Operationalizing the Last Mile: A Sentiment Analysis Approach to Delivery Quality in Indonesian Mobile Commerce

Abstract

Last-mile delivery is a critical determinant of customer satisfaction in platform-based e-commerce. However, specific delivery failures that generate negative customer sentiment mechanisms remain insufficiently understood. Especially in rapidly expanding Southeast Asian markets, which are characterized by island geography and seasonal demand fluctuations. This study examines last-mile delivery failures and their effects on customer sentiment. With analyzing 67,600 Google Play Store reviews on Shopee Indonesia Apps collected from January to April 2026. Using keyword extraction, 12,797 delivery-related reviews were identified, of which 54.6% expressed negative sentiment. These reviews were classified into nine sub-themes of delivery failure. The results showed that the highest scores were for delivery delay (41.6%), courier misconduct (32.7%), and SPX Express infrastructure failure (32.70%), which were the most prevalent. Temporal analysis reveals that the Ramadan shopping season significantly increased negative delivery reviews, resulting in a sixfold rise between January and March 2026 and an increase in the delivery negativity rate from 42.7% to 58.3%. Drawing on the Logistics Service Quality framework, the Expectation-Disconfirmation Model, and Service Failure and Recovery Theory, the findings indicate that delivery failures are structurally compounding. Courier misconduct leads to delays, delays result in inaccurate estimated times of arrival, and these inaccuracies intensify customer dissatisfaction before recovery measures are implemented. This study extends the LSQ framework to platform-mediated logistics by introducing the concept of attribution asymmetry. Namely, by validating the use of user-generated reviews as a source of real-time data and documents the forms of delivery failures