Retention curve
The line chart showing what percent of viewers are still watching at each second of a video.
The retention curve is the second-by-second graph of how many viewers are still present in your video. It starts at 100 percent at second 0 and decays over the length of the video. Every dip is a moment where a meaningful number of viewers swiped or clicked away, and every flat or rising section is a moment where curiosity is doing its job.
YouTube Studio exposes it as 'Audience retention'. TikTok and Shorts surface a simplified version in the analytics dashboard. The shape of the curve, not the absolute average, is what matters for diagnosing what to fix.
Three patterns recur across viral videos: a sharp drop in the first 10 percent (hook problem), a slow gradual decline (pacing problem), and sharp mid-video drops (content delivery problem). Each pattern requires a different fix.
Why it matters
The retention curve is the only honest feedback you get from your audience. View count tells you how well the thumbnail performed. The retention curve tells you whether the video itself deserved those views, and is the single best predictor of whether the next video will be pushed or buried.
How to improve it
- 01Identify the steepest drop and rewatch the 5 seconds before it. The fix is almost always in the 5 seconds before the drop, not at the drop itself.
- 02Aim for a curve that stays above 50 percent at the midpoint. Below 50, the algorithm reads the video as underdelivering.
- 03If retention rises during a section, study what you did and repeat that pattern in future videos.
- 04Re-edit and re-upload is rarely worth it on a single video. Apply the lesson to the next 3 uploads instead.
See your own retention curve in seconds.
Paste any YouTube, Shorts or TikTok link and our analyzer will surface the metric, the curve, and the move that lifts it next.
Related terms
Percent of impressions that keep watching past the first 3 seconds.
The average number of seconds a viewer spends inside a video before leaving.
AVD expressed as a percentage of total video length.
A sharp decline at a specific point on the retention curve.