Hook Architect

1. Purpose

Treats the hook (first 3 seconds of a video ad) as a standalone variable. Filters out down‑funnel noise — offer, landing page, targeting, body script — to answer one question: is the creative relevant enough to stop the scroll? Every other creative skill optimizes what comes after the stop; this one optimizes the stop itself.

Run when:

• Auditing a batch of new creatives to see which hooks are actually stopping scroll.
• fatigue-detector flags declining CTR — a hook audit disambiguates "hook fatigued" from "body fatigued".
• Briefing a new creative via creative-brief — hook‑architect contributes the recommended archetype.

2. Primary Metric — Hook Rate

Hook Rate = 3‑second video views / impressions

Benchmarked against the rolling account average over the trailing 30 days.

| Tier            | Threshold                                | Meaning                                                             |
| :-------------- | :--------------------------------------- | :------------------------------------------------------------------ |
| 🏆 **Elite**    | Hook Rate **> account avg + 1σ**         | Repeat and clone this hook archetype.                               |
| 🟢 **On‑par**   | Within **±1σ** of account avg            | Keep running; not a priority for iteration.                         |
| 🔴 **Below**    | Hook Rate **< account avg − 1σ**         | Flag for hook‑swap — the body may still be fine (see §3).           |

Account average and σ are recomputed each run from the last 30 days of video ads with ≥ 1,000 impressions each. Ads below the impression threshold are excluded from both the baseline and the scoring.

3. The Hand‑off — Hook × Hold Quadrants

A high Hook Rate alone can be misleading. Cross‑reference with Hold Rate to diagnose why a creative is or isn't working.

Hold Rate = 15‑second views / 3‑second views (retained attention among those the hook stopped).

| Hook Rate | Hold Rate | Diagnosis                                                  | Recommended Action                                                            |
| :-------- | :-------- | :--------------------------------------------------------- | :---------------------------------------------------------------------------- |
| 🟢 High   | 🟢 High   | **Complete winner** — hook and body both working           | Scale via [budget-shifter](budget-shifter.md); clone the hook archetype.     |
| 🟢 High   | 🔴 Low    | **Clickbait** — hook stops scroll but body disappoints     | **Keep the hook, replace the body.**                                          |
| 🔴 Low    | 🟢 High   | **Weak hook, strong content** — those who stay love it     | **Keep the body, swap the hook.** Lowest‑risk, highest‑leverage iteration.    |
| 🔴 Low    | 🔴 Low    | **Full failure** — hook and body both unengaging           | Pause; brief a fully new concept via [creative-brief](creative-brief.md).    |

4. Hook Archetype Tagging

Every scored hook is tagged with an archetype + variable set so patterns can be compared across the account.

4.1 Archetypes

| Archetype              | Example opener                                           |
| :--------------------- | :------------------------------------------------------- |
| **Direct Question**    | *"Are you still losing $40K a year to manual reporting?"* |
| **Negative Constraint**| *"Don't launch another ad until you've seen this."*       |
| **Social Proof**       | *"2,400 agencies switched to this last quarter."*         |
| **Direct Problem**     | *"Your CFO is drowning in CSV exports."*                  |
| **Visual ASMR**        | Satisfying visual (pour, reveal, cut) with no voiceover   |
| **Pattern Interrupt**  | Unexpected cut, color flash, or contradiction in frame 1  |
| **Character Intro**    | Founder / user on camera, direct address                  |
| **Curiosity Gap**      | *"Nobody told me this about Meta ads…"*                   |

4.2 Variable Extraction

For each hook, also capture:

• Text overlay in seconds 0–3 — presence, size, contrast/color
• Creator persona — founder · customer · voiceover · no‑face · animated character
• Pattern interrupt type — visual cut · zoom · color flash · text pop · contradictory statement
• Scene count in 3s — number of cuts (higher = more pattern interrupt)
• Audio — voiceover · music · ASMR · silence

These variables are what make iteration briefs specific ("reuse the handheld kitchen‑background UGC persona from AD_388"), not abstract ("use a testimonial").

5. Iteration Brief — Output Logic

Comparing archetypes by both Hook Rate and Hold Rate reveals recombination opportunities. The strategic output is an iteration brief that pairs the best stop‑power with the best retention.

Template:

"The {winning-hook-archetype} hook is outperforming {other-archetype} by X% on Hook Rate. However, {other-archetype} has a Y% higher Hold Rate."

Action: "Produce N new variations using the {winning-hook} opener, transitioning at ~3s into the {other-archetype} body script — to maximize both stop‑power and retention."

Only issue an iteration brief when:

• Both archetypes have ≥ 3 ads in the sample window (statistical floor).
• The Hook Rate gap is ≥ 15% and the Hold Rate gap is ≥ 10% (otherwise the recombination isn't worth the production cost).

6. Operational Output: Hook Audit + Iteration Brief

Hook Architect Audit — Acme Ads · 30‑day window · 42 video ads scored

Account baseline: Hook Rate 28.4% (σ = 5.1) · Hold Rate 31.2% (σ = 7.3)

🏆 Elite hooks (7) — Hook Rate > 33.5%

• AD_388 · Social Proof · Hook 41.2% · Hold 38.1% · "2,400 founders already made the switch" → complete winner
• AD_412 · Pattern Interrupt · Hook 38.9% · Hold 22.0% → 🟡 clickbait risk (hook high, hold below avg) — keep hook, swap body
• AD_554 · Visual ASMR · Hook 36.1% · Hold 44.3% → clone the archetype

🔴 Below‑avg hooks — iteration targets (6)

• AD_238 · Direct Problem · Hook 19.1% · Hold 41.8% 🟢 → weak hook, strong content — swap hook, keep body
• AD_501 · Curiosity Gap · Hook 17.3% · Hold 18.9% → full failure — pause, new concept via creative-brief

Archetype performance

| Archetype | Avg Hook Rate | Avg Hold Rate | Sample (ads) | | :------------------- | :------------ | :------------ | :----------- | | Social Proof | 34.8% 🏆 | 32.1% | 9 | | Visual ASMR | 31.2% | 38.4% 🏆 | 6 | | Direct Problem | 22.1% | 39.7% 🏆 | 8 | | Negative Constraint | 29.4% | 28.3% | 5 |

🎯 Iteration Brief

The Social Proof hook archetype is outperforming Direct Problem by 34% on Hook Rate (34.8% vs. 22.1%). However, Direct Problem has a 12% higher Hold Rate (39.7% vs. 32.1%).

Action: produce 3 new variations opening with a Social Proof hook ("2,400 agencies already switched") that transition at ~3s into a Direct Problem body script ("…because their CFOs were drowning in CSV exports") to maximize both stop‑power and retention.

Next steps: pass the pairing Hook: Social Proof / Body: Direct Problem into creative-brief for 3 visual concepts, then copy-rotation for 3 headlines + 3 body variants per concept.