In design thinking, empathizing and defining stages are part of UX research. The goal is to analyze pain points or complaints experienced by users using qualitative data. However, this process is always done manually, which can be time-consuming and resource-intensive. The objective of this research is to develop a system for clustering qualitative data based on problem topics using K-Means clustering and several evaluation methods, namely silhouette score, Davies-Bouldin Index, and Calinski-Harabasz Index, implemented in Python programming language and run on Google Colaboratory. User review data for the Gojek app version 4.9.3 from November 2021 to January 2024, obtained from Kaggle and preprocessed, will be used as the object for system development. Based on testing for each cluster number, the results obtained are 14 clusters or problem topics with a silhouette score of 0.65, Davies-Bouldin Index of 0.35, and Calinski-Harabasz Index of 40.7, where each evaluation method has good accuracy. The system requires a computation time of 127.4 seconds. The K-Means algorithm is effective when clustered user review data based on complaint topics. UX researchers can utilize the system from this research to assist them in analyzing pain points more quickly and efficiently.