15% AVP on long videos beats 50% on short ones
Creators shifted focus to absolute watch time as the metric that moves the algorithm. Retention graphs show where minutes are gained or lost.
Blog
Notes from the edit bay
Retention benchmarks, hook formulas, and script fixes. The same patterns HookProof checks for, written out so you can apply them by hand.
TikTok Shop scripts that hit 10k orders with timed demos
Break down the exact 60-second structure creators use to turn psychological hooks and real-life demos into repeated TikTok Shop sales without hard pitches.
vidIQ retention graphs versus TubeBuddy SEO and Retti_AI scripts
vidIQ now maps drop-offs from any link while TubeBuddy skips curves entirely and script tools like Retti_AI train on 5000 graphs to hit 68 to 74 percent retention before filming starts.
AI Script Drafts Require 60 Percent Human Cuts
Creators report keeping only 40 percent of AI script lines after voice matching and editing passes to remove generic phrasing and lost quirks.
Markup prompts that fix AI voice scripts
One prompt template adds pauses, caps, and emotion tags to AI drafts so ElevenLabs and Murf output matches a creator's cadence instead of flattening it.
AI video scripts that don't sound like AI
The tells viewers notice in AI-written scripts, how to voice-match a model with your past transcripts, and a de-slop checklist to run before recording.
Hook formulas that survive the first 3 seconds
Cold opens, specific numbers, contrarian claims, and open loops, with before and after rewrites, plus the math on what 'Hey guys, welcome back' costs.
What good YouTube retention looks like in 2026
Typical retention curves by video length, what 50% at 30 seconds actually means, and how to read your own graph before you panic about it.