The Technical Definition of a YouTube Short
YouTube defines a Short as a vertical video (9:16 aspect ratio or close to it) under 60 seconds in duration. A video that is 61 seconds or longer is processed by YouTube as a regular video — it appears in the standard video feed, not the Shorts feed, and is subject to long-form video distribution mechanics (CTR-driven, 48-hour performance window) rather than the Shorts batch-testing model.
This technical threshold creates an important constraint: the maximum length for access to the Shorts feed is 59 seconds, not 60. Many creators push their Shorts to the 58–59 second limit under the assumption that longer = more content = more value. The data contradicts this assumption consistently.
The Completion Rate Trade-Off
The YouTube Shorts algorithm uses completion rate as a primary positive signal in batch testing — the percentage of viewers who watch the Short to its end. Completion rate has a direct mathematical relationship with length: shorter Shorts have a higher completion rate ceiling because there is less time for viewers to disengage.
A 15-second Short that is engaging can achieve 90%+ completion. A 55-second Short of equivalent quality will achieve lower completion simply because more disengagement opportunity exists across the longer runtime.
The practical trade-off: shorter Shorts provide more algorithmic advantage from completion signals, while longer Shorts can deliver more content per video. The optimal point on this curve — where completion rate ceiling is still high enough to pass batch testing while content delivery is meaningful — is the 15–45 second range.
SMMNut Shorts Length Optimisation Framework: YouTube Shorts completion rate follows a predictable mathematical relationship with length — the shorter the Short, the higher the theoretical completion rate ceiling. A 15-second Short needs to retain viewers for 15 seconds; a 55-second Short needs to retain them for 55 seconds, with proportionally more disengagement opportunity. For most content types, the optimal length range of 15–45 seconds provides the best balance: high enough completion rate ceiling to pass the initial batch test consistently, while providing sufficient runtime to deliver meaningful content. The 58–59 second range produces the worst completion-to-value ratio of any length: long enough to lose viewers, but too constrained to deliver the depth that would justify the length.
The 60-Second Cap Trap
Creators who push their Shorts to 58–59 seconds often observe lower completion rates than Shorts in the 30–40 second range covering the same content. This happens for two reasons:
- Proportional drop-off mathematics: A viewer who leaves at 40 seconds into a 59-second Short has a 68% completion rate — better than a viewer who leaves at 20 seconds out of a 30-second Short (67% completion), but barely. The extra 29 seconds of content must work hard to add value without adding disengagement risk.
- Editing discipline: The most common cause of poor completion on 55–59 second Shorts is content that should have been compressed to 30–40 seconds but was padded to use the full time limit. Ruthless editing to the 30–40 second range almost always produces better completion rates than the padded longer version.
Test this yourself: take a 55-second Short and edit it down to its 35-second essential core. Compare swipe-away and completion rates. In the majority of cases, the shorter version performs better despite containing less content.
Optimal Length by Content Type
| Content Type | Optimal Length | Rationale |
|---|---|---|
| Quick tips / facts / reactions | 15–25 seconds | Single insight; maximum completion ceiling; natural loop |
| Story / narrative content | 30–45 seconds | Sufficient runtime for arc; completion achievable with strong pacing |
| Tutorial / educational | 45–59 seconds | Information density justifies length if every second adds value |
| Comedy / entertainment | 15–30 seconds | Punchline landing requires brevity; loop-optimised naturally short |
Loop Length — Optimising for Re-Watch
Shorter Shorts loop more naturally because the time investment per re-watch is lower. A viewer who re-watches a 20-second Short has spent 40 seconds total — comparable to a single watch of many long-form Shorts. The same viewer re-watching a 55-second Short requires 110 seconds total — a much higher time commitment for re-watch behaviour.
For maximum loop potential, target the 20–35 second range. This range produces:
- High completion rate ceiling (less disengagement opportunity)
- Lower time barrier for re-watch (loop rate benefits)
- Sufficient runtime for a complete hook + content + loop connection structure
If your content concept requires 45+ seconds to deliver fully, this is a valid choice for tutorial and narrative content — but be aware that loop rate will be lower than equivalent 20–35 second content. For the full Shorts strategy framework, see our Shorts strategy guide.
SMMNut Shorts Loop Rate vs Length Trade-Off: Loop rate (the percentage of viewers who re-watch a Short) decreases predictably with length. A 20-second Short with 30% loop rate means 30% of viewers spend 40 seconds with the Short — equivalent to a 12-second extended watch. The same 30% loop rate on a 55-second Short means viewers spend 110 seconds — a much higher re-watch commitment. YouTube’s algorithm rewards loop rate as a secondary positive signal. For creators who want to optimise for loop rate alongside completion rate, the 20–35 second length range provides the best combination of both signals. Tutorial and educational content that genuinely requires 45–59 seconds is exempt from this optimisation — the information density trade-off justifies the reduced loop potential.
Editing for Length — Practical Compression Techniques
Hitting the optimal length range (15–45 seconds) requires disciplined editing. Most unedited Shorts run 10–30% longer than they should. Compression techniques:
- Jump cuts: remove all pauses, filler words, and transitions between points. Jump cuts that feel jarring in long-form are entirely natural in Shorts, where viewers expect rapid pacing.
- Eliminate the intro: in long-form, you introduce yourself and the topic. In Shorts, the hook IS the intro. Starting with “hey guys, today we’re going to talk about…” before the actual hook wastes 3–5 seconds and increases swipe-away rate before the hook even lands.
- Single-point structure: a Short should make one point, demonstrate one technique, or tell one story. Multiple-point Shorts almost always exceed the optimal length range and produce lower completion rates than single-point Shorts.
- End at the resolution: there is no need for a conclusion, call-to-action, or summary in a Short. End the moment the point is made — the loop will drive re-watches if the ending creates a question about the beginning.
Testing Length Variants — A Scientific Approach
If you have a content concept that could work at 25 seconds or at 55 seconds, the fastest way to find the optimal length for your specific audience is to test both versions:
- Edit the concept down to its 25-second essential core. Publish it.
- After 500+ views, record the Average Percentage Viewed and swipe-away rate at 3 seconds.
- Edit the same concept as a fuller 50-second version. Publish it.
- After 500+ views, compare both versions on the same two metrics.
- The version with higher Average Percentage Viewed and lower swipe-away is the better length for your audience and content type.
Do not make this decision based on personal preference about content depth. Make it based on the analytics data — which length your audience actually watches to completion. For the full Shorts content framework, read our Shorts strategy guide.
SMMNut Shorts Length Decision Framework: When deciding between a shorter and longer version of the same content concept, the default should always be the shorter version — compression improves completion rate ceiling and loop rate potential. The exception is tutorial and educational content where information density genuinely requires longer runtime to deliver value without sacrificing steps or context. For every second added beyond 30s, ask: does this second add information that the viewer cannot infer, or does it exist because it was easier to leave in than to edit out? Every second that passes the ‘can I infer this?’ test should be cut.