Instagram Sentiment Analysis for Customer Engagement Improvement in SMEs

Abstract

Small and Medium Enterprises (SMEs) increasingly rely on Instagram as a primary digital marketing channel; however, customer engagement is frequently evaluated through quantitative metrics such as likes and reach, with limited integration of emotional orientation embedded in user-generated content. This study addresses this gap by positioning sentiment analysis as an evaluative tool within digital marketing analytics for SMEs. The research aims to identify customer sentiment patterns in SME Instagram content, explain the role of sentiment analysis in assessing customer engagement, and provide a data-driven foundation for improving engagement performance. The study adopts a quantitative design using naturally occurring Instagram data, including captions and comments collected from SME accounts within a defined period. Text preprocessing is conducted prior to sentiment classification using a lexicon-based approach with the VADER (Valence Aware Dictionary and Sentiment Reasoner) model to capture polarity and intensity of sentiments in social media text. Furthermore, statistical analysis is performed using correlation analysis and one-way ANOVA to examine the relationship and differences between sentiment categories and engagement indicators (likes, comments, and interaction rate). The findings reveal systematic variation in engagement metrics across sentiment polarity categories, indicating that emotional orientation represents a relevant analytical dimension in engagement evaluation. The study contributes theoretically by reinforcing the multidimensional perspective of customer engagement and methodologically by integrating text mining techniques with engagement analytics in the SME context. Practically, it offers a replicable and scalable analytical framework that enables SMEs to incorporate sentiment-based evaluation into Instagram content strategy development.