How Long Should a YouTube Video Be? (What the Data Says)

There is no magic number. Not 8 minutes, not 10, not “as short as possible.” The question is not how long a video should be, it is how much total watch time it can earn. That single shift changes almost every decision you make about length.

Here is how to actually pick a length, instead of guessing at a round number.

The metric YouTube actually rewards

YouTube does not rank videos on percentage watched. It ranks on total watch time, and total watch time is a function of both how long the video is and how much of it people stick around for. A 3-minute video with 90 percent retention gives you about 2.7 minutes per view. An 11-minute video with 55 percent retention gives you roughly 6 minutes per view, more than double. Same effort to click, more than double the signal to the algorithm.

That is the whole reason “longer” videos so often outperform “shorter” ones in recommendations, even though the retention percentage looks worse on paper. Percentage is a vanity number. Minutes watched is the number that moves distribution.

Why 8 to 12 minutes often beats 3 minutes

For most mid-form topics, a video padded out past 3 or 4 minutes with real content, not filler, tends to out-earn a tight 3-minute cut, because there is simply more runway for watch time to accumulate. This is not permission to stretch a 4-minute idea to 12 minutes with recaps and stalling. It only works if the extra minutes hold attention. Add a fluff minute and you lose more retention than you gain in length, and your total watch time drops.

The honest version: length is a lever you can pull only as far as your content supports it. Find the point where the topic runs out of real material, and stop there.

When short-form wins anyway

Some topics do not have 8 minutes of substance, and forcing them there kills the video. Short wins when:

The answer is genuinely quick. “Does X work” or “which setting to use” often only needs 60 to 90 seconds.
You are testing a new topic or hook. Shorts and quick videos are cheap ways to find out if an idea has an audience before you invest in a long version.
The format is inherently fast. Highlights, reactions, and single-tip clips lose people if you stretch them.

If you are cutting a video down to hit a length target and it feels thin, that is the content telling you it should have been short in the first place.

Match length to the promise

The real rule is not a minute count, it is a promise match. Your title and thumbnail make a promise. The video should deliver that promise as fast as the topic allows, and not one second slower. A tutorial that promises “the fastest way to do X” earns distrust if it opens with a 90-second intro. A deep dive that promises “everything you need to know” earns distrust if it wraps up in 2 minutes with half the topic untouched.

Ask before you film: what did the title promise, and how much runway does that promise actually need? That answer is your target length, not a number you picked because a video you watched was 10 minutes long.

How to keep a long video tight

If your topic supports length, protect the watch time instead of assuming the extra minutes are free:

Cut the intro to one sentence that states what the video delivers.
Remove any section that restates something you already said.
Watch your own retention graph and mark exactly where it drops, that is where the boring part starts.
Re-cut or re-script that section before your next upload, not after five more videos with the same problem.
If a section could be its own video, it probably should be, split it out instead of cramming it in.

The one-line version

Do not chase a length. Chase the longest runtime your topic can hold attention for, and cut ruthlessly to protect every minute you keep.

Your own retention graph already has the answer for where your videos lose people, most creators just have not looked closely. A free channel audit lines your videos up side by side so you can see exactly which lengths and topics are actually holding your audience, instead of guessing from a general rule.