Apartment buildings and residential buildings in Taiwan have become more and more community oriented. Precise control of residential buildings operational performance is increasingly important. Using network data envelopment analysis (Network DEA) and existing data, this study develops a three-phase based residential building operational performance assessment model and use it to assess the operational efficiency of 26 cases to obtain the performance index of cases and dimensions. The assessment results can be referred by the property management firm to establish its operation strategy and improvement. It is found that the model is equipped with three dimensions (with its weight) named staff quality (0.26), customer satisfaction (0.35) and operational efficiency (0.39) respectively. Among the various dimensions, the top three efficient cases are differentiated as: staff quality- B1, A2, A7; customer satisfaction- A8, A2, B1; operational efficiency- B1, C7, D5. Worthy to be referred to, the top three benchmark cases, in terms of overall efficient, are B1, A8, and B6 respectively. The sensitivity analysis is conducted to clarify the evaluation models input and output of various dimensions. The slack-based measures and variation rate are used to explain the influence of sensitivity. The slack-based measure is used to calculate the suitable decreasing ratio which is used to adjust non efficiency cases. It is found that A2, D8 and C8 are the top three cases. All of them have more than 100% improvement range. Among the three, the satisfaction of management committee beneath operational efficiency has the largest improvement range. Additionally, the study also uses BCG (Boston Consulting Group) matrix to analyze 26 cases. Based on the analysis results, several strategies are proposed to improve those inefficient cases.