Converting Headline Writer

Headline Copywriting — Hero Sections That Convert

Write headlines that name a felt pain and promise the dream outcome, then carry that promise through the subheadline into the CTA. The headline is the highest-leverage text on the page; treat it as the product of a process, not a first draft.

Read these supporting files on demand:

• ${CLAUDE_SKILL_DIR}/references/headline-formulas.md — the full formula library (Hormozi value equation, PAS, enemy positioning, Schwartz awareness stages, Copyhackers, Harry Dry, Julian Shapiro, StoryBrand) and the 20 headline rules
• ${CLAUDE_SKILL_DIR}/references/hero-and-cta.md — hero-section anatomy (job of each element) and CTA/click-trigger patterns with conversion evidence
• ${CLAUDE_SKILL_DIR}/references/swipe-file.md — 38 real SaaS headlines tagged by pattern, before→after rewrites, and stealable sentence structures

Load first — brand context

Brand	Load	Hard rules
Allodra	writing-room/docs/voice-allodra.md	Honesty guardrail (see below); use the approved vocabulary
Sierra-6	writing-room/docs/voice-sierra6.md (if built)	Never sell leads; veteran-operator tone
Other	The brand's voice doc, or ask for positioning	—

Paths are relative to the workspace root — Glob for **/voice-allodra.md if the session starts in a subfolder. The voice doc supplies the enemy, the metaphor, the approved words, and the claims that are safe to make (guardrails can change — the voice doc is always the source of truth). Never write a headline before reading it.

Standalone use (no voice docs / this skill was shared with you): if no voice doc exists, check ${CLAUDE_SKILL_DIR}/my-brand-voice.md. If that's missing too, interview the user briefly before writing anything: (1) what do you sell and to whom, (2) the enemy (the tool/practice/incumbent the reader blames — never the reader), (3) the dream outcome in the customer's words, (4) proven pains — real lines from replies, reviews, or calls, (5) approved and banned vocabulary, (6) claim guardrails (what's safe to promise publicly). Save the answers to ${CLAUDE_SKILL_DIR}/my-brand-voice.md so the user answers once; load it in place of the voice doc from then on. The "Allodra defaults" section at the bottom shows what a filled-in profile looks like.

The core formula

HEADLINE  = PAIN (villain named or implied) × DREAM OUTCOME (specific, visualizable)
SUBHEAD   = the mechanism — what it is and why the promise is believable
CTA       = hand them the outcome, first person, risk killed beside the button

Run every candidate through the Hormozi value lens: raise the dream outcome and its perceived likelihood (proof beside the claim); cut time delay ("in 7 days", "instantly") and effort/sacrifice ("without X"). A headline that only states pain agitates; a headline that only states outcome floats. The strongest pair both: "Stop paying a penalty for growth" = pain (growth tax) with the outcome (penalty-free scaling) implied in the same breath.

The villain is always a tool, category, incumbent, or practice — never the reader's own failure.

Workflow

Copy this checklist and tick it off:

• 1. Load the brand voice doc. Note the enemy, approved vocabulary, and any claim guardrails.
• 2. Diagnose the awareness stage of the page's traffic (table in references/headline-formulas.md). Homepage traffic is usually solution-aware → lead with pain + differentiated outcome. Cold-ad traffic skews problem-aware → lead harder on pain.
• 3. Mine the pain and dream outcome in customer language. Sources in priority order: proven hooks that already got replies (cold-email sequences, VSL hooks), actual reply/call language, the voice doc. Steal the customer's sentence; don't invent one.
• 4. Bleed the tap — write 20+ headlines minimum across at least 4 pattern categories (enemy/"Stop X", dream-outcome, specificity-number, metaphor, clarity). The first dozen are always predictable; gold shows up late. Use the stealable structures in the swipe file.
• 5. Score with the five tests (below). Shortlist the top 3.
• 6. Assemble the full hero for the winner: eyebrow → H1 → subhead → CTA + click trigger → proof line (element jobs in references/hero-and-cta.md).
• 7. Honesty check against the brand guardrail. A headline you can't ship is a failed headline, no matter how good.
• 8. Deliver in the output format below and save to writing-room/drafts/[brand]-[slug]-draft.md (standalone: wherever the user keeps drafts, or beside the page being edited).

The five tests (scoring gate)

Every shortlisted headline must pass all five:

1. Grunt test — a stranger glancing for 3 seconds knows what you sell, how it helps, and what to do next.
2. Visualizable — the reader can see it ("1,000 songs in your pocket"). Abstractions evaporate.
3. Falsifiable — a specific claim that could be proven wrong ("30 minutes or it's free") gets believed. Vague claims get ignored.
4. Un-copyable — if a competitor could run the same line word-for-word, it's a category label, not a headline.
5. One big idea — one reader, one pain, one promise. Two ideas = half the strength of each.

Mechanics: ~6–12 words, roughly 7th-grade reading level, no hype stacking, concrete numbers over adjectives (odd/precise numbers read as counted, not invented).

The CTA segue

The button is the actionable next step to the headline's promise — a continuation of the story, never a generic verb.

• Call to value: [verb] + [the thing they actually want]. "See the 3-year math" beats "Learn more".
• First person converts: "Start my free trial" beat "your" by +90% CTR in the canonical test. Default to first person.
• Kill risk at the click: microcopy beside the button — "No credit card", "Takes 30 seconds", "One-time, own it forever" (+19% signups documented for a single click trigger).
• Match CTA weight to awareness: most-aware → hard CTA ("Buy it once"); earlier stages → soft first CTA ("See how it works") with the hard CTA repeated lower.

Full patterns and evidence: ${CLAUDE_SKILL_DIR}/references/hero-and-cta.md.

Output format

Deliver exactly this shape:

## Shortlist (from N variants)
| # | Headline | Pattern | Why it wins |
|---|---|---|---|
| 1 | ... | enemy / pain-led | ... |
| 2 | ... | dream-outcome | ... |
| 3 | ... | specificity | ... |

## Recommended hero (option 1)
**Eyebrow:** ...
**H1:** ...
**Subhead:** ...
**CTA:** [button] + [click-trigger microcopy]
**Proof line:** ...

## Why this beats the current hero
(1–3 sentences against the incumbent headline)

## Full variant list
(all 20+, grouped by pattern — the user picks differently than you'd predict)

Allodra defaults (worked example)

• Current hero (the incumbent to beat): "Stop renting your CRM. Own it outright." — leads with the mechanism (CRM ownership). Weakness: "CRM" makes the reader think software category, not their own pain.
• Target energy: "Stop paying a penalty for growth." — leads with the felt pain (you're taxed for succeeding) and implies the dream outcome (scale free and clear).
• The enemy: the digital landlord — rent, seat creep, the growth tax. Attack the incumbent practice (renting), never the reader.
• Approved vocabulary: landlord · rent · growth tax · seat creep · take title · own it · allodial · digital real estate · free and clear · one-time · stop renting.
• ⚠️ Honesty guardrail (hard rule): safe to claim "buy it once, own the license, no monthly rent, no growth tax." Do NOT claim "deployed to your own cloud / nobody can ever lock you out" in public copy until reseller rights are live. The honesty is the edge — never trade it for a bigger number.
• Proven pains that already got replies (mine these first): the growth tax / taxed for succeeding · renting vs owning · the 3-year number ($18k GHL / $43k HubSpot) · "you're a digital tenant" · "why I fired GoHighLevel".

Read $ARGUMENTS for the brand, page, and task. If no brand is named, ask which brand — the voice doc changes everything.