Sources first. Scores with context.

Learn helps filmmakers find useful videos from a curated creator set. Our recommendation score organizes discovery; it does not certify that a creator’s claims are correct.

Who makes the notes?

The original videos belong to the credited creators. ReelStack uses AI to extract takeaways and workflow steps from source material. A generated summary is not independent testing or human review. Notes link back to the original source, with timestamp links where available.

How recommendations are calculated

Current formula: radar-v5. While measured momentum is unavailable, performance contributes 65%, freshness 25% and engagement 10%. Once sufficient momentum history exists, 20 percentage points move from performance to momentum. Relevance determines inclusion, not additional score points.

  • Performance: compares observed views with the creator’s recent median, switching to a comparable-age baseline when enough observations exist. Small view samples receive less credit.
  • Freshness: uses a 14-day half-life. An older video can still rank well through performance.
  • Momentum: uses measured view changes between snapshots, compared with channel history. Missing data earns no bonus.
  • Engagement: uses likes and comments relative to views, with smoothing for small samples.

Top picks may reduce similar-topic repetition among closely scored videos. Other videos remain available below. “Just published” sorts by date; “Views vs. usual” sorts by the lifetime channel-relative multiple. The displayed score is rounded; ordering uses the precise value.

Limits of the numbers

Public views are a snapshot, not a live counter. Channel history is a bounded sample, and same-age comparison improves only as observations accumulate. The lifetime views-per-day average is different from views gained during the last day. A high score is not a promise of accuracy, instructional quality, compatibility or value for your project. Scores are ReelStack calculations, not YouTube ratings.

What becomes a search landing page?

Being in the feed is not enough. Standalone insights must be editorially approved, have traceable source evidence and practical application, and add something beyond a rewritten summary. Incomplete or unapproved video notes are kept out of the search index. We do not publish full transcripts as search pages.

Daily briefings combine source material and compare recent editions to reduce repetition. Automated quotation and originality checks are safeguards, not proof that every claim is true. Their pages disclose how they were made.

How to use Learn responsibly

Read the source before relying on a consequential claim. Verify current provider documentation, permissions, availability and costs. Test a representative shot before committing a whole production. Internal tool-profile links provide context; outbound partner links, where used, should be identified as such.

If a source or note is wrong, contact ReelStack with the learning URL, the disputed claim and the original source or timestamp.

Powered by ReelStack

Help keep this running

Your tip funds servers, models, and the time it takes to ship new tools faster. Set any amount below — every bit helps.