Hook Angle Library

Seven archetypes, ranked roughly by ceiling in the calibration dataset. Every post opens with exactly one. When generating the three-angle set in Step 3, pick the three that best fit the user's proof profile — an angle that demands a metric the user doesn't have is disqualified.

Exemplars below are original demonstrations of the pattern, not real posts.

1. Tool-powered result

A named AI tool + an outsized outcome. Highest ceiling when the tool is having a moment. Requires: a real workflow and a genuine result.

"Claude + one cron job now writes my entire content pipeline. (steal the full setup, completely free)"

2. Metric shock

One hard, precise number doing all the work. Requires: a real metric.

"We booked 27 sales calls in 48 hours for one client... All from 1 LinkedIn post (grab the playbook free below)"

3. Before/after contrast

Two timestamped states, the gap is the hook. Requires: an honest before.

"7 months of daily posting: 800 followers, 1-2 calls a month. 30 days of THIS system: 21 booked calls."

4. Curation flex

You did the tedious collecting so they don't have to. Requires: the collection genuinely exists. Numbers make it ("166 posts across 8 niches").

"I saved every viral giveaway post I found for 6 months. 166 of them. One database. It's yours..."

5. Contrarian pattern-interrupt

Open by attacking the default behaviour of the ICP. Works with zero metrics — the strongest angle for users without proof points yet.

"STOP automating individual tasks in 2026. It's why your 'AI stack' saved you 20 minutes and changed nothing."

6. Experiment reveal

You ran a controlled test; the result is counterintuitive. Requires: the test actually happened.

"I posted the exact same lead magnet twice. 1st time: 40 comments. 2nd time: 1,158. The only thing I changed..."

7. Borrowed authority

Apply a famous operator's framework to your channel and report what happened. Requires: you genuinely applied it. The famous name carries reach; your result carries trust.

"In Feb I applied Hormozi's lead magnet framework to LinkedIn. In 3 weeks, here's what changed:"

Selecting the three for Step 3

• User has strong metrics → lead with 1 or 2, add 3, add 5 for contrast.
• User has no metrics yet → 5, 4 (if they own a collection), 7.
• Magnet IS a collection/swipe file → 4 is near-mandatory as one of the three.
• A named AI tool is central to the magnet → 1 is near-mandatory.

Never present two angles from the same archetype.

Narrative frames (for multi-variant runs)

Hook archetypes decide the opening claim; narrative frames decide who is telling the story and how. Crossing the two is what keeps five variants genuinely different rather than one post re-hooked five times.

• The Mirror — second person; the reader lives the scenario line by line and lands on "Sound familiar?" Strongest when the pain is a shared, repeated moment the ICP recognises instantly.
• The Confession — first person; admit the embarrassing cost you paid before the fix (the hours, the unbilled work, the night-before panic). Vulnerability reads as proof.
• The Vivid Contrast — one sharp metaphor carries the whole gap ("senior-partner thinking, ransom-note handover"). Pick one image and commit; two metaphors cancel each other.
• The Intervention — contrarian frame aimed at the reader's default behaviour. Pairs naturally with archetype 5; the frame supplies the "here's the loop I see everywhere" body.
• The Fork — two artefacts or outcomes side by side, present tense, no invented event. The safest frame when the user has no story or metrics — the contrast is the story.

In a multi-variant set, no two variants share a frame, and each gets its name and a one-line mechanism when presented.