The graph is a symptom
A dip at four minutes is rarely caused by anything at four minutes. It is caused by a story that started too late, a section that repeated something the viewer already understood, or a promise the video made in the first minute and never came back to.
By the time the graph exists, the decisions have already been made. That is why we treat retention as an editing problem rather than an analytics problem.
Three questions on every cut
- What is the viewer still waiting to find out?
- Has this idea already been said once?
- Is the next thing better than the thing they just watched?
If the answer to the first question is nothing, the video has stopped moving. If the answer to the second is yes, something has to go, no matter how good the footage is.
A cut is not where footage ends. It is where attention is asked to continue.
Momentum is not speed
Fast cutting is often mistaken for pace. Cutting faster on a section with nothing to say only makes the emptiness arrive sooner. Pace comes from the order of information: give the viewer a reason to stay, then answer it slightly later than they expect.
The result is a video that feels shorter than it is. That is the only retention trick that survives contact with a real audience.