How long should a short video actually be?
Everyone wants a number. Thirty seconds, forty-five, under a minute. But the feeds do not rank you on length - they rank you on how much of your video people finish. Length only matters because it changes that.
The metric that decides everything is completion, not duration
A short-form recommendation system has one question to answer: if I show this to the next person, will they stay? The strongest available signal is what happened to the last thousand people who saw it - how far through they got, whether they looped, whether they swiped away in the first second.
This is why a 15-second clip watched to the end beats a 60-second clip abandoned at 20 seconds, even though the second one has more total watch time. Percentage is the currency. Seconds are just the denominator.
So the real question is: how long can you hold them?
Take the honest version of your content. If you have one genuinely interesting idea and it takes 18 seconds to land, make an 18-second video. Padding it to 45 to hit some rumoured ideal is how you turn a 90% completion rate into a 40% one.
The mistake runs the other way too. If your idea needs 50 seconds of setup to pay off, cutting it to 30 does not make it snappier - it makes it incomprehensible, and people leave because they are confused rather than bored.
What actually changes between the platforms
TikTok will happily push a three-minute video if it holds. Instagram Reels rewards loops, so clips under about 20 seconds that are watched twice count generously. YouTube Shorts sits inside a library where the same viewer also watches long-form, and Shorts over a minute now compete for a different kind of attention.
The practical consequence is not 'make a different length for each'. It is that the same clip can be cut at two lengths - a tight version and a fuller one - and you find out which travels, per platform, with data rather than opinion.
A rule of thumb that survives contact with reality
Cut to the shortest length that still makes sense on its own. Then check your retention graph. If it is flat to the end, you had room to say more. If it falls off a cliff at a specific second, go and look at what happens there - usually a pause, a tangent, or a sentence that repeats something already said.
That one habit - reading the drop-off point instead of guessing at durations - will teach you more in a month than any recommended number.
- Feeds rank completion percentage, not seconds
- Cut to the shortest length that still stands alone
- A flat retention graph means you could have said more
- Test two lengths of the same moment before deciding
Questions
Is there a minimum length worth posting?
Below about 7 seconds there is rarely enough room for a hook and a payoff, so completion looks great while nothing is communicated. Treat 7 to 10 seconds as a practical floor unless the clip is purely visual.
Should I post the same clip at different lengths?
Yes, and it is one of the cheapest experiments available. Cut a tight version and a fuller version of the same moment, post them a few days apart, and compare completion rather than views.
Does adding subtitles change the ideal length?
It tends to extend it. Subtitles keep people who are watching with the sound off, and those viewers are the ones who otherwise drop out earliest, so a subtitled clip usually holds a longer runtime.
Free coins every day. No card needed.
More articles
Why nobody watches past your first second
Most short videos are lost before the hook arrives. What a first second has to do, and the four openings that reliably fail.
How to turn one long video into a week of posts
A practical routine for getting eight to twelve short videos out of a single recording - without posting the same clip twice.
Burned-in subtitles or platform auto-captions?
Platform auto-captions are free but you do not control them. When that matters, and how to make burned-in subtitles that are actually readable.
When should you actually post?
There is no universal hour. What decides is when your own followers are awake - here is how to find that, and what matters more than timing.
Why nobody is watching your videos
Views are not random. Four things decide almost every flat video, and you can check all four from your own analytics in ten minutes.
Shorts, Reels, TikTok: what really changes
The same clip does not perform the same on all three. What is genuinely different - and what you can safely post everywhere unchanged.
Captions and hashtags, without the folklore
Captions do a different job on each app, and hashtags do far less than their reputation suggests. What to write, and what to stop bothering with.