In recent years, COVID-19 policy measures massively affected university teaching. Seeking an effective and viable way to transform their lecture material into asynchronous online settings, many lecturers relied on prerecorded video lectures. Whereas researchers in fact recommend implementing prompts to ensure students process those video lectures sufficiently, open questions about the types of prompts and role of students' engagement remain. We thus conducted an online field experiment with teacher students at a German university (N = 124; 73 female, 49 male). According to the randomly assigned experimental conditions, the online video lecture on topic "Cognitive Apprenticeship" was supplemented by (A) notes prompts (n = 31), (B) principle-based self-explanation prompts (n = 36), (C) elaboration-based self-explanation prompts (n = 29), and (D) both principle- and elaboration-based self-explanation prompts (n = 28). We found that the lecture fostered learning outcomes about its content regardless of the type of prompt. The type of prompt did induce different types of self-explanations, but had no significant effect on learning outcomes. What indeed positively and significantly affected learning outcomes were the students' self-explanation quality and their persistence (i.e., actual participation in a delayed posttest). Finally, the self-reported number of perceived interruptions negatively affected learning outcomes. Our findings thus provide ecologically valid empirical support for how fruitful it is for students to engage themselves in self-explaining and to avoid interruptions when learning from asynchronous online video lectures.