How long should a short be? The retention math
"Shorter is better" and "longer earns more" are both true, which is why the question never dies. Here's the arithmetic the feeds actually run, what each platform's length rules change, and a way to pick a length per video instead of per superstition.
Click to enlarge
Platforms judge a short mostly by how much of it people watched — as a share, not in seconds — plus whether they watched it again, shared it or followed. That one fact settles most length arguments.
The ratio, not the clock
Two videos, same clip:
- A 9-second cut watched to the end: completion 100 %, and a good chance of a replay, which counts again.
- A 40-second cut where people leave at 0:20: completion 50 %, no replay.
Same footage, same viewers, and the first one tests far better — even though the second one delivered more watch seconds. Retention is a fraction. Every second you add has to earn its place or it lowers the score of everything before it.
That's the rule of thumb: a video should end at the last moment that still pays for itself. Not at a round number.
When longer wins anyway
- Payout programs. TikTok's Creator Rewards wants videos over a minute; YouTube's monetization counts engaged views and watch time (TikTok, YouTube Shorts). A page built on platform payouts makes longer videos on purpose and accepts a lower completion rate — a business decision, not a retention one.
- Stories that need the time. A 45-second clip that holds viewers to the end beats a 15-second clip that doesn't. Length isn't the enemy; dead time is.
- Search-driven platforms. Pinterest and YouTube's search tail reward "the full answer" over the fastest cut.
"longer" is a business decision, never a retention one. Decide it before the cut, not while editing — a 15-second idea stretched to 61 seconds loses both arguments.
When shorter wins
- The first test. New pages, new formats, cold audiences: a tight clip gets a clean yes/no from the test group (stuck at 200–300).
- Loops. Under ~12 seconds a clean loop gets replays for free.
- Reply-driven feeds. X, Threads and Bluesky skim; the clip is the proof for the line above it, not the main event.
Per-platform reality check
One render rarely fits all: a 90-second piece is fine on TikTok and over the line on X and Snapchat. These are the conservative ceilings VidVertex's pre-flight warns against before a run — the platforms' own rules move, so treat them as a batching guide, not gospel.
| Platform | Duration the pre-flight accepts |
|---|---|
| TikTok | up to 10 minutes |
| Instagram Reels | up to 15 minutes |
| YouTube Shorts | up to 3 minutes — longer is not treated as a Short |
| Facebook Reels | 3–60 seconds |
| X (Twitter) | up to 140 seconds |
| Snapchat Spotlight | 5–60 seconds |
| Threads | up to 5 minutes |
| Bluesky / Pinterest | up to 3 minutes / up to 15 minutes |
A shorter cut in its own stack, or a per-channel output template, solves the overflow cleanly (rendering templates). Details per platform in the post formats hub.
Picking a length per video
- Find the end. Watch your own cut and mark the moment the point is made. Cut there. Then cut two seconds earlier.
- Find the start. If the best moment is at 0:08, the video starts at 0:08 (hooks).
- Remove the air. Pauses between sentences, breaths, "so, yeah". Silence-cutting does this mechanically — VidVertex's silence cut trims spoken pauses out of the main clip before compositing, and the captions follow (captions) — and it's the single biggest retention gain for talking-head clips.
- Check the graph. After a week, the retention curve tells you where people leave. If it's a cliff at the end, you ran long. If it's a cliff at the start, length wasn't the problem.
- Decide the business case separately. If the page is built for a payout program, set the floor (over a minute) and make the content earn it; don't stretch a 15-second idea.
The length the platform judges is the rendered length. A hook overlay costs nothing, but a card hook, an endcard and an outro clip each add real seconds in front of and behind your cut.
Two numbers worth remembering
Completion rate and average watch time are the two metrics every platform shows. Read them together: high completion with low watch time means your cut is tight and might afford a few more seconds; low completion with high watch time means a long video with a weak ending. Everything else about length follows from those two.
THE CRAFT · 3 MINWhy burned-in captions lift retentionA huge share of feed scrolling happens with the sound off. Captions aren't decoration — they're the sound-off version of your video.THE CRAFT · 4 MINHooks: the first two secondsMost videos don't fail in the middle. They fail before they start — in the two seconds where a scrolling thumb decides whether you exist.START HERE · 3 MINReading your analytics: the retention curve is the bossEvery platform hands you a dashboard full of numbers. One graph matters more than all the rest together — and most people never open it.