youtube-strategy

name

youtube-strategy

description

Turn a pain point, topic, or rough idea into a compelling YouTube package — title, thumbnail, intro hook, motivation/story, outro, and video description — for Shaw's channel. Use any time Shaw is writing, brainstorming, or iterating on YouTube content. Triggers include "title ideas", "punch up this title", "thumbnail ideas", "design the thumbnail for X", "package this video", "write the intro/hook", "draft the motivation", "write the outro", "iterate on the script", "Ed Lawrence-style intro", "write the (YouTube) description", and casual mentions like "title this for me", "thumbnail this", "I'm making a video about X", or "help me write the open". Do NOT use for LinkedIn posts (use linkedin-post-writer), newsletters (use newsletter-writer), or sales copy (use conversion-copy or sales-letter-writer) — this skill is specifically for YouTube packaging and scripting.

YouTube Strategy

Generate compelling YouTube packages — titles, thumbnails, hooks, motivation/story sections, and outros — that earn the click and the watch from Shaw's target avatar without being clickbait or punky.

Avatar (canonical source)

The current avatar definition lives in the 2026 Q2 page (parent: Strategy) in Notion. Pull from there as the first move on every YouTube task — the avatar evolves quarterly, and assumptions baked into this skill will go stale.

What lives on the strategy page:

• The current ICP definition (industry, business size, role, psychographic)
• The five verbatim pain points, in customer voice
• Revenue benchmarks
• Mission and the "sell outcomes, not training" framing

When the strategy page disagrees with anything in this skill or its references, the strategy page wins.

The 3-filter framework

Every title, thumbnail, hook line, motivation beat, and outro line passes three filters:

1. A real avatar would search/click/pause-scroll for this. Discoverable, recognizable, relevant to their world.
2. It captures attention with at least 2 click-driving principles stacked.
3. It's specific enough to be ownable. A competitor, generic creator, or Anthropic's homepage couldn't sign the same line or use the same image.

For sentence-level work — "does this exact phrasing land?" — defer to the visualize / falsify / unsignable filter from the conversion-copy skill. That's the canonical home for the line-by-line filter. This skill applies it; doesn't restate it.

Iteration order (top-down)

Lock the higher-level decisions first. Each level constrains the next.

1. Title — the verbal hook
2. Thumbnail — the visual hook (paired with the title)
3. Script:
   • Intro / hook — first 10 seconds (~25 words)
   • Motivation / story — next 30 seconds (relevance + credibility)
   • Body — the substance the title promised
   • Section cards (optional, when the body is chaptered) — full-frame chapter slates at each section boundary; inherit the thumbnail's visual world
   • Outro — the next-video bridge (~10-15 seconds)

Skipping levels leads to wasted iteration. A locked title makes thumbnail concepts faster to generate; a locked title + thumbnail clarifies what the intro promises.

Visual beats are annotated inline

Shaw's scripts notate visual moments as sub-bullets at the line that triggers them:

• \[ANIMATION\] <what happens on screen> — motion graphics, B-roll overlays, morphs, builds
• \[CARD\] <what appears on the chapter slate> — full-frame section intro slates

The annotation describes content — subject, glyph, label, motion only if essential. Don't redefine the visual system at this level. Type, palette, lighting, and motion language are locked at the thumbnail (references/thumbnails.md) and propagate down to every card and animation. If a \[CARD\] and a \[ANIMATION\] land on the same beat and carry the same visual idea, drop one — spend each visual beat once.

Two non-obvious iteration moves

These come from real sessions. Both apply when the work feels stuck.

Iterate the first hook line alone, before the full hook

The first line of the intro carries the most weight — it has to anchor in real customer language and pass the filter on its own. Lines 2 and 3 keep shifting until line 1 is locked. Don't try to perfect a full hook in one shot when line 1 is still wrong. See references/intros.md.

Pull verbatim customer language before drafting

Hooks fail when language is invented; they land when it's real. Before generating hook candidates (or motivation copy), pull verbatim phrases from:

• ABA Calls (Notion) — sales call notes
• Workshop intake form responses (Google Drive)
• Testimonial form responses (Google Drive)
• Workshop one-pager (Notion)

Pull at minimum 3 verbatim phrases before drafting. Quote the source for each candidate so the grounding is auditable.

Routing — when to load which reference

Task	Reference
Generate / sharpen titles	references/titles.md
Generate / sharpen thumbnails	references/thumbnails.md
Write or iterate on the intro hook (Ed Lawrence-style)	references/intros.md
Write motivation/story section or outro	references/script-body.md
Write the video description (CTA, summary, links, chapters)	references/descriptions.md
Sentence-level filter on a single line	conversion-copy skill (canonical)

For paired packaging (title + thumbnail together), load both titles.md and thumbnails.md. For a full script pass, load intros.md and script-body.md.

Output format (high-level)

When the task spans multiple levels (e.g., "package this video"), generate top-down: title → thumbnail → intro → motivation → outro. Each level explicitly references the level above so the package coheres.

When the task is one level only (e.g., "give me title ideas"), use the output format inside the relevant reference doc.

Channel positioning context

Shaw's channel is in transition. The existing audience (~100k+ subs) skews technical/builder. The new avatar from the strategy page skews founder/CEO. Most videos in this period straddle both. Default to the strategy-page avatar; fall back to the existing audience when the topic is fundamentally technical (a tool review, a code walkthrough, etc.).

The thumbnail and title visual/verbal language should always match the new avatar — even on technical-leaning content — because that's where the channel is going. See references/thumbnails.md for the strategic shift (Technical Implementation → Strategic Authority).

When in doubt

Prefer the choice that:

• Uses a real number over a vague claim
• Names a specific entity (company, person, mechanism, logo) over a category
• Is grounded in language a real founder actually used (in a call, in an intake response, in a testimonial)
• Stacks fewer principles cleanly over more principles awkwardly
• Respects the avatar's intelligence and time

Lead with substance. The avatar is sophisticated. They click on packages that promise real, specific information — not on packages that try to manipulate them.