2.6 times as many YouTube videos end just after 8 minutes as just before

How long a video runs is a decision someone made before uploading it, and it is one of the few such decisions anyone outside the channel can measure exactly. It also runs straight into one of YouTube's own rules: a video has to be longer than eight minutes before it can carry ads in the middle, a line YouTube moved down from ten minutes in 2020. If a rule like that changes what creators make, the evidence should show up in the lengths themselves. It does.

We took every video uploaded in October 2026 by a channel that has 10K or more subscribers in our 5 Nov 2026 rankings: 9,512,779 videos from 916,291 channels. Live streams are left out (2.3M of them), because a stream runs as long as it runs rather than to a length someone chose. So is anything under a minute, because we cannot reliably tell those apart from Shorts.

Longer videos are rarer — until eight minutes#

Most of these videos are not short. A quarter run under 5:51, half run under 12:40, and one in ten runs past 1:02:41. 41% finish inside ten minutes, and 17% go past forty.

Within that range, each extra half minute of length holds fewer videos than the half minute before it. The group just short of eight minutes, 7:30–7:59, holds 118,491 videos — fewer than every group before it. The very next group, 8:00–8:29, holds 320,701, nearly three times as many, and after that the steady decline picks up again as if nothing had happened.

Measured right at the line rather than in those half-minute groups, it is 124,295 videos in the 30 seconds before eight minutes against 319,403 in the 30 seconds after. The difference between those two numbers is 195,108 videos, in one month, from this one set of channels.

The jump happens at the line, and it stays up#

Counted one second at a time, the shape is hard to miss. Up to 479 seconds the count holds steady at a few thousand videos per second. At exactly 480 seconds — eight minutes — it is already 10,262. At 481 seconds, the first length that is longer than eight minutes rather than equal to it, it is 16,931. Three seconds later it is still 13,627, far above where it was before the line.

Ten minutes behaves differently, and that difference is the useful part. There the busiest second is the round one itself: 11,436 videos at 600 seconds against 6,513 at 599, and by 604 seconds it is back down to 7,457. That is what aiming at a round number looks like — a spike, then a return to normal. A rule written as "longer than eight minutes" leaves the other shape: the count steps up as the line is crossed and stays up.

Nothing about taste, viewing habits or our own measuring changes between one second and the next, so something tied to that exact second is doing the work, and the ad rule is the only thing we know of sitting there. What the data cannot tell us is how any creator gets past the line — a slower opening, one more point, a longer sign-off — or who is doing it deliberately.

The pile-up moved when YouTube moved the rule#

We ran the same two counts for every October from 2017 to 2026, over the same channels. Up to 2019 there is no bunching at eight minutes at all: 0.9 as many videos just after the line as just before it, meaning slightly fewer, which is simply what a steady decline looks like. Ten minutes in that same October shows 2.7.

In 2020 — the year YouTube lowered the threshold — the two trade places: 2.0 at eight minutes and 1.5 at ten. The eight-minute pile-up has held or grown since; the ten-minute one has drifted down to 1.1. The old line did not empty out the day it stopped paying. It took years to fade, which is what a habit looks like in aggregate.

Two things to keep in mind about the earlier years. A channel is in this set because it is above 10K subscribers today, so the earliest Octobers describe the channels that survived rather than YouTube as it was at the time, and we hold fewer videos from those months. Neither affects the comparison within a single year, which counts the same videos on both sides of the line.

Gaming bunches hardest, music barely at all#

Every category we could measure shows the pattern, but not equally. Gaming is the strongest at 4.0, followed by Pets & Animals at 3.4. Music is the weakest at 1.3, which makes sense: its typical upload runs 7:35, and a song lasts as long as the song.

Channel size barely matters by comparison. The pile-up runs from 2.5 among channels with 10K to 100K subscribers to 3.0 above 10M, even though their typical lengths are far apart (13:10 against 8:58). Whatever is happening at the line, small channels do it about as much as the biggest ones.

Videos just past the line also collect more views than videos just before it: typically 2,980 views against 2,095, with 2,730 across the whole set. We would not read that as length earning views. Channels that place a video just past a payment threshold are more likely to be the ones already earning from ads and already established, and this measurement cannot separate the video's length from the channel behind it.

What this means#

If you compare video lengths — your own against a niche, or one roster of channels against another — compare inside a category rather than against a site-wide average. A category's typical length and its pile-up at the line are two different facts: gaming channels bunch hardest at eight minutes without having the longest videos, while music channels have the shortest videos in our set and barely bunch at all. Where a peer group clusters a few seconds past eight minutes, that is a platform rule showing up, not the length its viewers asked for.

The same caution applies to anything built on length: cost per minute, watch-time models, any claim about an "ideal length". They all inherit a set of lengths that a rule has bent in one spot, and an average taken across the bend hides it. For the rest of the picture, the category mix of channels shows which categories these uploads come from, and quiet channels covers how many channels in this set are still uploading at all. The most watched recent uploads are on trending, and you can search upload titles in the video lists.