How Long Should a Short-Form Video Be?

It's the question every creator asks eventually: how long should a short-form video actually be? Seven seconds? Thirty? A full minute? You'll find a confident answer for every one of those numbers, which is a good sign that the number itself is the wrong thing to obsess over.

The honest answer is that length is an output, not an input. The right duration is however long it takes to deliver one idea well — and not one second longer. This guide walks through how to reason about that, platform by platform and goal by goal, so you stop guessing and start measuring.

The metric that actually decides length

Platforms don't rank your video by how long it is. They rank it by how well it holds and satisfies attention. The two signals that matter most are:

Average watch time / completion rate — did people stay to the end?
Rewatches and replays — was it dense enough to watch twice?

Here's the tension. A longer video has more total watch time to give, but more room to lose people in the middle. A shorter video is easier to finish, but a near-100% completion rate on a three-second clip tells the algorithm very little. Your job is to find the length where completion stays high and total watch time is meaningful.

That's why "how long should it be?" has no universal answer — it depends on how much genuine substance your idea has. Padding a five-second idea to thirty seconds tanks your retention graph. Cramming a rich, two-part idea into eight seconds leaves value (and rewatches) on the table.

Rough starting ranges by platform

Use these as starting points to test, not rules. Every account's audience behaves differently.

TikTok: The format rewards both extremes — punchy 7–15 second clips that get rewatched, and longer 30–60 second (or more) storytelling when the payoff earns the runtime. Start around 15–34 seconds for most content and let the topic pull you shorter or longer.
Instagram Reels: Similar to TikTok, but the audience tends to reward tight, well-produced clips. 15–30 seconds is a sensible default to test first.
YouTube Shorts: Shorts can run up to a minute (and longer as limits expand), and the audience is often more tolerant of a slightly longer, more informative beat. Educational and how-to content frequently does well in the 30–60 second range.

Notice the overlap: 15–30 seconds is a defensible default across all three. If you have no data yet, start there, then let the specific video's substance stretch or compress it.

Let the hook set the floor, the payoff set the ceiling

A more useful way to think about length than "pick a number" is to think in beats:

The hook (first 1–3 seconds). This is non-negotiable and roughly fixed in length. If the opening doesn't stop the scroll, nothing after it matters. The hook doesn't make the video longer — it earns the right to the rest of it.
The delivery (the middle). This is the part that flexes. One idea, one demonstration, one story. Its natural length is your video's natural length. If you're struggling to fill it, your video is shorter than you wanted. If you're rushing, it's longer.
The payoff / CTA (last 1–3 seconds). A clear ending — a takeaway, a punchline, a reason to follow. This caps the video.

When you frame length this way, the "how long" question dissolves. The video is as long as the delivery beat needs, bookended by a fixed hook and a fixed payoff.

Match length to your goal

Different objectives justify different runtimes:

Reach / discovery: Lean shorter and denser. High completion and rewatches are what push a video into more feeds. A tight 10–20 second clip with a strong loop is built for this.
Teaching / building trust: Lean a little longer. If someone needs 45 seconds to actually learn the thing, giving them 15 helps no one — a confused viewer doesn't convert. Substance beats brevity here.
Selling / converting: Long enough to earn the ask, short enough to keep momentum. Front-load the value, then make the call-to-action unmistakable in the final beat.

Common length mistakes

Padding to hit a number. Adding filler to reach some "ideal" duration is the fastest way to create a mid-video drop-off cliff. Cut ruthlessly.
One video, one length, forever. Your best-performing length is discovered, not decided. Treat it as a hypothesis you keep testing.
Ignoring the retention graph. The single most useful artifact you have is the audience-retention curve. A sharp drop at second three means a weak hook; a slow bleed means the middle drags; a flat line to the end means you could probably have said more.

Test length like everything else

The reason length debates never end is that people argue from anecdotes instead of their own numbers. The fix is boring but reliable: produce the same idea at two lengths and compare. Make a 15-second cut and a 30-second cut, post them a few days apart to comparable audiences, and read the retention and watch-time data. Do that across a handful of topics and your account's real preferred range emerges — no guru required.

The catch is that most people never run this test, because producing two versions of everything is tedious. That's exactly the kind of grunt work worth automating. With a tool like Vicreon, you can generate multiple cuts of the same concept — different runtimes, different hook framings, different pacing — from a single brief, publish them on a schedule, and let the performance data tell you which length your audience actually rewards. The point isn't to guess the perfect number once; it's to make testing cheap enough that you never have to guess again.

The short answer

How long should a short-form video be? Exactly as long as it takes to deliver one idea well — usually somewhere between 15 and 60 seconds, biased shorter for reach and longer for teaching. Start with a 15–30 second default, let each idea's substance flex the middle, protect a strong hook and a clear ending, and then let your own watch-time data — not a blog post, not a rule of thumb — decide the rest.

Stop optimizing for a number. Optimize for a finished idea and a full retention graph, and the right length takes care of itself.