How to edit TikTok videos people actually watch

The short answer

Short-form video is edited backwards from the first second. Everything else is subordinate to whether someone keeps watching past it.

The structure that works:

1. Second one: the hook. The result, the problem, or the strangest frame you have.
2. Seconds two to five: the promise. What they get if they stay.
3. The middle: deliver it, with a visual change every few seconds.
4. The end: a payoff or a reason to rewatch. Not a lengthy sign-off.

The opening second

Cut whatever you filmed first. Almost always.

Beginners open with a greeting, a logo, or a lead-in explaining what they are about to do. All three are places where viewers leave.

Instead:

• Show the result first, then explain how.
• Start mid-sentence if the sentence is interesting.
• Ask the exact question your video answers.
• Show the mistake you are about to fix.

Film a few different openings and try them. The rest of the edit can stay the same.

Pacing

Something has to change visually every two to four seconds: a cut, a reframe, a zoom, text appearing, a change of angle.

That does not mean chaos. It means never leaving a static frame long enough for someone to remember they were scrolling.

Practical moves:

• Cut every pause, including the small breaths. Short-form is not conversational pacing.
• Punch in on a static talking shot to create a cut where you only have one angle.
• Use speed ramps to compress boring middles rather than deleting them entirely.
• Change the frame when the subject changes.

Captions, and the rule nobody follows

Many people watch with the sound off. Captions are not an accessibility extra here, they are the content.

Auto captions are the starting point, and read every line before you post. Automatic transcription gets names, jargon and numbers wrong, and a caption that says the wrong thing is worse than no caption at all.

Keep them:

• In the middle third of the screen, out of the way of platform interface elements at the top and bottom.
• Large enough to read on a phone at arm's length.
• Short lines, two or three words at a time for fast content.
• In one consistent style, because a recognizable look is worth more than a novel one each time.

If you have ever posted a caption reading "let me show you my new pen ultimate setup", you have learned this lesson. Everyone learns it once.

Sound

Even for silent watchers, sound decides whether the people who do listen stay.

• Voice clear and loud, music well underneath.
• Trending audio helps discovery on some platforms and is not worth using if it fights your message.
• Sound effects on transitions and text appearances, used lightly, make an edit feel produced.
• Consistent volume across your posts, so people do not have to adjust each time.

What to stop doing

• Long intros. There is no room for them.
• Every effect in the library. One or two consistent looks beats a showcase.
• Filler words left in. Cut them all.
• Text that appears and disappears too fast to read.
• Ending with a long sign-off. End at the end.

Format and export

Vertical, 1080 by 1920, at the frame rate you shot. Export at high quality, because the platform compresses again on upload and starting from a compressed file compresses it twice.

Set the canvas to vertical before you start editing. Reframing horizontal footage afterwards wastes an hour and loses the sides of every shot.

Which tools

CapCut is the default for a reason: vertical templates, auto captions, effects and speed ramps in one place, on phone and desktop. CapCut's own site covers the current apps and what each plan includes.

If you are also making longer videos, DaVinci Resolve handles both, and Blackmagic's product page covers what the free version includes.

Where to learn it

The CapCut certificate covers vertical editing end to end, and the Video Editing & Motion Graphics school covers the longer-form tools when you want them.

One subscription opens all of them.

Hook in one second, change something every three, check your captions before posting.