Published 29 September 2026 · Figures measured 15 and 25 September 2026 · Every figure below carries the number of Shorts behind it
First, what the number actually is
Average percentage viewed is the average time someone spent on the Short, divided by the length of the Short. It is a ratio, not a share of an audience, and a ratio can exceed one.
So a 10 second Short with an average view duration of 12.7 seconds reads 127%. Nobody watched 127% of a video. The average viewer watched it through once and then kept watching into a second pass. On a Short, which restarts automatically when it ends, that is easy to do without deciding to.
This is also why the number behaves differently from a long-form video's retention. A 12 minute video is very unlikely to be watched one and a quarter times. A 9 second one is.
Mostly, it is a fact about length
The first thing our data says about a number above 100% is that it is not unusual.
On 737 Shorts across 24 channels, measured 25 September 2026, 264 of them (35.8%) had an average percentage viewed above 100%. Ten days earlier, on a smaller snapshot of 575 Shorts, the figure was 223 (39%). Roughly one Short in three.
And it concentrates almost entirely in short Shorts. Of that 575, 200 of the 325 Shorts under 20 seconds were above 100%, against 13 of 92 at 20 to 40 seconds, 6 of 53 at 40 to 60 seconds and 4 of 105 over a minute. The median length of a Short above 100% was 10 seconds.
That has a consequence worth stating plainly, because it is the part most easily misread. A whole length band can sit above 100% for the same reason a single Short does. On that same 575, the median Short under 20 seconds read 109.7% viewed (n = 325). That is not a band of unusually gripping videos. It is a band short enough to be watched twice inside the time it takes to lose interest. 109.7% is not a target. It is an artefact of the band, and a longer Short that never reaches it is not failing at anything.
A correction we owe our own notes. We used to say that every Short above 100% in our data was 20 seconds or shorter. On the 25 September base that is no longer true: 34 of the 264 run longer than 20 seconds. Among the 294 Shorts that also carry a stored second-by-second curve, 114 exceed 100% and 14 of those are longer than 20 seconds, including a 58 second Short at 276.4%. The generalisation held on a much smaller catalogue and stopped holding as the catalogue grew.
What being above 100% actually predicted
The common advice is that a loop tells YouTube viewers wanted more, and that YouTube answers by showing the Short to more people. We cannot check the first half of that. We do not have any view into what the platform concludes, and we will not guess.
We can check the second half against outcomes, because we hold both the retention and what the Short went on to do. Here is every Short under 20 seconds in the 15 September base, split by average percentage viewed.
| Average % viewed | Shorts | Median views | Reached 1,000 views | Reached 4,000 | Median length |
|---|---|---|---|---|---|
| 0 to 70% | 43 | 14 | 2.3% | 0.0% | 16s |
| 70 to 90% | 51 | 127 | 25.5% | 0.0% | 15s |
| 90 to 100% | 31 | 107 | 29.0% | 3.2% | 11s |
| 100 to 120% | 70 | 1,090 | 54.3% | 2.9% | 11s |
| 120 to 150% | 49 | 320 | 40.8% | 8.2% | 10s |
| 150 to 200% | 29 | 59 | 27.6% | 6.9% | 9s |
| 200% and above | 52 | 65 | 7.7% | 0.0% | 9s |
It rises, it peaks between 100% and 150%, and then it comes back down. Above 200%, the Shorts in this band reached 1,000 views 7.7% of the time, which is worse than the Shorts sitting at 70 to 90% (25.5%) and far worse than the peak (54.3%).
The obvious objection is that this pools seven different channels, and a band could simply be one channel's habits. So the same split was run inside a single channel, with no pooling at all: one creator's 151 Shorts under 20 seconds, the largest single block in the base.
| Average % viewed | Reached 1,000 views |
|---|---|
| 0 to 70% | 0.0% |
| 70 to 90% | 27.3% |
| 90 to 100% | 38.1% |
| 100 to 120% | 71.8% |
| 120 to 150% | 66.7% |
| 150 to 200% | 25.0% |
| 200% and above | 9.1% |
Same shape, one channel, no pooling.
Does a looping Short beat a non-looping one on the same channel?
Put more simply than the bands: take one creator and one length band, split their Shorts at 100%, and ask which half did better. We found five such groups with at least four Shorts on each side.
| Length band | Above 100% | Median views | At or below 100% | Median views | Higher |
|---|---|---|---|---|---|
| Under 20s | 92 | 860 | 59 | 89 | Loopers |
| Under 20s | 30 | 19 | 26 | 6 | Loopers |
| Under 20s | 46 | 160 | 7 | 177 | Non-loopers |
| Under 20s | 6 | 326 | 19 | 496 | Non-loopers |
| 20 to 40s | 4 | 1,648 | 10 | 2,126 | Non-loopers |
Two out of five. On five groups that is a coin flip and we are not going to call it anything else. Whatever separates a creator's best Short from their worst on our data, crossing 100% is not it.
The limits, before you use any of this
The fall above 200% is the weakest part of this page, and we would rather say so than have it found. The Shorts in that band have a median of 65 views. Average percentage viewed is noisy when very few people watched, and that noise would produce exactly this collapse on its own. We cannot separate the two. Treat "above 200% did worse" as a result we measured and cannot yet defend as real.
What would settle it: fifteen or more Shorts in each of the 150 to 200% and 200%+ bands, restricted to Shorts above 500 views, where curve noise cannot drive the result. We do not have that yet. When we do, we will publish whichever way it lands.
The peak has a confound too, and it points the other way. A Short that gets distributed further is shown to better-matched viewers, and its measured retention rises with it. So some of why the 100 to 120% band looks good is that those Shorts had already travelled. That bias favours "higher retention is better", which makes it notable that the top two bands still fall.
These are small channels. The median Short under 20 seconds in this base took 131 views. Nothing here describes a catalogue that routinely reaches six figures, and it should not be read as if it does.
A test you can run on your own curve in about ten seconds
There is a check that tells you whether your number is measuring retention at all, and it needs no threshold and no benchmark. Open the Short in YouTube Studio and look at its audience retention graph.
Find the lowest point of the whole curve. Not the start, not the end. The lowest point anywhere.
- If the lowest point is still above 100%, then on average every single second of that Short was watched more than once. There is no drop-off in it to diagnose, because the curve never falls below one complete viewing. The number you are reading is a measure of how much the Short looped, not of how well it held anybody.
- If the curve dips below 100% anywhere, that dip is a real moment where real people left, and it is the part worth looking at.
We checked how tightly those two things track. Across 33 Shorts on 11 channels (generated 17 September 2026, measured 25 September 2026), the rank correlation between average percentage viewed and the lowest point of the curve was +0.956. Above roughly 100%, average percentage viewed is close to a direct measure of looping.
It is not a hypothetical. 5 of those 33 Shorts had every point of their curve above 100%. On 4 of the 11 channels, the Short with the highest average percentage viewed in its length band was one of them. One was 94 seconds long, read 195.6% viewed, never dropped below 186%, and had 122 views: the deepest looper on its channel and one of its quietest Shorts.
Scope. 33 Shorts across 11 channels is a small set, and the +0.956 is measured on it and nothing wider. It is offered as a test you can apply to one Short of your own, not as a population statistic.
What this page does not say
- Not "looping is bad." 100 to 150% was the best-performing band we measured, by a distance. If your Short sits there, the honest reading is that it was replayed, and the Shorts that were replayed a little did well on our data.
- Not "engineer a loop to get distribution," and not "avoid loops" either. We measured what happened after the fact on somebody else's videos. That is not the same as knowing what causes what, and we are not going to dress it up as though it were.
- Nothing at all about what YouTube does with a replay. We cannot observe it. Every sentence on this page is about Shorts and view counts we can see.
- Not a benchmark. There is no number on this page you should be aiming at. The most useful thing here is the curve test, and its output is "stop reading this figure as retention", not a score.
And the thing we genuinely cannot see
Every serious account of how Shorts get distributed points at the same metric: viewed versus swiped away, the share of people who were shown the Short and stayed rather than flicking past. We do not have it. It is visible in YouTube Studio and it is not in the data a creator can authorise us to read, and there is no honest proxy for it. Average percentage viewed is not one: it describes the people who started watching, and says nothing about how many never did.
So we will not publish an estimate of it, and we will not tell you whether you clear anybody's threshold for it. We explain what happened to the people who started. Studio tells you how many never did.
Questions people ask about this number
Taken from public creator forums and search results in September 2026, close to the wording they were asked in. Every answer carries the number of Shorts behind it, and where the honest answer is that we do not know, that is the answer.
I have 103.9% average percentage viewed on a Short and it still stopped at 10,000 views. Why?
We cannot tell you why one particular Short stopped where it did. What our data can say is that crossing 100% is not the thing separating a creator's better Shorts from their worse ones. In the five groups above, each one channel and one length band split at 100%, the looping half had the higher median views in two of five.
It also says the figure can be high for a reason that has nothing to do with how well the Short held anybody. The 94 second Short described above read 195.6% viewed, never dropped below 186% at any point of its curve, and took 122 views (n = 33 Shorts across 11 channels, measured 25 September 2026).
So before treating a high number as a puzzle, run the curve test above on that Short. If its lowest point is still above 100%, the number is measuring replays and there is no drop-off inside it to explain. And see the section above on what we cannot see: how many people were shown the Short and swiped past it is not in the data a creator can authorise us to read, and it is not something we will estimate.
Is an average percentage viewed above 100% a mistake in YouTube Studio?
No. It is a ratio of time watched to video length, and a ratio can go above one. The first section of this page explains how that happens on a Short.
It is also ordinary rather than exceptional, which is the part worth knowing. On 737 Shorts across 24 channels, measured 25 September 2026, 264 of them (35.8%) were above 100%. Roughly one in three. If yours is above 100%, the most likely single explanation is that the Short is short.
Does building a loop into a Short get it shown to more people?
We do not know. We have no view into what YouTube does with a replay, and we are not going to guess at it.
What we can report is what happened afterwards, on Shorts we neither designed nor influenced. The best performing band we measured was 100 to 150% viewed (n = 325 Shorts under 20 seconds, measured 15 September 2026), and the two bands above it fell back rather than climbing further, though the limits section above says plainly why we do not yet trust the size of that fall. Splitting single channels at 100% inside one length band, the looping half won two of five groups.
That is a record of outcomes we observed after the fact. It is not a mechanism, and none of it tells you that adding a loop to your next Short would cause anything.
Which of my Shorts should I study to work out what I did right?
Not the one with the highest average percentage viewed, which is the advice sitting on most of the pages that answer this question.
Across 33 Shorts on 11 channels, measured 25 September 2026, 5 had every point of their retention curve above 100%. On 4 of those 11 channels, the Short with the highest average percentage viewed inside its length band was one of those 5. On four channels in eleven, picking a study subject by that number would have handed the creator a Short with no drop-off in it to study.
33 Shorts on 11 channels is a small set, and this is offered as a caution about a selection rule rather than as a population statistic. Run the curve test first, then pick the Short.
Two questions this page deliberately does not answer. What average percentage viewed is normal for a Short of a given length has a different table behind it and is not this page's question. Why a Short with better retention lost to one with worse retention is a wider measurement than anything here and is not about the 100% line at all.
How this was measured
Why we looked at this at all
ShortLens reads the measured retention curve of each of your Shorts and tells you where viewers left, to the second. When a curve runs above the 100% line we draw that part as a labelled replay band, and we leave it at that: it is an observation about the video, not a score, and this analysis is the reason it stays that way.