Why a metric can go over 100%
In the Shorts report, average percentage viewed is calculated against the length of the Short, and YouTube counts a replay as more viewing of the same Short. A 30 second Short that a viewer watches once and swipes away from at the end sits at 100%. The same Short watched one and a half times sits at 150%. That is not a rounding artefact, it is the single clearest signal available that the content was worth staying for.
This is why the loop is a distribution lever and not a stylistic preference. Two Shorts can have identical hooks, identical retention curves through the body, and wildly different distribution, because one of them ends in a way that makes the viewer's thumb hesitate and the other ends in a way that says the transaction is complete.
It also explains something creators find maddening: a Short that seems worse can outperform a Short that seems better. The better one had a satisfying, closed, well produced ending. The worse one ended mid thought, the feed restarted it, and the viewer watched the opening again before deciding to leave.
- Percentage viewed is measured against the Short's length, and replays add to it.
- A 30 second Short watched 1.5 times reports 150% viewed.
- Two Shorts with the same hook can perform very differently on the ending alone.
- A closed, satisfying ending is often worse for distribution than an open one.
What a loop actually is: the seam, not the content
A loop is not a Short about a circular topic. It is a Short whose last frame and first frame belong to the same moment, so that when the platform restarts it the viewer does not perceive a restart. There are three seams to close, and most Shorts close none of them.
The visual seam is the easiest: the composition, the framing and the brightness of the final frame should be close to the opening frame. A Short that ends on a wide dark shot and opens on a bright close up announces the cut. A Short that ends on the same framing it opened with reads as continuous motion.
The audio seam is the one people forget. If the narration ends mid sentence and the music stops dead, the restart is audible even when it is invisible. Music that carries under the cut, and narration whose last words feed grammatically into the first words, hide the seam completely. The third seam is logical: the final sentence should raise the question that the first sentence answers, which is what makes the second watch feel like new information rather than a repeat.
- Visual seam: last frame framed and lit like the first frame.
- Audio seam: music continues under the cut, narration feeds into the opening line.
- Logical seam: the last line raises the question the first line answers.
- A loop is a construction, not a topic choice.
The four endings that guarantee a Short never loops
The first killer is the fade to black. It is the standard ending in video editing and it is exactly wrong here, because darkness on the final frame is the universal signal that something is finished. The viewer's thumb moves before the restart happens.
The second is the outro card, the branded end plate with a logo and a subscribe prompt. On long form it belongs there. On a Short it converts the last second, the most valuable second you have, into a frame with no content. The subscribe decision on Shorts is made during the content, not on a card at the end.
The third is the verbal full stop: thanks for watching, see you in the next one, let me know in the comments. Each of those is an explicit instruction to leave. The fourth, and the most common on faceless channels, is the trailing silence: two or three seconds of nothing after the last word because the render length was set by the music rather than by the narration. Those two seconds are a dead zone that reads as an ending even when nothing on screen says so.
Three of those four are decisions and cost nothing to change. The fourth, the trailing silence, is usually a rendering artefact rather than a choice, and it is the one most likely to survive unnoticed. In FalconVid the Short comes out of the pipeline with the narration and the music aligned, and the Studio lets you watch what you have and trim the tail, shorten the intro or change the track without regenerating anything, which is how a dead two seconds gets found and removed instead of shipped.
- Fade to black: darkness reads as finished, the thumb moves first.
- Outro card: spends the most valuable second on a frame with no content.
- Sign off lines: an explicit instruction to leave.
- Trailing silence: two dead seconds that end the Short before the file does.
Three loop structures that hold up across niches
The question loop is the most portable. The Short opens with a claim, delivers the body, and closes by raising the specific question that the opening claim answers. The viewer hears the question, the feed restarts, and the first line lands as the answer. This works in finance, in history, in science and in list content without any change of format.
The reveal loop suits process and transformation content. The opening shows the finished result, the body shows how it was arrived at, and the ending returns to the same finished result on the same framing. The restart shows the result again and it now means something different, because the viewer knows how it was made.
The countdown loop suits list content. The Short counts down to a number one that is stated but not fully explained, and the opening line is the explanation. It is the highest replay structure of the three and also the easiest to overdo, because a viewer who feels tricked twice does not come back for a third.
- Question loop: end on the question your first line answers.
- Reveal loop: open on the result, end on the same result.
- Countdown loop: state the number one, explain it in the opening line.
- All three depend on the last sentence, which is the cheapest thing to change.
Where the loop and the call to action collide
The honest tension in this article is that a loop asks you to remove the ending, and a channel needs viewers to subscribe and to move to the long form video. Removing the outro card and the sign off feels like removing the business case, which is why most channels keep them and quietly lose the replay.
The resolution is placement rather than deletion. The call to action belongs in the body, not the tail: a spoken line at around 60% of the way through, while attention is high, plus a persistent on screen element that lives through the whole Short rather than appearing at the end. The pinned comment carries the link. The description carries the link. The last second carries the loop.
This is also where the Short earns its place in the funnel. A Short that loops three times has delivered three times the exposure of the same content on a single watch, and that exposure is what makes the subscribe decision during the body plausible in the first place.
This placement is a template decision, not a per video one, which is exactly the kind of thing that drifts when a person is publishing five times a week. FalconVid generates the CTA layer together with the Short, so the spoken call to action and the persistent on screen element sit where you decided they sit on every upload, and the last second stays free for the loop. The CTA layer and the 9:16 Shorts generator are on every plan, from Starter at $47 upward.
- Put the spoken CTA at around 60% of the Short, not at the end.
- Use a persistent on screen element instead of an end card.
- Links live in the pinned comment and the description.
- The last second belongs to the loop, and nothing else.
How FalconVid makes the loop cheap enough to actually test
The reason most channels never optimise the ending is not ignorance, it is cost. Testing three endings by hand means three export and upload cycles on top of everything else, and by video 30 nobody is doing that. With FalconVid the Short comes out of the pipeline rather than out of an editing session: 9:16 Shorts, karaoke captions in more than 15 styles and the CTA layer are generated with the video, on every plan from Starter at $47 upward.
The arithmetic is what makes testing realistic. Rendering a Short costs 5 credits, against 1,008 credits for an entire 12 minute video in economy mode, 8,676 in balanced and 26,760 in premium. Three ending variants of the same Short cost 15 credits in render terms, which is a rounding error against any plan on the table, and the Studio lets you watch the version you have and adjust it, shortening the intro, swapping media or changing the music, without regenerating the whole thing.
Volume matters here too, because the loop is a structure you learn by repetition. One long video can be repurposed into several Shorts, and FalconVid produces them with AI specialists working in parallel and a video ready in up to 30 minutes, publishing to YouTube, Instagram, TikTok, Rumble and Facebook from the calendar you approved once. Starter runs 2 simultaneous generations, Pro 5, and Scale 50, so the number of endings you can test in a week is a plan decision rather than a scheduling problem.
And because every creation feature is on every plan, the pieces this article depends on, the caption layer, the CTA layer, the Shorts generator and the Studio, are not features you upgrade into. What changes between plans is volume, channels, simultaneous generations, the Senior Analyst and support.
- Rendering a Short costs 5 credits, against 1,008 for a 12 minute economy video.
- Three ending variants are 15 credits of render, a rounding error on any plan.
- The Studio adjusts intro, media and music without a full regeneration.
- 2 simultaneous generations on Starter, 5 on Pro, 50 on Scale.
How to read whether the loop is working
The number to watch is average percentage viewed in the Shorts report, and the threshold to beat is 100%. Below 100% the average viewer is leaving before the end and the loop is irrelevant, because the ending is not being reached. Between 100% and 130% the loop is closing for some viewers. Above 130% you have something worth copying into the next twenty Shorts.
Compare against yourself rather than against published benchmarks. Shorts performance varies so violently by niche and by length that any external number is close to useless, whereas your own last ten Shorts are a fair control group. Change one thing, the ending, keep hook and length steady, and read the difference.
One caution that saves wasted effort: a Short with a weak hook cannot be rescued by a loop. If the first two seconds are losing 60% of viewers, fixing the ending changes a metric that only a minority ever reach. Fix the top of the Short first, then spend a week on the bottom of it.
- Under 100% viewed: the ending is not being reached, fix the hook first.
- 100% to 130%: the loop is closing for part of the audience.
- Above 130%: copy that ending structure into the next twenty Shorts.
- Compare with your own last ten Shorts, not with published benchmarks.