["British Journal of Educational Technology, EarlyView. ", "\nAbstract\nInstructional videos increasingly supplement formal education, yet the efficacy of design elements (e.g. pre‐questions and feedback) remains underexplored, particularly regarding interactions with learners' prior knowledge. This study involving 352 Chinese university students employed a 2 (prior knowledge: low/high) × 4 (instructional intervention conditions: no pre‐questions; pre‐questions without feedback; pre‐questions with simple feedback; pre‐questions with elaborated feedback) experimental design, using eye‐tracking to measure attention allocation (time to first fixation, dwell time, total fixation duration) and analysing cognitive load, learning efficiency, retention and transfer via two‐way ANOVA. Research has indicated that intrinsic cognitive load was mostly influenced by existing knowledge and was not significantly impacted by the instructional intervention. Extraneous cognitive load and germane cognitive load showed a prior knowledge × intervention interaction: under low prior knowledge, pre‐questions with elaborated feedback yielded lower extraneous load and higher germane load, whereas no pre‐questions yielded higher extraneous load and lower germane load; under extensive previous knowledge, no pre‐questions or simple feedback produced lower extraneous load without suppressing germane load, while elaborated feedback increased extraneous load due to redundancy. Pre‐questions universally reduced time to first fixation, indicating quicker attention capture, while simple feedback increased dwell time on critical content, fostering deeper engagement. Retention was highest for low‐knowledge learners using pre‐questions without feedback, whereas transfer performance depended more on prior knowledge than instructional design. Findings suggest tailoring videos to learners: Novices benefit from pre‐questions with simple or no feedback, while experienced learners gain equally from simpler designs without needing elaborated feedback. Pre‐questions effectively guide attention, making them ideal for scalable online education.\n\n\nPractitioner notes\n\nWhat is already known about this topic\n\n\n\nPre‐questions embedded in instructional videos can function as cognitive scaffolds that direct learners' attention to critical content and improve academic performance.\n\nFeedback enhances teaching and learning when properly designed, though its effectiveness varies by type (e.g. simple vs. elaborated feedback) and learner characteristics (e.g. learners' prior knowledge).\n\nEye‐tracking technology has been widely employed to investigate attention‐allocation mechanisms in multimedia learning. By recording metrics such as time to first fixation, percentage dwell time in areas of interest, and total fixation duration, researchers can assess how much learners focus on key information and how this relates to learning performance.\n\n\n\n\nWhat this paper adds\n\n\n\nEye‐tracking data reveal the universal effectiveness of pre‐questions in video learning, particularly in significantly reducing first‐fixation time and extending total fixation duration.\n\nInstructional strategies interact with prior knowledge: For low prior knowledge learners, pre‐questions without feedback or with simple feedback optimize learning efficiency and reduce cognitive load, whereas elaborated feedback may cause cognitive overload; for high prior knowledge learners, the performance differences across instructional intervention conditions are minimal.\n\n\n\n\nImplications for practice and/or policy\n\n\n\nFor practitioners, feedback should be matched to learners' prior knowledge and current performance. A graduated scheme is recommended: default to knowledge of results (KR)/knowledge of correct response (KCR) and layer brief, on‐demand explanatory cues only when needed. Provide concise corrective and key‐concept prompts for low‐knowledge or unstable responders, and succinct confirmation for high‐knowledge or stable performers. In practice, this argues for adaptive, profile‐matched feedback rather than unguided learner self‐selection, which may help manage cognitive load (e.g., lowering extraneous load and supporting germane processing).\n\nIn large‐scale online courses and other scenarios where personalized teaching is challenging, it is recommended to set pre‐questions as a default component. This design can not only adapt to learners with different knowledge levels but also ensure that all learners benefit even without personalized guidance.\n\nWhen using instructional videos in class, teachers can refer to the findings by embedding pre‐questions before key concepts, thereby capturing students' attention and promoting deeper learning.\n\n\n\n\n\n"]