delivery, and others. These segment-specific insights imply that producers should prioritize reliable pricing and delivery for low-tier markets, ensure strict price fairness and value consistency for mid-tier consumers, and, for high-tier segments, focus on integrating diffuser compatibility as a basic requirement while leveraging bonuses as emotional value-adds to enhance customer delight. Theoretically, this research introduces a scalable, Explainable AI-based approach for applying the Kano model to unstructured textual data, overcoming limitations of traditional survey methods. Despite its strengths, this study is limited by the absence of validation for the Kano categorization through survey-based procedures. Future work will address this limitation by conducting perception-based surveys or interviews to validate and refine the inferred categorizations. Nonetheless, this research contributes a methodology and provides actionable strategies for essential oil producers to align product improvements with consumer expectations in digital commerce environments.