Free Script

YouTube Outlier Finder Setup Guide

By Giuseppe Gaspari: Zio Advertising

One Python file, a free Google API key, and your first outlier report in about 10 minutes.

Download the script

outliers.py, the README, the MIT license, and the .env template. One zip, no account needed.

Download youtube-outlier-finder.zip

MIT licensed. No account, no watermark.

What's in the Zip

Unzip it and paste your key into .env.

# YOUTUBE OUTLIER FINDER - Setup
# What's in the zip:

youtube-outlier-finder.zip
└── youtube-outlier-finder/
    ├── README.md          # Full usage + the options table
    ├── outliers.py         # The scanner (one file, one dependency)
    ├── .env.example        # Copy to .env, paste in your API key
    ├── .gitignore
    └── LICENSE             # MIT

# Unzip and set up your key:
unzip youtube-outlier-finder.zip
cd youtube-outlier-finder
cp .env.example .env
# open .env and paste in your YouTube Data API key

Get a Free YouTube Data API Key

Google Cloud Console, about 3 minutes, no credit card.

# GET A FREE YOUTUBE DATA API KEY (about 3 minutes)

1: "Go to console.cloud.google.com"
2: "Create a project (or pick an existing one)"
3: "APIs & Services -> Library -> search 'YouTube Data API v3' -> Enable"
4: "APIs & Services -> Credentials -> Create Credentials -> API key"
5: "Click the new key -> Restrict key -> API restrictions -> YouTube Data API v3 only"
6: "Paste the key into .env as YOUTUBE_API_KEY=..."

note: >
  Free tier = 10,000 units a day. No credit card needed to turn the API on.
  This is a Google quota, not a Zio one - there is nothing to buy from us.

Run Your First Scan

Three example commands to start with.

# RUN YOUR FIRST SCAN
# Needs uv (docs.astral.sh/uv) - it runs the script with its one
# dependency, youtube-transcript-api, with no manual pip install.

# Long-form, last 90 days, channels under 100k subs:
uv run outliers.py "claude code" "ai agents"

# Shorts only, last 30 days:
uv run outliers.py "seiko mod" --shorts --days 30

# Also save transcripts of the top 5 hits (for script research):
uv run outliers.py "claude code" --transcripts 5

# Reports land in ./research/outliers/ as Markdown + CSV.

Useful Options

Narrow the scan by subs, views, date range, or format.

# USEFUL OPTIONS (full table in README.md)

--max-subs 100000   # skip channels bigger than this (default)
--min-views 5000    # ignore videos below this (default)
--days 90           # only videos published in the last N days (default)
--shorts            # Shorts (<=3 min) instead of long-form
--pages 2           # more search pages per keyword (50 results / 100 units each)
--top 25            # how many rows land in the report (default)
--transcripts 5     # also save transcripts of the top N videos
--no-median         # skip the channel-median check to save quota

That's the whole tool.

One file, one free Google API key, and a report in ./research/outliers/ every time you run it. No login, no credits to buy, no dashboard to check.

If you want this kind of research system built into your own content or marketing workflow, that is exactly the kind of thing I build for businesses.

Want this kind of system built for your business?

Research tools, dashboards, and AI workflows, built for your stack and owned by you. Book a call.

Book a Call

Built by Giuseppe Gaspari: zioadvertising.com | giuseppe.builds on YouTube