retention-lab

Your retention data,
decoded.

Deconstruct any YouTube video's retention patterns with AI. Get your next winning hook, cold-open, and outline — calibrated to your real audience data.

Join the waitlist → ★ Star on GitHub

Open-source CLI available now · Hosted version coming soon

Live demo

What the tool actually outputs.

Analyzed with retention-lab v0.1 (open source). Input: one YouTube URL.

Veritasium · 20:57 · 18.4M views

The Surprising Secret of Synchronization

Retention shape (proxy)

0%50%100%

Key moments

  • 0:00-0:15

    hook

    Opens with a stated impossibility — viewer wants to see the resolution. Classic hook pattern.

  • 3:45

    retention dip

    Retention dips ~8% here — matches a slow mathematical explanation. Consider a visual metaphor or callback.

  • 8:20

    recovery

    Reveal moment + camera cut to reaction shot recovers retention.

Edit suggestions

  1. Tighten 3:30-4:15 by 20%. The math detour loses ~8% of viewers with no recovery until 8:20.
  2. Add a visual callback to the opening impossibility at 6:00 to re-anchor the narrative thread.
  3. Your cold-open works — replicate the 'stated impossibility → reveal' pattern in your next video.

Run it yourself: pip install retention-lab && retention-lab deep-dive <url>

Join the waitlist

Be first when the hosted version ships.

Early supporters get 30% off their first 3 months.

No spam. We'll email you once when the hosted version is ready.

Pricing (planned)

Simple, creator-friendly.

Free

$0

  • · Local CLI forever
  • · Community support
  • · Open source

Pro

$29/mo

  • · Real retention curves
  • · AI Copilot drafts
  • · Web workspace
  • · Email support

Studio

$99/mo

  • · Everything in Pro
  • · Managed agents (later)
  • · Unlimited channels
  • · Priority queue