ChumokuA Shorts retention curve shows what share of viewers are still watching at each moment of the video. It is the only honest record of where a Short loses people, and it is read by shape rather than by any single number. A vertical drop at the start is an opening problem. A steady slope is normal. A bump upward means a moment is being rewatched, because Shorts loop and the loop counts. The most common mistake is reading the average percentage instead of the shape: two videos with the same average can have completely different problems, and only one of them is fixable in the edit.
What the curve actually measures
The horizontal axis is position in the video, not clock time for any one viewer. The vertical axis is the share of views still playing at that position.
It is a record of a crowd, not of a person. A dip at 0:07 does not mean one viewer left at 0:07; it means a noticeable share of everyone who started the video stopped around there. That is what makes it worth acting on: a single viewer leaving is noise, a step in the curve is a repeatable reaction to something you can go and look at.
The shapes and what each one means
A cliff in the first second or two. The opening is the problem. The video has not done anything wrong yet in the sense of craft, it has simply not given a reason to stay. This is the most common shape and the most fixable one.
A steady, gentle decline. Normal, and not a defect. Every video loses people gradually. If the curve declines evenly with no steps, the video is working and the way to improve it is to make it shorter, not to patch a moment.
A step partway through. Something specific happened there. Go and watch those two seconds. Usually it is a shot held too long, a second idea starting before the first has paid off, or the video visibly changing subject.
A rise, or a section above the starting level. People are rewatching. Shorts loop by default, so a moment people scrub back to or watch twice shows up as retention going up. This is a good sign and worth understanding, because whatever is there is the part of the video that worked.
A flat line that ends abruptly. The video held everyone and then stopped. If that ending is the payoff, the video is doing its job.
Why it can go over 100 percent
On Shorts this is normal and not a glitch. The format loops, and replays of a section count as views of that section, so a heavily rewatched moment can exceed the number of unique viewers who started the video.
Treat anything above the line as a marker rather than a score. It tells you which two seconds people wanted again. That is more useful for the next video than the average percentage is.
What the curve cannot tell you
Why. The curve shows where, never the reason. The reason is in the footage at that timestamp, and reading one off the other without watching is guessing.
Whether the video was good. A short video with a high average and forty views has not been tested. Retention is a measure of how a video holds the people who saw it, not of whether it reached anyone.
Anything about the algorithm. Retention shape is not a shadowban signal, a penalty, or a ranking. A video with a bad curve and few views has one visible problem and one unknown, and treating the first as the cause of the second is the fastest way to fix the wrong thing.
Anything at all, on a video with very few views. A curve built from a handful of views is noise with a line drawn through it.
Where to find it
YouTube Studio, open the individual Short, then the analytics for that video and its engagement or audience-retention section. The curve is per video; there is no channel-level version of it that means anything.
It is not available through the public YouTube Data API. Any tool that shows you a retention curve is either reading it from your own connected account through the Analytics API, with your permission, or estimating something else and calling it retention.
Common questions
What is a good retention percentage for a YouTube Short?
There is no single number worth chasing, because it depends heavily on length: a 15 second Short and a 55 second one are not comparable. The useful comparison is against your own previous videos of similar length, and the useful reading is the shape rather than the average.
Why is my Shorts retention above 100 percent?
Because Shorts loop, and replays of a section count toward that section. A moment people watch twice can exceed the number of people who started the video. It is a sign that part of the video worked.
Can I see the retention curve for someone else's Short?
No. Audience retention is private analytics, available only to the channel that owns the video. Any figure presented as a competitor's retention is an estimate built from public data, not the real curve.
Does a drop in the curve mean the algorithm stopped promoting the video?
No. The retention curve describes how the people who watched behaved inside the video. It says nothing about distribution, and reading it as a ranking signal leads to fixing things that were never broken.
Retention · Analytics
Published 12 September 2026
Chumoku does this reading for you: send it a Short and it watches the whole thing, marks where attention drops, and tells you what to change in the edit. With your channel connected it reads your real retention curve rather than guessing at it.
Try itKeep reading
20 September 2026 · 6 min read
What view jail on Shorts actually tells you
Shorts that keep stopping at the same view count. What the ceiling narrows the problem down to, and which of the usual explanations for it nobody can actually check.
Distribution · Retention
12 September 2026 · 4 min read
Why Shorts get swiped in the first two seconds
A Short is judged in a feed, not after a click. The reasons viewers swipe in the first two seconds, and what to change in the edit.
Hooks · Retention · Editing