The use of mobile banking services in Indonesia continues to increase along with the development of information technology, including the BRImo application owned by Bank Rakyat Indonesia (BRI), which has reached more than 50 million downloads and one million reviews on the Google Play Store. These reviews serve as an important data source for understanding user perceptions and experiences. This study analyzes the performance of the Ensemble Learning method for sentiment classification of BRImo reviews by combining Support Vector Machine (SVM) and Decision Tree. The data was obtained through web scraping techniques, then processed through preprocessing stages including cleaning, case folding, normalization, tokenization, stopword removal, and stemming. Next, a lexicon approach was used for sentiment labeling, while TF-IDF was used for feature extraction. The dataset consists of 8,002 reviews, split with a ratio of 80:20. The study results show that SVM achieved the highest accuracy at 92.63%, due to its strong ability to optimally separate high-dimensional text data. The Ensemble model combining SVM and Decision Tree achieved an accuracy of 89.38%, slightly lower than SVM, but still providing stable predictions. This is because the Ensemble leverages the strength of two algorithms, making it capable of reducing result variance. Meanwhile, the Decision Tree recorded the lowest accuracy at 86.45%, indicating its limitations in handling the complexity of text data. Thus, although the Ensemble does not surpass SVM, the model combination still produces a more balanced and consistent performance. This study has limitations in terms of data coverage and a lexicon approach that is sensitive to context. The findings have implications for the development of the BRImo application based on user perceptions. The novelty of the research lies in the application of the SVM–Decision Tree Ensemble in sentiment analysis of mobile banking applications in Indonesia.