mbfinotti/advertising-skills / skills/ad-hook-analyzer
ad-hook-analyzer
VerifiedScore and force-rank the openings of candidate video ads - from a script, transcript, storyboard, shot list, or a description of a finished cut - to decide which hooks deserve test budget, returning a ranked shortlist within the batch rather than an absolute performance prediction. Use whenever the user mentions a hook, the first 3 seconds, an ad opening, hook rate, thumbstop, attention scoring, or asks which hook to test or whether a hook works - even if they never say 'hook analyzer'. Covers B2B and B2C on any platform. Ends at the ranking: writing the scripts themselves is mbfinotti/advertising-skills@ugc-ad-scripts and static copy is mbfinotti/advertising-skills@ad-copy-variants.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 18.9 KB |
| evals/evals.json | config | 39.4 KB |
| references/examples.md | markdown | 12.1 KB |
| references/hook-patterns.md | markdown | 6.1 KB |
| references/platform-notes.md | markdown | 6.2 KB |
| references/scoring-rubric.md | markdown | 10.0 KB |
references/hook-patterns.md
View on GitHub# Hook Patterns - Vocabulary and Anti-Patterns
**How to use this file.** The taxonomies below are a generative and diagnostic vocabulary: use them to name what a candidate opening is doing, to spot that a batch is five flavours of one mechanism, or to suggest what a revision could try.
**Never use them as a scoring axis.** They overlap, they mix executional formats with psychological mechanisms and narrative structures, and none has falsifiable category boundaries - the same ad can be tagged under several types depending on who tags it, which is exactly why human hook-taxonomy tagging has poor inter-rater reliability. A "this batch covers 6 of 8 hook types" claim is not evidence of anything.
## Motion "Hook Tools" - eight psychological types
Source: Motion (motionapp.com, current 2026), framed as "the psychological triggers that top creative strategists use to stop the scroll." Named mechanisms, not measured categories.
1. **Contrarian** - challenges something the viewer believes is true.
2. **Identity Call-Out** - names the viewer's role, situation, or tribe directly.
3. **Confession** - the speaker admits something costly or embarrassing.
4. **Pain Agitation** - opens on the problem, sharpened.
5. **Curiosity Gap** - opens a question with a defined shape that the ad resolves.
6. **Loss Aversion** - frames what the viewer stands to lose by not acting.
7. **Unspoken Truth** - says what the audience thinks but nobody in the category says.
8. **Pattern Interrupt** - an unexpected visual or framing that breaks the feed's texture.
## Motion 2026 empirical top hook/headline categories
Source: Motion 2026 Creative Benchmarks - 578,750 creatives, 6,015 accounts, $1.29B in Meta spend, 1 September 2025 - 1 January 2026. **Caveat Motion itself states**: the window spans BFCM and the holidays, so offer/urgency/newness framings are seasonally inflated and the ranking would differ in another season.
Top-performing categories by hit rate: **Offer only** (highest, ~9.29%), Storytelling, Question, POV, Listicle, How-to, Explainer, Curiosity, **Confession** (~8.74%), Bold claim. Note what the top of the list is: the plain offer, not cleverness - a useful corrective when a batch is all conceptual and nothing states the deal.
## Mark Jung - structural B2B hook types
Source: Mark Jung (Authority), B2B LinkedIn taxonomy via Goldcast. Structural rather than psychological - these describe _which layer of the opening does the work_, useful for B2B where UGC/unboxing executional hooks rarely fit:
- **Composition hook** - the framing/visual arrangement of the shot stops the eye.
- **Physical hook** - a physical action or object in motion.
- **Audio hook** - a sound or line that arrests (remember: fails muted feeds - pair with a subtitle layer).
- **Subtitle hook** - the on-screen text line carries the stop.
Jung's stated rule - "You need to have 80% of your time go to your hook" - is a practitioner's emphasis claim, not a measured allocation.
## Anti-pattern list
Name the pattern when diagnosing; a named pattern ("this is a generic category open") is more actionable than "this is weak."
| Anti-pattern | What it looks like | Why it fails |
| ----------------------------------------- | ---------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Logo-first** | Logo card, brand jingle, or "we're excited to announce" as the opening beat | Signals "this is an ad" at the exact moment the viewer decides whether to leave; spends the highest-attention seconds on the advertiser, not the viewer |
| **Audio-dependent** | The promise lives in the voiceover or sound design; first frame + text say nothing | Feed video autoplays muted on Meta and LinkedIn - the meaning never arrives for muted viewers |
| **Slow build** | Setup, greeting, or context first; the real hook sits in sentence three | Attention decays to roughly half the audience by second 2 - the "real" hook plays to a fraction of the viewers who saw second 1 |
| **Generic category open** | "The smarter way to X," "revolutionize your workflow" | Interchangeable with any competitor; gives the viewer's brain nothing checkable to grab |
| **Hook/body mismatch** | Arresting promise, unrelated body and offer | The documented high-hook-rate/mediocre-ROAS failure: the scroll stops, nothing downstream converts, and the metric looks great while the ad loses money |
| **Cosmetic-only variation** | Same visual, new first line, shipped as a "new hook" | The platform does not treat it as a different variation (Savannah Sanchez) - the test cell is an illusion and the batch's real concept count is lower than it looks |
| **Clickbait that stops the wrong viewer** | Universal-curiosity spectacle unrelated to the buyer | Maximises raw hook rate while attracting unqualified attention - the pattern behind Barry Hott's warning that a higher hook rate does not mean the ad performs better |
Diagnosis procedure:
1. Check every candidate against this table _after_ the gates.
2. Name any pattern found in the scorecard's band rationale.
3. Pair each named pattern with the **cheapest rung of the skill's Raising a Rank ladder that actually removes it** - a text rewrite where the wording carries the fault, a re-cut or a reshoot where the footage itself does.
That revision is the candidate's "raise-the-rank" note.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the work | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixes | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation code | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |