How to Review Attention and Retention Signals

An AI-generated attention heatmap is an editorial view of your video. Its colors or scores may flag a scene worth inspecting, but they do not measure what viewers looked at, felt, or skipped. For observed viewer behavior, use the audience retention report for your published video in YouTube Studio.

Start with the actual graph

Open YouTube Studio → Content → select a video → Analytics → Overview or Engagement → Audience retention. YouTube says the report is available at the video level and normally takes one to two days to process. The graph can show flat sections, gradual declines, spikes, and dips. A dip means viewers skipped or stopped watching at that point; a spike can mean rewatching or sharing, including rewatching because something was unclear. See YouTube's retention documentation.

Review one timestamp at a time

Write down the timestamp and watch the preceding and following 30 seconds. Compare the moment with the title and thumbnail promise. Did the video change topic, repeat an explanation, reach a natural ending, or introduce an unfamiliar term? List at least two plausible explanations before choosing an edit.

An AI heatmap may suggest a transition or visual that feels slow, but it cannot tell you how many viewers left there. Treat the suggestion as a hypothesis to check against the Studio graph, then test one concrete change in a future video of similar format.

Keep a simple audit log

Record the video, timestamp, observed Studio pattern, possible cause, and change to try. When the next video is live, compare its measured results with care; topic and audience differences can affect the graph. Our audience retention guide explains the signals in detail, and the video audit checklist covers the full review. You can also analyze a video with VidFeel for editorial observations.