Ideas worth stealing.

How outliers are scored

For each channel we pull its 40 most recent long-form uploads (Shorts and members-only videos excluded) and take the median view count. Every video's multiple is views ÷ channel median.

• 3× and up is the working definition of an outlier here, the amber badge. It means the title or idea did the work, not the channel's size.
• Views per day = views ÷ days since upload, so a two-week-old video can compete with a two-year-old one.
• Engagement rate = (likes + comments) ÷ views, only for videos where likes were fetched (top outliers and anything from the last 3 weeks).
• Upload dates come from YouTube's relative labels ("3 weeks ago") unless a verified date is shown, so recent-window filters are approximate by a few days.

Refresh by rerunning scrape.py → details.py → build.py in the project folder and republishing.

Add video