YouTube has been openly transparent that watch time, not views, is the dominant ranking input. Watch time is retention rate multiplied by video length, weighted by session contribution. Retention is the lever a creator actually controls; length is a tradeoff. This guide is the playbook for the retention half of that equation. For the scripting side, see our how to script YouTube videos for retention guide; for the analytics layer, see YouTube analytics explained.
Every video produces a retention curve in YouTube Studio. The shape of the curve, more than the average, tells you what to fix.
The cliff (bad). A near-vertical drop in the first 30 seconds, then a slow flat trail. The video lost everyone who was not already a subscriber in the hook. Fix: rewrite the first 15 seconds.
The slide (average). A smooth decline with no dramatic drops. The video has no killer moment but also no killer flaw. Fix: add a rehook at the 60-second mark.
The plateau (good). A drop in the first 15 seconds, then a flat or slightly declining line. Viewers who survived the hook stayed for the rest. Fix: tighten the hook.
The spike (great). A flat line with one or two upward bumps where viewers rewatched a segment. The bumps mark the moments that earned shares and re-engagement. Fix: do more of those moments.
The wave (excellent). A flat line with multiple rehook spikes throughout, each one corresponding to a curiosity beat. This is the structure of top-decile creators (MrBeast, Veritasium, Tom Scott). Fix: study the moments that did not spike and figure out why.
Aggregating retention data across thousands of channels, three drop-off zones dominate:
0 to 15 seconds: 20 to 35 percent of viewers leave. They watched the title and thumbnail, clicked, and made an instant decision. This is the hook zone. If your video does not answer "what is this and why should I care" in this window, retention is permanently capped.
60 to 90 seconds: another 10 to 20 percent leave. These are viewers who were on the fence after the hook. They give the video a minute to deliver value; if it has not, they abandon. This is the rehook zone.
After 90 seconds: drop-off rate flattens dramatically. Viewers who survived past 90 seconds are committed; they usually watch to the natural ending or the next rehook. The job past 90 seconds is rhythm and curiosity-loop maintenance, not retention from scratch.
This is why MrBeast openly says he edits the first minute more than the rest of the video combined. The math agrees: a 5 percent improvement in first-minute retention compounds across every viewer who reaches the rest.
The structure that top-decile creators converge on, across niches:
Hook (0:00 to 0:15). State the question or promise. Preview the payoff visually. Skip the channel intro, the logo flourish, the welcome-back greeting, the "in this video we will" sentence. Examples: "I tried 7 productivity apps and one of them genuinely changed how I work." "The answer to this is not what most fitness YouTubers tell you." "Here is what happens when you reverse-engineer a Drake song."
Rehook (0:50 to 1:30). Reframe the promise with new urgency. New angle, surprising twist, or doubling down on the original stakes. Examples: "But here is what nobody told me before I started." "And the reason this matters now, in 2026 specifically, is..." "Before we get to the answer, there is one thing you need to understand."
Body (1:30 to T-2:00, where T is total length). Deliver the promised content with rehooks every 90 to 180 seconds. Each rehook reopens a smaller curiosity loop: "And it gets weirder." "Here is where I was wrong." "What I expected was X; what actually happened was Y."
Payoff (T-2:00 to T-0:30). The original promise from the hook is delivered. Resolve the curiosity loop. This is the moment that earns shares and rewatches.
Ending (last 30 seconds). One sentence of closure, one sentence of next-step (subscribe, watch the related video, try the thing). Skip the outro animation and the long sign-off. The last 30 seconds also feed YouTube's end-screen and suggested-video ranking.
B-roll every 3 to 5 seconds in the first 60. Visual change is what keeps the eye on the screen. Even a single zoom-in or a graphic overlay counts as a visual cut. The first minute should have 12 to 20 visual changes; the rest of the video can settle into a slower 1-per-10-seconds rhythm.
Cut the pauses. Every "um," every pause longer than 0.3 seconds, every restart. Descript and CapCut both have automated silence removal that gets you 80 percent of the way; do the final pass by hand.
Pattern interrupts at the 60-second mark. Music cue, sound effect, hard cut to a different angle, a graphic, a callout. Something that signals "we are entering a new beat." This catches the 60-second drop-off zone.
Pace the rehooks. If your video is 8 minutes, you need 3 to 4 rehooks (at roughly 1:00, 2:30, 4:30, 6:30). If your video is 20 minutes, you need 8 to 10. Mark them in the script before you record; do not try to add them in post.
The cheapest retention edit is one that happens before the camera turns on. Paste your script (or any video transcript) into the LensPOV Watch-Time Estimator and the tool grades the script across hook strength, rehook density, pacing, and ending. Iterate on the script until the score crosses 70, then record.
This catches three problems before they become recording problems: hooks that are too long, videos that lose curiosity in the middle, and endings that fade instead of land. Each of those is a 10-to-20-minute script edit. Each of those is also a 5-percent retention swing on the final video.
Related: How to script YouTube videos for retention and Best video editing software for YouTube.
One email, every week. Retention tactics, growth analytics, and the one insight worth reading.
Above 50 percent average view duration is good. Above 60 percent is excellent. The median YouTube video sits around 40 to 45 percent. Long-form videos (over 15 minutes) that hold above 50 percent retention typically perform extremely well in the algorithm because absolute watch time is the dominant ranking signal. Short videos (under 5 minutes) need higher retention (60 to 75 percent) to compete on watch time.
The biggest drop is in the first 15 seconds. Across YouTube's aggregate data, 20 to 35 percent of viewers leave before the 15-second mark. The second biggest drop is at the 60-second mark, where viewers who are not committed to the video tend to abandon. After 90 seconds, drop-off rates flatten significantly. This is why the first 15 seconds (the hook) and the 50-to-90-second window (the rehook) matter more than any other segment.
YouTube favors high watch time, which is the product of retention rate and video length. A 20-minute video with 50 percent retention earns 10 minutes of watch time per view; a 5-minute video with 80 percent retention earns 4 minutes. Long videos earn more watch time per view when retention holds, but they need to actually hold retention. Padding a short video to 10 minutes with filler tanks retention and hurts ranking. The right length is the length where retention stays above 50 percent.
Under 15 seconds, ideally under 10. The hook is the segment from second zero until the viewer is committed to watching the rest. It should answer three implicit viewer questions: what is this video about, why should I care, what will I learn or see if I keep watching. Skip the long intro animation, the channel logo flourish, and the welcome-back greeting. Those are retention killers.
Hook-rehook-payoff is a three-beat structure used by high-retention creators. The hook (first 10 to 15 seconds) opens a curiosity loop and previews the payoff. The rehook (between 60 and 90 seconds) reframes the original promise with new urgency or a surprising twist, catching viewers who would otherwise abandon at the 60-second drop. The payoff is the moment where the original promise is delivered, ideally in the back half of the video. Strong videos have multiple rehooks every 90 to 180 seconds throughout, each one reopening a smaller curiosity loop to carry attention to the next.