In today's classrooms, videos are essential because they make learning more adaptable. However, a big challenge is keeping students engaged with these videos and enhancing their learning experience. This study investigates the effectiveness of instructional videos by analyzing audience retention and applying mathematical analysis. If a teaching video effectively maintains its audience, indicated by a smooth retention curve, it signifies effective engagement. This effectiveness is quantified by a low Mean Squared Error (MSE) in a linear regression model that fits the retention curve. A lower MSE means the model's predictions are closer to the actual viewer retention data, signaling a better match between expected and observed viewer behavior. We examined 14 instructional videos from a network course, employing two approaches to analyze the data: one method divides the video into equal segments (MSE1), and the other segments it based on where viewers tend to drop off (MSE2). Our findings suggest that the MSE1 method more accurately reflects actual outcomes, though it overlooks the impact of the number of segments used. Further analysis revealed a close similarity in performance between MSE1 and MSE2, despite their different calculation methods. The MSE2 approach is particularly effective at revealing shifts in student engagement, highlighting the most captivating parts of the lecture. The MSE1 approach, on the other hand, is easier to handle, good for a first look or when dealing with lots of videos. Additionally, we found that both MSE1 and MSE2 scores have a moderate negative correlation with student satisfaction, indicating that higher MSE scores may correlate with lower satisfaction levels. This information aids educators in refining their video content and improving teaching strategies.