Jump cut
A jump cut is a video edit that removes time between two similar shots, creating a visible discontinuity in the speaker's position or framing. In short-form video, jump cuts compress speech (cutting out pauses, filler words, slow segments) to maintain the fast pacing the format demands. A well-executed jump cut every 3–5 seconds keeps retention high.
Jump cuts violate traditional continuity editing rules — they're intentionally disruptive. In long-form film, this is a craft choice; in short-form, it's the default. TikTok and Reels viewers are trained to expect 3–5 second pacing. A 30-second clip with no cuts feels slow regardless of content quality.
The technique is straightforward: edit out every "um", "uh", every breath, every pause longer than 0.3 seconds, every restatement. What remains is the speaker delivering content at unnaturally fast cadence — which reads as confidence and competence on short-form. The visible jump on each cut becomes a feature, not a bug; the brain treats each cut as a reset point that re-engages attention.
AI clip generators handle some jump-cut compression automatically (silence removal, filler-word detection). The human pass after AI selection should add additional jump cuts where the speaker restated themselves or got off-track. Clipzing.com includes filler-word detection in its caption editor.
Related terms
B-roll
B-roll is supplementary footage cut into a video to support or illustrate the main A-roll (the speaker on camera). In short-form video, B-roll is typically 1–4 second inserts of relevant imagery, gameplay, screen recordings, or stock footage that visualize what the speaker is describing. Used well, B-roll boosts retention; used to cover the punchline, it kills replay.
Viral hook
A viral hook is the first 1–3 seconds of a short-form video clip — the line, image, or question designed to stop the thumb-scroll and earn the next 5 seconds of attention. In 2026, hooks that lead with a specific number, a stand-alone declarative sentence, or a contradiction outperform vague openers by an order of magnitude in retention.
Watch time
Watch time is the cumulative duration viewers spend watching a video, expressed either as average watch duration per view or as percentage retention curve across the clip's length. For short-form algorithms (TikTok, Reels, Shorts), watch time is the dominant ranking signal — clips with high retention get distributed to more feeds regardless of like/comment counts.
Auto-captions
Auto-captions are AI-generated subtitles that transcribe spoken audio into on-screen text in real time during video playback. In short-form video, auto-captions are mandatory — roughly 85% of social feed playback happens with sound off, and uncaptioned clips lose retention immediately. Modern auto-caption quality on English audio runs 95%+ accurate.