Retention rate

The curve is the most useful screen in any analytics dashboard, because it does not tell you that a video underperformed. It tells you the second it happened.

How to read the shape

• A cliff in the first one or two seconds. The hook failed. Nothing after it matters, because almost nobody saw it.
• A cliff at four to six seconds. The hook worked and the second beat did not arrive. Usually a pacing problem. See pattern interrupt.
• A steady slope with no cliff. Normal decay. The video is working; the way to improve it is to make it denser, not shorter.
• A bump upward. People rewound or the video looped. Whatever happens at that point is worth repeating.
• A cliff just before the end. The payoff arrived too late, or the video kept going after it.

Retention against length

A 30-second video at 70 percent retention and a 90-second video at 40 percent produce roughly the same average view duration, but they are not the same result. Short-form feeds respond to the share of the video watched, so the shorter video usually gets pushed harder while the longer one collects more total watch time.

The practical loop

Post, open the curve, find the first cliff, change only the thing at that timestamp, post again. Changing five things at once tells you nothing.

MakeViral renders variations of the same brief in one run, so a hook test is one session rather than a week. See how to go viral on TikTok.