A script tells you where a video will lose viewers before you film it, if you know how to read it. Most creators do not — they read for accuracy and flow, not for exits. The Viewer Retention Script Analyzer reads for exits: it marks every danger zone, estimates the drop-off probability at each, rewrites the zones, scores the script's engagement, and estimates the retention gain from applying the rewrites.
This guide explains what a danger zone looks like on the page, how the rewrites work, and how to use the analyzer as the last step before recording.
Danger zones on the page
A paragraph of context before any payoff. A list of caveats. A section that explains a method before showing a result. Three sentences that say the same thing. A transition that announces "now let's move on to" — which tells the viewer a natural stopping point has arrived. A sponsor read placed before the first payoff. The analyzer marks each with a drop-off probability based on where it sits and how long it runs.
The rewrites
Each danger zone gets a rewritten version that keeps the information and removes the exit: result first then method, caveats folded into one line, the transition replaced with a forward tease, the sponsor moved. Rewrites preserve your voice — they restructure, they do not paraphrase into a different register.
Last step before recording
Paste the final draft. Apply the rewrites you agree with. Check the engagement score moved. Then record. The analysis takes seconds and catches the structural mistakes that no amount of editing energy fixes after the fact.
Tips
- Paste the whole script, including the intro and outro; the first and last minutes are where the most fixable zones live.
- Mark your intended timestamps or section breaks; the probability model uses position.
- Run it a second time after applying rewrites; new zones sometimes appear at the seams.
Common mistakes
- Analysing a summary instead of the script — Exits live in the sentences.
- Rejecting every rewrite to protect voice — Restructure, then restore your phrasing.
- Skipping it for "short" videos — A six-minute video has the same first-30-second cliff.
Worked example
Input: 2,600-word grinder-comparison script.
What the report returned: Engagement 61. Zones: 0:00–0:50 (setup before promise, 34% drop), 4:10–5:30 (method before result, 22%), 7:00 sponsor before payoff (18%), 10:40 "so to wrap up" (15%). Rewrites for each. Projected engagement 79; est. AVD +9 points.
Frequently asked questions
How accurate is the drop-off probability?
It is a structural estimate calibrated on typical retention shapes; it is reliable about where the risks are, less exact about the percentages.
Does it change my writing style?
It restructures; phrasing stays yours. You can accept or reject each rewrite.
Is this different from the Retention Heatmap?
Heatmap works from a structure description; this works from the actual script text and rewrites it.
Can it analyse a transcript of a published video?
Yes — useful for diagnosing why a video underperformed and planning the next.