How Long Should a YouTube Short Be?

This question usually comes with an expectation of a number. The honest answer is that the platform's maximum has changed more than once — Shorts began at a minute and has been extended since — so any guide quoting a fixed ceiling is probably out of date. Check the current limit in YouTube's own help pages, then forget about it, because the maximum is almost never the right length anyway.

Length is a decision about attention, not about what the platform permits.

Why longer clips usually underperform

Short-form feeds optimise for completion. A viewer who watches a thirty-second clip to the end signals something different from a viewer who abandons a two-minute clip at forty seconds, even though both watched the same forty seconds. Longer video makes completion harder by arithmetic alone: to finish, someone has to stay twice as long.

That does not mean short is always better. A clip that ends before the idea lands gets abandoned too, just sooner. The useful question is not "how long should this be" but "what is the shortest version that still delivers the thing I promised in the first three seconds".

Three durations that do different jobs

In practice, short-form video clusters into three lengths, each with its own purpose.

Fifteen to twenty-five seconds: the single beat. One claim, one demonstration, one punchline, one reaction. This length rewards a strong opening line and punishes any setup. It is the easiest length to complete and the hardest to make memorable, because there is no room for a turn.

Thirty to forty-five seconds: the single idea with evidence. A claim, a reason, an example. This is the workhorse length for tutorials, product demos and commentary, and it is roughly where a spoken paragraph fits comfortably.

Sixty to ninety seconds: the short story. Setup, complication, resolution. Useful for personal stories, case studies and anything where the shape of the narrative is the point. Completion rates drop here, so the opening has to be unusually strong.

Anything beyond that is usually two clips wearing one coat.

When to split a clip instead of trimming it

If a clip needs three examples to make its point, that is not one ninety-second video. It is three thirty-second videos, each with one example, published across three days. You keep the argument intact, you triple the number of things you can post, and each piece gets its own chance in a feed.

The exception is a story whose power depends on accumulation — a build-up, a reveal, a payoff. Splitting that destroys it. Ask whether the parts work alone. If they do, split.

The rule of one idea

Before touching a timeline, write the clip's claim as a single sentence. If the sentence needs an "and", the clip has two ideas and should be two clips.

This rule does more for retention than any editing trick, because a viewer can only hold one promise at a time. A clip that promises one thing and delivers it feels satisfying. A clip that promises three things and delivers two feels like it wasted the viewer's time, even though more happened.

End on purpose, not on exhaustion

Two endings work, and they are opposites:

• The loop. The last frame flows into the first, so a viewer who reaches the end starts again without noticing. This inflates watch time and works best for visual or comedic clips.
• The hard stop. The clip ends on the strongest sentence, immediately after it. No outro, no "follow for more", no trailing laugh.

What does not work is drifting to a stop: the speaker finishes the thought, then says something that weakens it, then the clip ends. Those last four seconds cost you the completion you had already earned.

Cut at the peak. If the sentence sounds abrupt to you, it is probably exactly right — you have heard the material twenty times and the viewer has heard it once.

How to test length honestly

You cannot A/B test the same clip at two lengths, because the shorter one is a different video. What you can do is compare across ideas:

• Same topic, different lengths, different weeks. Take a subject you have covered at forty-five seconds and make a twenty-second version with a different opening. Compare completion, not views.
• Read the retention curve, not the headline number. If most viewers leave in the first three seconds, length is not your problem — the opening is. If they leave at eighty percent, the ending is.
• Watch your own clips on mute. If you cannot follow the point without sound, the pacing is being carried by audio you will not have in a feed.

What matters more than length

Three things, in order:

1. The first three seconds. A clear claim, stated or shown immediately. This determines almost everything downstream.
2. The captions. Most feeds autoplay muted; the words on screen are how the clip is understood.
3. The framing. Vertical, subject in the upper two thirds, text clear of the interface. A great clip cropped badly looks like an accident.

Length is fourth. Get the first three right and the length mostly chooses itself.

A practical default

If you want a rule of thumb rather than a rule: forty seconds, one idea, one piece of evidence, ending on the strongest sentence. Then vary deliberately — cut a twenty-second version when the idea is a single beat, allow ninety when the story needs it, and split anything that needs a second "and".

The clip is the right length when removing anything else would break it. That is a better test than a stopwatch.