“How long should it be?” is the first question almost everyone asks about Shorts, and the honest answer is: long enough for the moment to make sense, and not one second longer. That sounds like a dodge, so in this post I’ll explain the default I actually use, 30–45 seconds, where it came from, and when I ignore it.
What YouTube allows now
The limit isn’t really the constraint any more. According to YouTube’s help pages, Shorts can now be up to three minutes long, and square or vertical videos up to three minutes are treated as Shorts. So you have plenty of room.
Having room is the problem. When the ceiling was low, the platform made the decision for you. Now an editor, or a clipping tool, has to decide, and the easy mistake is to fill the space because it’s there.
One more platform detail worth knowing: YouTube changed how Shorts views are counted in 2025, so that a view counts when a Short starts playing or replays. That makes the view count a looser signal of whether people actually watched. It doesn’t tell you the right length on its own, which is another reason I lean on what the moment needs rather than on chasing a number.
How I landed on 30–45 seconds
I didn’t start there. In one of my first pilots, for an anime fight-analysis channel, I cut the first batch at roughly 11 to 23 seconds. My thinking was that short means punchy, and punchy means people stay.
The creator’s feedback was immediate: too short. The clips had lost their setups. His format is commentary: he builds a little argument, then lands the joke or the verdict. At fifteen seconds you get the verdict without the argument, and it just sounds like someone shouting a name.
For the second batch I cut every clip to 30–45 seconds and kept the context in. His review said that range was right, that anything under 30 was too short, and that in the past he’d gone longer than 45 and felt it was too much. That’s one creator’s preference, not a law. But it matched what I saw in the clips themselves: that window is long enough for a setup, a turn and a payoff, and short enough that nothing has to be padded.
I also compared my first batch against an AI clipping tool on the same video. Its clips ran from about 15 to 73 seconds, mostly in the high 30s to around a minute. I liked my own picks and packaging better, but on length the tool was closer to what the creator wanted than I was. I took that lesson on board for the next batch.
What the extra seconds actually buy
When I say 30–45 seconds “buys” you something, this is what I mean:
- The setup. One or two sentences that tell a stranger what’s going on. Without them, the payoff has nothing to land on. I wrote more about this in how to find moments that work without the full episode.
- The turn. The surprise, the disagreement, the moment the story changes direction. This is usually what makes people stay past the first few seconds.
- The payoff with room to breathe. If the payoff is a laugh, you need a beat after it. If it’s a statement, you need the whole sentence, not the first half.
- A clean ending. The same creator pointed out that a short which ends exactly on the last word tends to get watched again, because the loop feels natural. That’s easier to get right when you’re not cutting into the payoff to save time.
What the extra seconds don’t buy you is an excuse to keep the slow parts. Thirty-five seconds of a complete idea is fine. Thirty-five seconds of a fifteen-second idea plus filler is worse than fifteen.
What length doesn’t fix
It’s tempting to treat length as the lever. In my experience it mostly isn’t.
I spent some time studying one public short-form account’s full set of 44 reels to see what separated the hits from the rest. The lengths in that set ran from about 24 to 65 seconds, and length had essentially no relationship with views. What did matter was the opening: reels that held one continuous shot for the first two seconds did much better than ones that cut early, and greetings or slow introductions tended to flop. That’s one account, in one niche, so I treat it as a hint rather than a rule. But it points the same way as the pilot: get the first seconds and the ending right, and length becomes a secondary decision.
So if a short isn’t working, I check these before touching the length:
- Does it open on the strongest line or the most interesting thing on screen?
- Does a stranger understand what’s happening by the end of the first sentence?
- Does it end on the point, not a beat after it or a word before it?
When I go shorter or longer
The 30–45 second range is a default, and different creators want different things.
Shorter. Some moments are complete in ten or fifteen seconds: a single reaction, a quick exchange, a one-line joke with its setup in the same breath. Another creator I’ve worked with, who runs an interview channel, was happy with anything from 8 to 45 seconds as long as each clip stood on its own. For him I let the moment set the length inside that range instead of stretching everything to 30.
Longer. A story with a real arc, or an argument that needs two or three beats, can earn 50 or 60 seconds. With three-minute Shorts available it can go further. My rule is that every extra section has to add something new: a twist, a second payoff, a stronger line. If the last twenty seconds are the speaker restating the point, cut them.
The creator’s own data. If a creator already posts Shorts, their own best performers beat any general advice. I look at what length their top clips are and start there.
A quick way to decide
When I’m trimming a clip, I go through this order:
- Find the payoff first and mark exactly where it ends.
- Walk backwards to the earliest line a stranger needs to understand it. That’s the start.
- Cut everything between the two that doesn’t move the moment forward.
- Check the total. If it’s in the creator’s range, done. If it’s far outside, the clip is either the wrong moment or two moments stuck together.
Length is the output of that process, not the input. For most talk-based content it ends up somewhere around 30–45 seconds, which is why that’s my default. Framing is the next decision after length, and I covered that in why a vertical crop isn’t always the right edit.
If you’d like to see how your own long-form videos cut down, I can make a free short from one of your videos, trimmed to whatever length suits the moment. You keep it either way.