YouTube Outlier Finder

Point it at any YouTube channel, get back the channel's recent videos scored against its own median views — so you can see at a glance which videos overperformed (2x, 10x, 50x the channel's normal) and which flopped.

Outliers are the single strongest signal in content research: a video that does 10x its channel's baseline earned that reach through packaging (idea + title + thumbnail), not subscriber count. That's the thing worth studying and adapting.

Use cases

• Content research — find what's actually working in your niche right now, normalized for channel size.
• Title & thumbnail studies — build swipe files from proven overperformers, not lucky big channels.
• Competitor analysis — see which of a competitor's bets paid off and which flopped.
• Small-channel scouting — a 5k-sub channel with a 40x outlier is a trend alarm no view-sorted list will show you.

Input

{
  "channel": "@MrBeast",
  "maxVideos": 30,
  "minMultiplier": 0
}

Field	Type	Notes
channel	string (required)	Handle (@MrBeast), channel ID (UC…), or any channel URL.
maxVideos	integer	Recent videos to analyze, up to ~30 (one page of the Videos tab). Default 30.
minMultiplier	integer	Only return videos at/above this multiplier. 0 (default) returns every analyzed video with its score.
proxyConfiguration	object	Preconfigured — residential proxy is used automatically.

Output

Dataset items sorted by multiplier, highest first:

{
  "videoId": "iYlODtkyw_I",
  "url": "https://www.youtube.com/watch?v=iYlODtkyw_I",
  "title": "Survive 30 Days Chained To A Stranger, Win $250,000",
  "views": 50000000,
  "publishedText": "4 days ago",
  "channelBaseline": 142000000,
  "multiplier": 0.35
}

• channelBaseline — the channel's median views across the analyzed videos.
• multiplier — this video's views ÷ baseline. ≥2 is an outlier; ≥10 is a monster.

How it works

Fetches the channel's Videos tab through Apify residential proxy and parses the embedded page data — no browser, no YouTube API key, no quota. View counts are YouTube's compact numbers ("1.2M"), which is plenty of precision for multiplier math.

Limitations (honest ones)

• Analyzes the ~30 most recent longform uploads (one page, no scroll). Shorts live on a separate tab and are not included.
• Baseline is the median of those same recent videos — for channels that recently blew up, the baseline reflects the new normal, not their history.
• One run = one channel. Loop over channels to scan a niche.