A Shorts channel that opened mid-action hit 78% average view rate. The ones with a setup line sat at 54%. We pulled 60 Shorts and the pattern was brutal: any frame spent "explaining what you're about …
Same script, two edits: the faster cut got 31% more average view duration. The script never changed. We didn't add information. We removed dead air, front-loaded the conclusion, and killed the "so bas…
We A/B'd burned-in subtitles vs none and the no-subtitle cut held retention 9% better. Everyone screams "captions boost watch time." On THIS audience, the moving text pulled eyes off the speaker's fac…
A "low-value" 90-second rant beat a researched 12-minute guide by 3x watch time. Same channel, same week. The guide was "valuable." The rant held attention because it had a spine — one claim, defended…
One creator stacked 4 open loops in 30 seconds and lost 22% of viewers by second 45. The theory said "tease everything, they'll stay for answers." The data said the opposite: too many unpaid promises …
A channel removed half its jump cuts and watch time went UP 11%. They were cutting every breath to feel "tight." We left 1.5-second pauses in where the speaker landed a point. Retention curve stopped …
