To determine the success rate of a service, it is necessary to conduct sentiment analysis to understand public opinion and the level of public satisfaction, both positive, neutral, and negative. Sentiment analysis was used to improve the quality of service on the NewSakpole application in android-based vehicle tax payments. In this study, several stages were used such as data crawling, data preprocessing, word weighting using TF-IDF (Term Frequency Inverse Document Frequency) and SVM (Support Vectore Machine) classification model for sentiment classification. By using the Confusionsion Matrix test with a data sharing of 20% for data training and 80% for data testing, the accuracy results were obtained of 81.08%. The accuracy value of each sentiment also showed a fairly good performance, with a positive sentiment of 81.11%, a neutral sentiment of 78.62%, and a negative sentiment of 83.33%. These results show that the SVM algorithm is able to provide a fairly stable classification performance in this case study.Keywords: Sentimen Analysis; Google Play Store; NewSakpole; Support Vectore Machine AbstrakUntuk mengetahui tingkat keberhasilan suatu pelayanan, perlu dilakukan analisis sentimen untuk memahami opini publik dan tingkat kepuasan publik baik positif, netral, maupun negatif. Analisis sentimen digunakan untuk meningkatkan kualitas pelayanan pada aplikasi NewSakpole dalam pembayaran pajak kendaraan yang berbasis android. Dalam penelitian ini menggunakan beberapa tahap seperti crawling data, preprocessing data, pembobotan kata menggunakan TF-IDF (Term Frequency Inverse Document Frequency) serta model klasifikasi SVM (Support Vectore Machine) untuk klasifikasi sentimen. Dengan menggnakan pengujian Confussion Matrix dengan pembagian data 20% untuk data training dan 80% untuk data testing memperoleh hasil akurasi sebesar 81,08%. Nilai akurasi pada masing-masing sentimen juga menunjukkan performa yang cukup baik, dengan sentimen positif sebesar 81,11%, sentimen netral 78,62%, dan sentimen negatif 83,33%. Hasil ini menunjukkan bahwa algoritma SVM mampu memberikan performa klasifikasi yang cukup stabil pada studi kasus ini.