The intended rise in Indonesia’s value-added tax (VAT) rate to 12% by 2025 has ignited considerable public debate on social media, especially on Instagram. This research performs a sentiment analysis of public reactions to the policy by examining comments on the Narasi Newsroom Instagram post through text mining methods. A collection of 1634 comments was obtained through the Instagram Graph API and underwent data cleansing, tokenization, and sentiment analysis using the VADER model. Statistical analyses, both descriptive and inferential, were conducted to evaluate sentiment distribution and levels of engagement. The results indicate that neutral sentiment prevails (598 comments), with positive (629) and negative (407) sentiments trailing behind. Negative feedback mainly raises worries regarding the policy’s effects on middle- and lower-income populations, whereas positive remarks emphasize gratitude for the focused application aimed at luxury items. Topic modeling reveals that conversations predominantly center around “luxury,” and “goods.” The sentiment classification model, developed with a Logistic Regression and Random Forest algorithm, reached an accuracy of 75% and 71%, although effectiveness differed among sentiment categories. These insights aid in formulating data-driven policies and public communication approaches that enhance public acceptance.