Why Length Is the Wrong First Question
Almost every creator asks how long a video should be, and almost nobody is helped by the answer, because length is not what any platform is judging. What is judged is whether you held the audience for the length you chose. A video that keeps people to the end of forty seconds has done better than one abandoned halfway through fifteen, even though the second is shorter.
The reason the question feels urgent is that length and retention are linked by arithmetic. Every additional second is another opportunity for someone to leave, so longer videos show lower percentages almost regardless of quality. That is why comparing a fifteen-second video and a ninety-second one on the same number tells you nothing at all.
So the useful version of the question is different: what is the shortest length that lets this particular idea land properly? Answer that and the retention takes care of itself. Answer it wrong in either direction, by padding or by rushing, and no amount of editing craft rescues it.
What Changes as a Video Gets Longer
Two things move in opposite directions. The percentage of the video watched falls, because there is more of it to abandon. The absolute seconds watched usually rise, because even a partial view of a long video can exceed a complete view of a short one. Both are true at once, and reading only the first is how creators talk themselves out of formats that were working.
Viewer intent changes too. Someone who starts a three-minute video has already made a small commitment that a fifteen-second viewer never had to make. That self-selection means longer videos often keep a more engaged, smaller audience, which is a different kind of success and worth recognizing as one.
What does not change is the opening. Whatever the length, the first seconds decide whether anything else is seen, and a video that loses people immediately fails identically at every runtime.
The Retention Target for Every Length, on Every Platform
These are the targets Retensis benchmarks a video against, arranged by length so you can find your row quickly. Within each length band the three platforms sit close together, with Instagram slightly lower throughout because it weighs saves and shares alongside watch time.
Read these as lines to aim at rather than pass marks. A single video below its target means very little. A run of videos consistently below the target for their length is the signal worth acting on, and it usually points at one of three places: the opening, the middle, or the ending.
| Platform | Video length | Aim for | Strong |
|---|---|---|---|
| TikTok | Under 15 seconds | Above 60% | Above 75% |
| YouTube Shorts | Under 15 seconds | Above 60% | Above 75% |
| Instagram Reels | Under 15 seconds | Above 55% | Above 70% |
| TikTok | 15 to 30 seconds | Above 50% | Above 65% |
| YouTube Shorts | 15 to 30 seconds | Above 50% | Above 65% |
| Instagram Reels | 15 to 30 seconds | Above 45% | Above 60% |
| TikTok | 30 to 60 seconds | Above 40% | Above 55% |
| YouTube Shorts | 30 to 60 seconds | Above 40% | Above 55% |
| Instagram Reels | 30 to 60 seconds | Above 40% | Above 55% |
Matching Length to What You Are Actually Making
Entertainment works shortest. Comedy, reactions, and satisfying videos generally belong between 7 and 20 seconds, because the payoff is the point and delay is the enemy. Stretching this kind of content past forty seconds nearly always hurts, since there is rarely enough material to fill the time without dead air, and dead air is where people leave.
Instructional content sits comfortably between 20 and 40 seconds: long enough to state the problem, deliver the answer, and add a supporting detail. Past about forty-five seconds you start losing people who already got what they came for, and that is a completed job rather than a failure. Judge this kind of video on whether people reached the part that answers the question, not on the average.
Storytelling can justify a full minute or more, but only with the structure to support it. That means something happening in the first few seconds, tension maintained through the middle, and a resolution that pays for the wait. Narrative videos without that shape lose people somewhere in the thirties, when it becomes clear the payoff is still not close.
Follow-along content is the exception to most of this. A workout or a recipe holds attention for reasons unrelated to editing, because the viewer is doing something in real time, so these can run long and still retain well.
How Long Should a YouTube Short Be?
Shorts behave much like TikToks, with one difference worth planning around: a Short can arrive from search or suggestions as well as from the feed, so some of your audience came looking for something rather than scrolling past it. That audience is more patient, which makes slightly longer Shorts viable when the subject warrants it.
The targets are the same as TikTok at every band: above 60% under 15 seconds, above 50% between 15 and 30, above 40% between 30 and 60. Completion matters more here than elsewhere, so aim for above 40% completion under 15 seconds, above 30% between 15 and 30, and above 25% between 30 and 60.
In practice, if you are cross-posting the same cut, the Short does not need to be a different length. It needs a title and opening frame that make sense to someone who found it deliberately rather than accidentally.
Finding Your Own Best Length
The targets on this page are a starting reference. Your own catalogue is a better one, and you can build it in an afternoon. Log your last twenty videos with their length, format, and retention, then group them into bands. The pattern is usually obvious immediately, and it is often not the one you expected.
Then test deliberately instead of assuming. Take one concept and publish it at two lengths, a week or two apart so they are not competing, and compare what happens. That single comparison tells you more about your audience than any published range can, including this one.
What you are looking for is the point where extra runtime stops paying for itself. Every creator has one, it differs by format, and knowing yours is worth more than knowing the platform average. Retensis's retention analysis scores each video against the target for its own length, which makes that boundary visible without the spreadsheet.
Frequently asked questions
As long as the idea needs and not one second longer, which in practice means 7 to 20 seconds for entertainment, 20 to 40 seconds for something instructional, and up to a minute when there is a story with a payoff worth waiting for. Length is not the thing being judged. Whether you hold the audience for whatever length you chose is the thing being judged, and the target falls as the video gets longer: above 60% under 15 seconds, above 50% between 15 and 30, above 40% between 30 and 60.
Roughly 7 to 20 seconds for entertainment and 20 to 40 seconds for how-to content, which is the same shape as TikTok. The targets sit slightly lower on Reels: aim for above 55% view-through under 15 seconds, above 45% between 15 and 30, and above 40% between 30 and 60. That is deliberate rather than a quirk of measurement, because Instagram weighs saves and shares alongside watch time.
Not directly. What happens is that longer videos have more opportunities to lose viewers, so their retention percentage falls, and retention is what distribution responds to. A three-minute video that genuinely holds people can travel a long way. A three-minute video padded out from a thirty-second idea will not, and the padding is the problem rather than the runtime.
No, and this is where the advice usually goes wrong. Cutting a video below the length its idea needs produces something that feels rushed and unsatisfying, which shows up as poor completion and almost no shares. The rule that actually holds is to remove what does not serve the hook, the content, or the payoff, then keep whatever remains, however long that turns out to be.
Find out exactly where viewers drop off in your videos
Upload any video and see your predicted retention curve in 90 seconds. Spot drop-off points before they cost you views. Free to start.
See your retention curve free →