TikTok Video Editing
built for the first 3 seconds

TikTok's ranking model weights completion rate and rewatches more heavily than likes or shares, so the first signal it collects is whether viewers who saw the first frame are still there at frame 90 (three seconds at 30fps). A video that loses half its audience before the three-second mark gets throttled out of the initial test pool of a few hundred accounts before it ever reaches a wider For You distribution ring. Average watch time as a percentage of total runtime, not raw seconds, is the metric that decides whether the algorithm escalates you to the next audience tier.

Session-continuation behavior matters almost as much as on-video metrics: if a viewer scrolls away and doesn't open the app again for several minutes, that session gets logged as a negative outcome attributed to your video. Loops (the viewer watching to the end and the video restarting without a swipe) are read as a completion-plus signal and are one of the few levers editors can pull deliberately by matching the final frame's action or audio cue to the opening frame.

The hook window is the first 0.5 to 1.5 seconds of visible motion, not the first line of dialogue; a static talking-head open with no camera movement, text, or visual change in that window is the single most common reason a TikTok underperforms. Cut density on native TikTok content typically runs one new shot, camera move, or on-screen text change every 1.5 to 3 seconds for talking content and every 0.5 to 1.5 seconds for montage or b-roll-driven edits, which is noticeably faster than YouTube pacing.

Captions are functionally mandatory: burned-in text should sit inside the safe zone and mirror spoken audio nearly word for word rather than summarizing it, since a large share of the audience watches muted in public or commuting contexts. B-roll ratio for talking-head content runs 20 to 40 percent coverage over the base shot, inserted to reset visual attention rather than to illustrate every claim literally.

Export at 1080x1920 (9:16), the platform's native full-bleed vertical frame, using H.264 in an MP4 container with a target video bitrate of 8 to 12 Mbps for 1080p at 30fps; TikTok re-compresses on upload regardless, so delivering above roughly 16 Mbps mainly wastes upload time without a visible quality gain. Frame rate should match source (commonly 30fps; 24fps content should not be force-converted, which introduces judder).

Loudness should be normalized to approximately -14 LUFS integrated with a true peak ceiling at or below -1 dBTP, matching the platform's internal normalization pass so dialogue doesn't sound quiet next to other creators' audio. Keep primary text and faces inside a safe zone roughly 130px from the top and 250px from the bottom edge at 1080x1920 to clear the caption bar, profile info, and interaction icons on the right rail (roughly the right-hand 100 to 150px). Native captions can be uploaded as SRT if you're using TikTok's caption tool rather than burning text in, which allows viewers to toggle them off.

The most frequent mistake is repurposing a horizontally-shot interview by center-cropping to 9:16, which either crops out gesturing hands or leaves distracting negative space above the subject's head; the fix is shooting a dedicated vertical pass or using a zoompan-style reframe that tracks the speaker rather than a fixed crop. A second common error is opening with a logo, intro animation, or slow zoom-in before any content starts, which burns the entire hook window on branding.

Teams also frequently mistime captions to appear a full sentence behind the audio because they're generated from a rough transcript and never resynced to the final cut; captions should be re-timed to the picture edit, not the original recording. Finally, exporting at 720p to save render time measurably softens on-screen text legibility on larger phones and should be avoided in favor of full 1080x1920 delivery.

Efficient TikTok production works from a single briefed shoot day or long-form source (podcast, webinar, interview) broken into a topic map before editing begins, so an editor can pull five to ten standalone hooks rather than searching for moments reactively. Clips are typically cut in batches of five to fifteen per source, versioned with a naming convention that tracks hook variant (v1, v2) so a low performer can be recut with a new opening line without re-editing the whole timeline.

A realistic turnaround for a batch of ten TikTok cuts from existing footage is two to four business days including one round of revisions; same-day turnaround is achievable for reactive or trend-timed content but should be scoped separately since it compresses the review cycle to a single pass. High-performing cuts are commonly re-versioned into Reels and Shorts with re-timed captions and adjusted safe zones rather than reused as-is, since safe zones and caption placement differ slightly across the three platforms.