Most of the studies in identifying the factors contributing to high academic performance of students focus on both traditional and blended learning environments. This study explores the online engagement behaviours of high-performing online students. Three years of data derived from students' online activities and academic performance across six computing courses was analysed. Our findings reveal a continuity of high-performance among students who excel in online introductory programming courses, extending their success to subsequent online computing courses. Furthermore, we establish a positive association between programming experience and improved performance. High-performing s tudents exhibit engagement levels that exceed their peers by at least 100% in o nline formative learning activities, with particular emphasis on forum participation and quiz contribution. This study also looked into the predictive potential of past academic performances in anticipating future achievements. It demonstrated a predictive accuracy of 75% for programming courses and 91% for non -programming ones. When combined with student engagement data, the predictive accuracy for programming courses increases to 81%. This approach requires substantial data until at least week 7 or 8 of a 10 -week course, potentially delaying intervention until the latter stage of the course. To address this, exploring models exclusively reliant on engagement data is recommended. Our research offers actionable insights for educators, enabling them to identify critical content design elements that enhance academic performance across diverse student profiles, fostering improved online engagement and narrowing performance gaps . Moreover, this study calls for a re-evaluation of learning analytics tools within educational institutions to better understand and enhance student behaviours and patters, ultimately improving the overall online learning experience.