Every LMS ships a video player; almost none ship attention. The median completion rate for course video sits near 40%, and the drop is not random — it clusters at the exact moments where cognitive load spikes and the student's thumb reaches for another tab. Timestamp quizzes attack that moment directly: you set a point inside the video, attach a question, and playback pauses until the student answers. Mandatory questions go further — skip or close the video, and it simply will not continue. The result is a small behavioral contract enforced by the player itself, not by the student's willpower.
Where to place questions: the 4-7-12 rhythm #
Placement is a pacing decision, not a coverage decision. The pattern that consistently moves completion metrics follows a rhythm: one mandatory question around minute four (before the first attention cliff), another at the seven-to-eight minute mark (where working memory saturates for dense material), and a final comprehension check near the twelve minute mark or video end. Cram a question every ninety seconds and you train students to resent the video; space them wider than fifteen minutes and you are back to passive watching. The four-minute anchor matters most because it arrives after the student has committed context but before they have decided the video is skippable — an answer given there is a micro-investment that raises the psychological cost of abandoning the session.
What to ask: recall that serves the next sixty seconds #
The question type determines whether the pause feels like a checkpoint or an interruption. Ask about what was just explained — a definition, a step order, a number the student heard ninety seconds ago — because retrieval practice at short intervals is what cements encoding. Avoid trick questions, avoid trivia, and avoid questions that require information from later in the course; a pause the student cannot answer becomes a punishment, and punished students close videos. Two to four options with one defensibly correct answer keeps the interaction under fifteen seconds. If your video demonstrates a procedure, one well-placed "what happens if you skip this step?" question outperforms five recall questions, because it forces the learner to model the system rather than memorize the narration.
- Minute 4, mandatory: one-question recall of the framing concept — the "why you are watching" check.
- Minute 7-8, optional: application question tied to the step just demonstrated.
- Video end, mandatory: one comprehension question whose answer the next lesson depends on.
- Never more than three pauses per video — the ceiling before engagement inverts.
Reading the analytics the pauses produce #
Once questions sit inside videos, your analytics stop lying to you. Watch time alone cannot distinguish a student from a paused tab; a per-question answer stream can. The pattern to watch is wrong-answer clustering: if 60% of students miss the minute-four question in lecture six, the problem is almost never the students — it is the ninety seconds of explanation before it, and that is the exact segment to re-record. Question-level data also predicts refund risk weeks before support tickets appear: students who skip mandatory questions across three consecutive videos disengage shortly after, which gives you a retention window to intervene with a message, a live session invitation, or a simpler supplementary video. Treat the timestamp quiz layer as instrumentation first and enforcement second — the enforcement creates the data, and the data tells you which parts of the course to fix.
The best interaction design is invisible until the moment it is needed — then unavoidable.