What good looks like
| Video length | Healthy average percentage viewed |
|---|---|
| Under 2 minutes | 70%+ |
| 2 – 5 minutes | 55–70% |
| 5 – 12 minutes | 45–55% |
| 12 – 25 minutes | 35–45% |
| Over 25 minutes | 25–35% |
These are guides, not targets. A 40-minute documentary at 30% delivers 12 minutes of watch time per viewer, which is excellent. Judge total watch time and the shape of the curve together.
Curve shapes and their fixes
| Shape | Diagnosis | Fix |
|---|---|---|
| Cliff in first 15 seconds | Intro does not match the promise | Delete the intro; start mid-action |
| Cliff at 30–60 seconds | Hook landed, then the video slowed | Move the first payoff earlier |
| Gentle steady decline | Normal | Nothing |
| Steep steady decline | Pacing | Cut 20% of runtime, tighten every transition |
| Mid-video cliff | One specific segment fails | Find the timestamp, cut or rework that section |
| Sawtooth spikes | People skipping ahead | Add chapters; the structure is unclear |
| Upward spike | Rewatched moment | Make more of that; it is your strongest material |
| Flat then end drop | Healthy completion | Move end screen earlier |
Writing a hook that holds
- 1Open on the outcome, the conflict, or the strongest visual — never on 'hey guys'.
- 2Restate the title's promise in the first ten seconds so the viewer knows they are in the right place.
- 3Give one concrete reason to keep watching that is not simply 'stay tuned'.
- 4Cut branding, animated logos and sponsor reads out of the first 60 seconds entirely.
Relative vs absolute retention
Studio can show how your video's retention compares with similar videos of similar length. Use it as a sanity check: a curve that looks bad in isolation may be perfectly normal for the format. Use the absolute graph to find timestamps, and the relative view to decide whether there is a problem at all.