The COVID-19 pandemic led to closure of all educational institutions globally, forcing the transition of teaching and learning process to go online. This transition required technological innovation and digitization This study aimed to group lecturers' performance based on online pedagogical practices with the K-Means Clustering method at Universitas Negeri Semarang (UNNES) based on online pedagogical practices. The study used six variables namely course content, teaching design, video quality, teaching service, teaching evaluation, and learning effect, to measure the performance of lecturers in online learning practices in pandemic times. The data was obtained through a questionnaire from a sample of 278 lecturers at Universitas Negeri Semarang (UNNES) and analyzed using K-Means with SPSS 24.00 software. The results revealed that three lecturer performance clusters were formed during the COVID-19 pandemic based on the results of data clustering: 48 lecturers were placed in low-performance cluster (17%), 117 of them were in moderate-performance cluster (42%), and 113 of them were in the high-performance cluster (41%). The study recommends to improve lecturers' performance by joining the video-making workshops.