A computer-implemented method overlays an alert over a pre-recorded video regarding an upcoming segment of the pre-recorded video. One or more processors receive a pre-recorded video that is directed to a main topic, and then partition the pre-recorded video into sub-topic segments, where each sub-topic segment is about a sub-topic of the main topic. The processor(s) identify previous entities' responses while previous entities viewed precursory sub-topic segments from the pre-recorded video, where the precursory sub-topic segments were presented before an upcoming sub-topic segment. The processor(s) receive a current entity's responses that occurred while a current entity viewed one or more of the precursory sub-topic segments from the pre-recorded video, then compare the current entity's responses to the previous entities' responses. In response to the responses matching, the processor(s) overlay an alert over the pre-recorded video advising the current entity that the upcoming sub-topic segment will be difficult to understand.