How to read a retention curve

The first thing to know is that the opening drop is not a problem. Every video on every platform loses people in the first moment, because a share of any audience was never going to watch anything. Reading that as a failure sends people back to rewrite a hook that was fine.

What matters is the shape after that. Our report marks the steepest drop after the opening for exactly this reason: pointing at the start would only tell you the video began.

A cliff in the first two seconds. The opening frame or first line is not doing its job. Fix the hook, not the edit.

A steady, gentle slope to the end. This is the healthy one. People are leaving at the rate any audience leaves. If the average watch percentage is decent, the video is working and the ceiling is distribution, not craft.

A single sharp step in the middle. Something specific happened at that second: a slow shot, a topic switch, an ad-shaped sentence, a pause where the energy dropped. This is the cheapest thing in video to fix, because you cut three seconds and republish rather than reshooting.

A drop right before the payoff. The build-up ran too long for what it delivered. Either shorten the build or move a piece of the payoff earlier so the promise stays credible.

Average watch percentage is more useful than raw watch time, because it scales with length: forty per cent of a fifteen-second clip and forty per cent of a ninety-second one mean the same thing about your pacing, while the second one has six times the watch time.

It also travels between your own videos in a way it does not travel between accounts. Comparing your fifty-two per cent to someone else’s eighty is meaningless when their videos are a third as long and shot for a different audience. Comparing this week’s fifty-two to last week’s forty-one on the same format tells you something real.

Short clips that loop can push watch percentage above a hundred, which looks like a triumph and is sometimes just a video that ends confusingly enough to be watched twice. Real loops are designed: the last frame matches the first, and the rewatch is voluntary.

Before you optimise for looping, check whether the loop is deliberate. Deliberate loops raise watch time; accidental ones raise it once and cost you a share, because nobody sends a friend a video they had to watch twice to understand.

Platform analytics only draw the curve after publishing, which means the first version of a video is always the experiment. Upload the clip to the checker and it predicts the curve from the video itself — cut rhythm and audio included — and marks the steepest drop after the opening, so the second worth re-cutting is identified while re-cutting is still free.

Then publish and track the post. The predicted range gets measured against real views, which is the part most tools skip: a forecast nobody checks is just an opinion with a number attached.