Capafy logo Capafy
开始使用
Retention Autopsy: Why Your Short Lost Viewers

Retention Autopsy: Why Your Short Lost Viewers

by: One File Tools
For a short you already posted. Send the video plus a screenshot of its retention graph (or your average % viewed) and get the curve redrawn with the drop-offs marked, the frame on screen at each drop, every leak ranked, a cut list with the new target length, and a verdict printed next to its threshold.
#视频#社交媒体#分析
评分
需要更多评价
订单数
0
如何使用
在 Capafy 上运行
也可用于外部应用
创作者提供
Claude Sonnet 5

Where your short lost people - and what was on screen when it happened

View counts on your own videos cluster together whether a video was good or not, because every upload gets a similar first test. What decides whether a video travels further is how much of it people watch. Your platform already shows that as a graph. It does not show why the line falls where it falls.

Retention Autopsy lines the graph up against the video itself. Every drop gets a timestamp, the frame that was on screen at that second, and a reason. Then it hands you a cut list.

What it looks like

autopsy_report.webp

A report built from a sample video and sample numbers: the curve with three drops marked, the frame at each drop, the leaks ranked and labelled structural or surface, and a cut list that takes 35.6 seconds down to about 25.

What you send

  • The posted video (mp4, mov, webm)
  • A screenshot of the retention graph - YouTube Studio Audience retention, TikTok Retention rate, or Instagram Retention
  • Average percentage viewed (or average view duration)
  • (optional) the script, the platform

Without a graph, the report draws your average as a flat line and says no curve shape was given. It never invents one.

What you get

  • A verdict with its threshold printed beside it: 75%+ Strong, 50-75% PASS, 35-50% REWORK, under 35% REBUILD (working thresholds; no platform publishes an official pass mark)
  • The curve, the drop-offs, and the frame on screen at each one
  • Every leak ranked, labelled structural (order, premise) or surface (pace, text, pauses), each with its fix
  • A cut list - cut, compress, move earlier - with the new target length, checked so the numbers add up
  • What to record next time, so the next autopsy compares like with like
  • retention-autopsy.png, one page

What it does not do

  • No account linking, no analytics access, no link downloads. It works from what you send.
  • It never fills a missing number with an estimate, and never promises views, reach or income.

Before the next one

Hook Teardown checks a new opening frame by frame before you post: https://capafy.ai/agent/x/9864413431


外部 API

PyPI (pypi.org)

View more