AgentSkillsCN

copy-optimizer

运用 AIDA、PAS、Fogg 等框架,优化标题、行动号召与文案内容。 触发条件:“优化标题”、“改进行动号召”、“文案创作”、“转化文案”、“价值主张”, “文案未能带来转化”、“无人点击”、“标题效果不佳”。

SKILL.md
--- frontmatter
name: copy-optimizer
description: |
  Optimize headlines, CTAs, and copy using AIDA, PAS, Fogg frameworks.
  Triggers: "improve headline", "better CTA", "copywriting", "conversion copy", "value prop",
  "copy isn't converting", "nobody clicks", "headlines don't work".

Copy Optimizer

BEFORE OPTIMIZING: UNDERSTAND CONTEXT (Critical!)

DO NOT jump straight to copy suggestions. First, understand WHY it's not converting.

Step 0: Diagnose the Problem

Ask yourself (or the user):

  1. What's the current copy? (Read the code/file)
  2. What page/component is it on? (Hero? Pricing? CTA section?)
  3. What's the visual context?
    • Is the CTA above fold? (304% better if yes)
    • Is it the only CTA? (13.5% vs 10.5% conversion)
    • Is there visual hierarchy guiding to it?
  4. What's the business context?
    • Industry? (affects tone, frameworks)
    • Target audience? (affects voice, jargon)
    • Goal? (signup, purchase, demo, etc.)
  5. What's the user awareness level?
    • Unaware → AIDA
    • Problem-aware → PAS
    • Solution-aware → JTBD
    • Most-aware → Direct Offer

If Copy Isn't Converting, Consider:

Possible CauseSymptomSolution
Wrong wordsGeneric, feature-focusedThis skill (copy-optimizer)
Wrong placementCTA below fold, lost in pageSuggest component-builder
Wrong designCTA blends in, small, hard to seeSuggest design changes
Missing trustNo social proof, guaranteesAdd trust signals
Too much frictionMultiple steps, unclear valueSimplify, clarify
Wrong frameworkAIDA for ready-to-buy audienceSwitch framework

If the problem is visual/design, suggest:

"I notice the copy might not be the only issue. The CTA is [below fold / competing with other CTAs / not visually prominent]. Would you like me to also check the visual design?"


The Stakes

  • 80% of readers never make it past the headline (Copyblogger)
  • Personalized CTAs convert 202% better than generic (HubSpot, 330,000 CTAs analyzed)
  • 95% of purchase decisions happen subconsciously (behavioral research)
  • Changing one headline increased email opens by 82% within 45 days

Every word matters. Generic copy is invisible copy.

Framework Selection (By Audience Awareness)

code
How aware is the audience?

UNAWARE (don't know they have a problem)
└── Use AIDA: Attention → Interest → Desire → Action
    Hook first, educate about problem, then present solution
    Example: "Your competitors are doing this. You're not."

PROBLEM-AWARE (know the pain, not the solution)
└── Use PAS: Problem → Agitate → Solution
    Name the pain, twist the knife, offer relief
    Example: "Tired of [pain]? It gets worse. Here's the fix."

SOLUTION-AWARE (comparing options)
└── Use JTBD: Jobs To Be Done
    Focus on the outcome they're hiring you for
    Example: "When you need X, you want Y. We deliver Y."

MOST-AWARE (ready to buy, need a push)
└── Use Direct Offer
    Skip the education, go straight to the deal
    Example: "Start free. Cancel anytime. No credit card."

Voice Calibration (By Audience)

AudienceToneSentence LengthJargon OK?Example
DevelopersDirect, technical, no fluffShort (8-12 words)Yes"Deploy in 30 seconds. No config files."
EnterpriseConfident, ROI-focusedMedium (12-18 words)Industry-specific"Reduce compliance risk by 60% while cutting audit prep time in half."
ConsumersFriendly, benefit-firstShort (6-10 words)Never"Never miss another moment."
StartupsBold, momentum-drivenPunchy (5-8 words)Moderate"Move fast. Ship faster."
HealthcareReassuring, preciseMedium (10-15 words)Clinical terms"Trusted by 2,400 hospitals. HIPAA compliant from day one."

Reading Level: Landing pages at 5th-7th grade level convert 56% higher than 8th-9th grade (Unbounce)

CTA Optimization (Research-Backed)

Friction Words (Never Use)

Bad CTAWhy It FailsConversion LossBetter Alternative
SubmitFeels like surrenderingBaseline"Get my results"
Sign upFocuses on what user gives-15% vs benefit CTAs"Start free trial"
Buy nowTriggers loss aversion-12% vs soft CTAs"Get instant access"
Click hereNo benefit, accessibility issue-8% vs specific"See how it works"
RegisterBureaucratic, cold-18% vs friendly"Join 10,000+ teams"

Sources: WiderFunnel, ConversionXL case studies

High-Converting CTA Patterns

PatternExampleWhy It Works
First-person possessive"Get my free report"90% higher CTR than "your" (Unbounce)
Benefit + timeframe"Save 5 hours this week"Specificity = credibility
Social proof embedded"Join 12,847 marketers"Bandwagon effect
Risk reversal"Try free for 14 days"Removes friction
Action + outcome"Start closing more deals"Focus on result, not action

CTA Placement Research

  • Above fold: 304% more effective than below fold
  • Centered: 682% more clicks than left-aligned
  • Single CTA: 266% higher conversion than multiple

Headline Transformation Patterns

Pattern 1: Feature → Benefit

code
BEFORE: "AI-powered analytics platform"
AFTER:  "See what's killing your conversions in 5 minutes"

WHY: Features describe product; benefits describe user outcomes
DATA: Benefit-focused headlines get 17.1% higher revenue (SaaS case study)

Pattern 2: Vague → Specific

code
BEFORE: "Grow your business faster"
AFTER:  "Add $50K MRR in 90 days"

WHY: Specificity creates credibility; vague claims are ignored
DATA: Headlines with numbers get 36% more clicks

Pattern 3: Company-focused → You-focused

code
BEFORE: "We help teams collaborate better"
AFTER:  "Ship projects 3x faster with async collaboration"

WHY: Users care about their outcomes, not your capabilities
DATA: "You" in headlines increases personalization effectiveness by 12%

Pattern 4: Passive → Active

code
BEFORE: "Meetings can be recorded automatically"
AFTER:  "Never take meeting notes again"

WHY: Active voice creates agency; passive voice creates distance
DATA: Active CTAs boost conversion by 122% (action words)

Pattern 5: Rational → Emotional

code
BEFORE: "Reduce costs by 40%"
AFTER:  "Stop losing money to inefficiency"

WHY: 95% of decisions are emotional, justified with logic
DATA: Emotional content is 31% more likely to be shared (Adweek)

Reference Loading

By Copy Type

WritingMANDATORY Read FirstDo NOT Load
Headlinespsychology/pas-framework.md, copywriting/headline-formulas.mdseo/meta-descriptions.md
CTAspsychology/fogg-model.md, copywriting/cta-patterns.mdseo/content-structure.md
Value Propspsychology/jtbd-framework.md, copywriting/value-propositions.mdpatterns/loading-states.md
Microcopycopywriting/microcopy.mdpsychology/ files
Error Messagescopywriting/error-messages.mdAll psychology files
Meta/SEOseo/meta-titles.md, seo/meta-descriptions.mdpsychology/ files

Transformation Examples

code
../../references/transformations/hero-transformation.md
../../references/transformations/cta-transformation.md
../../references/transformations/copy-transformation.md

Anti-patterns (With Data)

MistakeExampleImpactFix
Generic benefit"Grow your business"Ignored (no differentiation)"Add $50K MRR in 90 days"
Feature headline"Built with React"Users don't care"Loads in under 1 second"
No numbers"Many customers love us"Unbelievable"12,847 teams switched this month"
Weak verbs"Get started"No outcome promised"Start closing more deals"
Missing "you""Our platform enables..."Company-centric"You'll close more deals..."
Long sentences20+ words-56% conversion vs 5th gradeKeep sentences under 15 words
Passive voice CTA"Account can be created"No urgency"Create my account"

Social Proof Guidelines

Testimonials increase conversions by 34% (WikiJobs). Optimize:

ElementBest PracticeEvidence
Quantity3-5 on homepage; 100+ total in library37% higher conversions with 100+
Rating displayShow 4.2-4.5 stars (not 5.0)Perfect ratings seem fake (Northwestern Spiegel)
FormatVideo testimonials+80% conversion vs text (VWO)
FreshnessReviews < 3 months old83% find older reviews unhelpful
Negative reviewsInclude some68% trust brands more with mixed reviews

MANDATORY PIPELINE (DO NOT SKIP)

This skill is part of a multi-step pipeline. You MUST follow these steps:

PHASE 1: Pre-Optimization (MANDATORY)

Before optimizing ANY copy, you MUST load these references:

  1. Psychology frameworks (choose by awareness level):

    code
    READ: ../../references/psychology/aida-framework.md     # For unaware audiences
    READ: ../../references/psychology/pas-framework.md      # For problem-aware
    READ: ../../references/psychology/jtbd-framework.md     # For solution-aware
    READ: ../../references/psychology/fogg-model.md         # For behavior triggers
    
  2. Copywriting patterns:

    code
    READ: ../../references/copywriting/headline-formulas.md
    READ: ../../references/copywriting/cta-patterns.md
    READ: ../../references/copywriting/value-propositions.md
    READ: ../../references/copywriting/microcopy.md
    
  3. Transformation examples:

    code
    READ: ../../references/transformations/hero-transformation.md
    READ: ../../references/transformations/cta-transformation.md
    READ: ../../references/transformations/copy-transformation.md
    

PHASE 2: Analyze & Optimize

  • Identify audience awareness level
  • Select appropriate framework (AIDA/PAS/JTBD/Direct)
  • Transform copy using patterns from references
  • Apply headline transformation patterns (5 patterns above)
  • Check CTAs against friction word list
  • Verify reading level (aim for 5th-7th grade)

PHASE 3: Post-Optimization (MANDATORY)

After optimizing, you MUST:

  1. Run copy-analyzer agent for validation:

    "Let me validate the copy with the copy-analyzer agent..."

    Launch using Task tool with copy-analyzer agent:

    • Headline effectiveness (benefit-focused, specific, you-focused)
    • CTA optimization (no friction words, first-person, action+benefit)
    • Value proposition clarity
    • Psychology trigger presence
  2. Verify transformations applied:

    • Generic → Specific (with numbers)
    • Feature → Benefit (user outcome)
    • Company-focused → You-focused
    • Passive → Active voice
    • Rational → Emotional hooks
  3. Check against anti-patterns (from table above)

PHASE 4: Completion

Only complete when:

  • Appropriate psychology framework was applied
  • Copy-analyzer agent validates (≥80 score)
  • No friction words in CTAs
  • Headline is benefit-focused and specific
  • Reading level is 5th-7th grade

INTEGRATION WITH OTHER SKILLS

When called FROM component-builder:

  • You are in the middle of a pipeline
  • Return optimized copy for the component
  • Don't skip validation

When called STANDALONE:

  • Run full pipeline above
  • Suggest component-builder if implementation needed

Cross-Skill Triggers (AUTOMATIC):

If user mentions conversion problems:

"Let me first understand why your copy isn't converting..."

  • Load design-audit mindset
  • Diagnose: Is it copy? Design? Placement? Trust?
  • Then apply appropriate fix

If visual issues discovered:

"I've optimized the copy, but I notice the CTA is [issue]. Would you like me to also fix the visual design?"

  • Suggest component-builder

If full page context needed:

"To give you the best copy recommendations, I need to understand the full page context. Let me audit it first."

  • Run mini design-audit

After Optimizing: ALWAYS Offer Next Steps

code
---
**COPY OPTIMIZED** ✓

I've improved your [headlines/CTAs/copy]. Here's what changed:
- [List changes with psychology rationale]

**Would you also like me to:**
1. Check CTA visual design and placement?
2. Audit the full page for other conversion issues?
3. Review social proof and trust signals?
4. Check the page on mobile (thumb zone, touch targets)?

Related Audits to Suggest

Copy Issue FoundAlso Check
Weak headlinesVisual hierarchy (is it prominent?)
Low CTA clicksCTA placement, color, size
No conversionsFull page audit (trust, friction)
Unclear value propFirst impression (5-second test)