Swipe-Away Rate and What Your Shorts Are Losing

Swipe-away rate is the percentage of viewers who left a Short by swiping instead of watching it through. On a per-second basis it becomes a map, because each spike points at the exact moment the clip stopped earning attention. It is the fastest way to find a bad line without guessing.

I watch this number more than any other during a content sprint. It reacts in hours, and it reacts to the writing rather than to the mood of the feed. When a swipe spike lines up with a sentence, I know which sentence to cut. Nothing else in the dashboard gives feedback that fast, and speed matters when you are trying to improve across a whole batch.

Swipe-away is also the number I trust most when a clip is new, because the feed has not yet decided how far to push it. Early read, early fix, and the second wave of viewers meets a better clip.

A spike is a sentence, not a vibe

Every swipe happened at a specific second of the clip, and every second holds a specific line. If the spike sits at four seconds, the fourth line is the problem. Naming the culprit turns a vague sense that the clip drags into an edit you can make in two minutes.

I keep a text version of the script next to the graph and mark the spike. The line under my finger is almost always setup, a repeated idea, or a promise the viewer already stopped believing. After a few months of this I could guess the culprit before opening the graph, and the guess was usually the line I was least sure about while writing.

Viewers do not swipe away from topics. They swipe away from sentences.

Common causes of an early exit

Each cause has a different repair, so pairing a spike with the wrong one wastes a rewrite. A repeated idea gets cut. A dead pause gets tightened. A missing payoff gets moved earlier. The graph tells you the second, and the script tells you the cause.

• The opening line names a category instead of a claim.
• A pause after the hook long enough to feel like the clip ended.
• A tangent that serves the creator and not the promise.
• A visual that does not change for several seconds while the voice keeps talking.
• A second topic introduced before the first one paid off.

Lower it without touching the hook

A common mistake is to fix a mid-clip spike by rewriting the opening. The opening did its job, since the viewer stayed long enough to reach the middle. The repair belongs at the spike, where attention leaked, and moving the payoff earlier is usually the fix.

Shortening the clip can lower swipe-away on its own. A tight twenty-second version of a dragging forty-second clip often holds the same people further in. I keep a rule that no beat gets more than eight seconds unless it earns a longer stay, and most clips get tighter the moment I enforce it.

I also read the spike against the beat before it. A sudden exit often means the previous beat ran long and soured the viewer. Sometimes the fix is not the line where the swipe happened, but the one that made them ready to leave.

Key takeaways

• Read swipe-away as a timestamp, not a single score.
• Match each spike to the sentence underneath it.
• Repair the middle instead of rewriting a working hook.
• Try a shorter cut before you rewrite the whole script.