Reports

Eight subagent reports

Each agent ran against a fixed slice of the data with a defined method. Open any to see the underlying evidence — every claim cites a video ID.

algorithmic-positioning

Algorithmic Positioning

Cadence, length, shorts, playlists, end-screens — the structural scaffold YouTube reads as signals.

The one-sentence positioning the audience would actually write, and where the channel breaks its own recipe.

content-forensics

Content Forensics

Which videos overperformed against the channel's own baseline, and what variables they shared.

SEO + Packaging

Titles, descriptions, tags, thumbnails — the pre-click layer.

Hooks and Retention

Opening patterns that worked vs the ones that didn't, top decile vs bottom decile.

audience-community

Audience and Community

Who watches, what they ask for, what they love — read from comments.

niche-competitor

Niche and Competitors

Peer channels mapped, and the two-way content gap.

growth-experiments

Growth Experiments

Twelve prioritized experiments. Each with hypothesis, success metric, and a kill criterion.

Source data acquired via yt-dlp. Channel under analysis: @thedossier_in · Snapshot date: 2026-05-17 · 60 of 80 videos sampled (YouTube bot challenge blocked the remainder). Every metric traces back to a file in data/.