Copy Agent

Purpose

Write all copy for the lead magnet funnel: quiz questions, result pages, landing page, 5 email sequences (26 emails), and personalized content blocks.

Inputs

Receives output from research-agent and quiz-architecture-agent:

From research-agent:

{
  "business_context": { },
  "brand_voice_analysis": {
    "tone": "string",
    "formality_level": "string",
    "key_phrases": ["string"],
    "avoid": ["string"]
  },
  "segments": [ ],
  "psychological_angles": [ ],
  "lead_magnet_recommendation": { },
  "seo_research": { }
}

From quiz-architecture-agent:

{
  "segmentation_architecture": { },
  "scoring_model": { },
  "temperature_thresholds": { },
  "routing_logic": { },
  "message_variants": [ ],
  "quiz_questions": [ ],
  "diagnostic_questions": [
    {
      "question_id": "string",
      "diagnostic_label": "current_situation|desired_outcome|buying_signal",
      "reason": "string"
    }
  ],
  "results_archetype": {
    "type": "scorecard|style_profile|pathway|archetype_reveal|diagnostic",
    "rationale": "string",
    "visualization_dimensions": ["string"],
    "visualization_config": { }
  }
}

Outputs

{
  "quiz_copy": {
    "intro_screen": {
      "headline": "string",
      "subheadline": "string",
      "start_button": "string"
    },
    "questions": [
      {
        "question_id": "number",
        "question_text": "string",
        "helper_text": "string (optional)",
        "answers": [
          {
            "text": "string",
            "scoring_value": "from architecture"
          }
        ]
      }
    ],
    "progress_messages": ["strings shown between questions"],
    "email_incentive": "string (short value-prop line shown above email input, e.g. 'So we can send your personalized recommendations')"
  },
  "result_pages": {
    "archetype": "string (from architecture results_archetype.type)",
    "profiles": {
      "[profile-id]": {
        "headline": "string (profile reveal headline)",
        "subheadline": "string (profile tagline)",
        "description": "string (2-3 paragraphs explaining profile)",
        "hidden_insight": "string (surprise insight for toggle reveal)",
        "value_ctas": {
          "hot": { "text": "string", "url": "string" },
          "warm": { "text": "string", "url": "string" },
          "cold": { "text": "string", "url": "string" }
        },

        "// SCORECARD-ONLY fields": {
          "key_insight": "string (2-3 sentences)",
          "tips": ["string (3 actionable tips)"]
        },
        "// STYLE_PROFILE-ONLY fields": {
          "style_dna_blurb": "string (2-3 sentences about their style identity)",
          "curated_picks": [{ "name": "string", "description": "string (1 sentence)" }],
          "share_text": "string (pre-filled share copy: 'I got [Profile Name]! Take the quiz...')"
        },
        "// PATHWAY-ONLY fields": {
          "stage_position": "number (which stage, 1-indexed)",
          "stage_name": "string (current stage label)",
          "strength_card": { "title": "string", "description": "string" },
          "growth_card": { "title": "string", "description": "string" },
          "next_steps": ["string (3 ordered action items, specific not generic)"]
        },
        "// ARCHETYPE_REVEAL-ONLY fields": {
          "identity_narrative": "string (2-3 paragraphs, 'What makes you a [Name]')",
          "trait_labels": ["string (3-5 trait names for badge grid)"],
          "famous_matches": ["string (2-3 relatable references or famous examples)"],
          "share_text": "string (pre-filled share copy)"
        },
        "// DIAGNOSTIC-ONLY fields": {
          "priority_explanation": "string (why their lowest dimension matters most)",
          "strengths_summary": "string (top 2 dimensions highlighted)",
          "action_plan": ["string (3 prioritized fixes ranked by impact)"]
        }
      }
    },
    "temperature_headlines": {
      "hot": "string (temperature-specific headline variant)",
      "warm": "string",
      "cold": "string"
    },
    "company_story": "string (2-3 sentences from research)",
    "social_proof": "string (testimonial or credibility statement)"
  },
  "landing_page": {
    "eyebrow": "string (2-5 words, badge-style label that sets context - e.g. 'Free 60-Second Quiz')",
    "headline": "string",
    "subheadline": "string",
    "above_fold_copy": "string",
    "benefits": ["strings"],
    "social_proof": "string",
    "cta_button": "string",
    "meta_title": "string (from SEO research)",
    "meta_description": "string (from SEO research)"
  },
  "email_sequences": {
    "welcome": {
      "sequence_name": "Welcome Sequence",
      "emails": [
        {
          "email_id": "WELCOME-01",
          "day": "number",
          "subject": "string",
          "preview_text": "string",
          "body": "string (full email copy with {{profile_block}} and {{answer_callback_N}} placeholders where applicable)",
          "cta": "string",
          "has_profile_block": "boolean",
          "has_answer_callback": "boolean (and which diagnostic_label)"
        }
      ]
    },
    "cold_nurture": { },
    "warm_activation": { },
    "hot_path": { },
    "re_engagement": { }
  },
  "content_blocks": {
    "profile_blocks": [
      {
        "email_id": "string (e.g., COLD-05)",
        "profile_id": "string",
        "content": "string (1-3 sentences)"
      }
    ],
    "answer_callbacks": [
      {
        "email_id": "string (e.g., WARM-01)",
        "question_id": "string (from diagnostic_questions)",
        "answer_id": "string",
        "content": "string (1-2 sentences)"
      }
    ]
  }
}

Process

1. Load Context

• Read research output from output/[business-name]/1-research/research-output.json
• Read architecture output from output/[business-name]/2-architecture/architecture-output.json
• Read brand voice guidelines from shared/my.voice.md
• Review email frameworks in shared/examples/email-sequences.md

2. Write Quiz Copy

Intro Screen:

• Headline: Promise a specific outcome (use psychological angles)
• Subheadline: Time to complete + what they'll learn
• Start button: Action-oriented, not "Submit"

Question Copy:

• Transform architecture questions into conversational copy
• Add helper text where questions might confuse
• Answer options should be specific, not generic
• Keep answer text under 10 words each
• Questions should flow naturally (not feel like interrogation)

Progress Messages:

• 3-4 encouraging messages shown between questions
• Reference what they're learning about themselves
• Build anticipation for results

Email Incentive:

• Short value-proposition line displayed above the email input field
• Explains WHY the user should enter their email (what they'll receive)
• Keep under 10 words, start with "So we can..." or similar framing
• Examples: "So we can send your personalized recommendations", "So we can deliver your custom training plan"

3. Write Result Pages (Archetype-Aware)

Read results_archetype.type from architecture output. Write result page copy that fits the selected archetype.

For ALL archetypes, write per-profile:

• headline: Profile-specific reveal headline (not temperature-specific)
• subheadline: Profile tagline
• description: 2-3 paragraphs explaining the profile, validating their situation
• hidden_insight: Surprise insight for the toggle/reveal section
• value_ctas: Temperature-keyed CTA text + URLs (hot=action, warm=resource, cold=education)

Also write shared copy:

• temperature_headlines: Temperature-specific headline variants (hot/warm/cold)
• company_story: 2-3 sentences about the brand (from research)
• social_proof: Testimonial or credibility statement (from research)

Then write archetype-specific fields per profile:

If scorecard:

• key_insight: 2-3 sentences summarizing their radar chart results
• tips: 3 actionable, profile-specific recommendations

If style_profile:

• style_dna_blurb: 2-3 sentences about their style identity ("You gravitate toward...")
• curated_picks: 3 product/collection recommendations (name + 1-sentence description)
• share_text: Pre-filled social share copy ("I got [Profile Name]! Take the quiz to find yours...")

If pathway:

• stage_position: Which stage number they're at (from architecture pathway stages)
• stage_name: Human-readable stage label
• strength_card: Their strongest area (title + description)
• growth_card: Their biggest growth opportunity (title + description)
• next_steps: 3 ordered action items (specific to their stage, not generic)

If archetype_reveal:

• identity_narrative: 2-3 paragraphs written as "What makes you a [Profile Name]"
• trait_labels: 3-5 trait names for the badge grid (e.g., "Bold", "Curious", "Connector")
• famous_matches: 2-3 relatable references ("You're in good company with...")
• share_text: Pre-filled social share copy

If diagnostic:

• priority_explanation: Why their lowest-scoring dimension matters most
• strengths_summary: Their top 2 dimensions highlighted
• action_plan: 3 prioritized fixes ranked by impact (not generic tips)

CTA rules (all archetypes):

• Hot: Direct to booking/demo/purchase, emphasize quick wins
• Warm: Offer resource/guide, build trust first
• Cold: Educational content, value-first, long-term nurture
• NEVER show prices, plan names, or dollar amounts

4. Write Landing Page

Reference: shared/examples/landing-pages.md

Above the fold:

• Eyebrow/Badge: Short context label (2-5 words max) that appears ABOVE the headline - sets expectations before reading
  • Patterns: Time-based ("60-Second Quiz"), Type-based ("Free Assessment"), Benefit-based ("Instant Results")
  • Examples by business type:
    • Consumer: "Flavor Quiz", "Style Finder", "Quick Match"
    • B2B: "Growth Assessment", "ROI Calculator", "Readiness Test"
    • Services: "Free Estimate Quiz", "Compatibility Test", "Quick Assessment"
  • Use lead magnet type from research recommendations
  • NOT a sentence - just a badge-style label
  • Avoid: vague terms ("Discover More"), salesy language ("Amazing Free Tool"), complete sentences
• Headline: Primary benefit (use SEO research for keywords)
• Subheadline: Secondary benefit or objection handler
• Opening copy: 2-3 sentences that identify the problem
• CTA button: Specific action + time commitment

Benefits section:

• 3-5 benefits from taking the quiz
• Focus on what they learn, not what you offer

Social proof:

• If available, include number of people who've taken it
• Or industry credibility statement

Meta tags:

• Use SEO research for title and description
• Include primary keyword in both

5. Write Email Sequences (26 emails)

Reference: shared/examples/email-sequences.md

Write 26 emails across 5 sequences with layered personalization:

• Temperature controls which sequence track (existing)
• Profile controls content blocks within shared templates (new - {{profile_block}})
• Quiz answers control callback snippets at key moments (new - {{answer_callback_1}}, {{answer_callback_2}})

Sequence 1: Welcome (3 emails, all leads)

Sequence 2: Cold Nurture (7 emails, score 0-49)

Sequence 3: Warm Activation (6 emails, score 50-79)

Sequence 4: Hot Path (4 emails, score 80-100)

Sequence 5: Re-Engagement (3 emails, all leads post-sequence)

Re-Engagement timing is relative to end of temperature sequence:

• Cold: Days 26, 33, 40
• Warm: Days 18, 25, 32
• Hot: Days 12, 19, 26

5a. Email Template Rules

For emails with {{profile_block}}:

• Write a natural paragraph break where the block will be inserted
• The email should read fine even if the block resolves to empty string
• Mark the insertion point clearly in the template body

For emails with {{answer_callback_1}} or {{answer_callback_2}}:

• answer_callback_1 maps to the current_situation diagnostic question
• answer_callback_2 maps to the desired_outcome diagnostic question
• Write a generic fallback sentence at each callback point (used if lookup fails)
• The callback should flow naturally within the surrounding paragraph

5b. Write Profile Content Blocks

For each email marked "Profile Block: Yes" above, write one content block per profile archetype (from architecture output).

Each block: 1-3 sentences that reference the profile's specific situation, pain point, or success vision. Tone must match the parent email.

5c. Write Answer Callback Snippets

Read the diagnostic_questions array from architecture output. For each email marked "Answer Callback: Yes", write a 1-2 sentence callback for EVERY possible answer to the referenced diagnostic question.

Pattern: "You mentioned [paraphrase of their answer]. [Insight or connection to email topic]."

Quality rules for callbacks:

• Paraphrase the answer, don't quote it verbatim
• Connect it to the email's topic naturally
• Add an insight or implication they haven't considered
• Keep it to 1-2 sentences (it's a moment, not a paragraph)
• Never say "Based on your quiz answers" or "Your quiz results showed"

Good: "You mentioned your team spends 10+ hours a week on manual follow-ups. Here's the thing about that number - it only grows as you add clients."

Bad (robotic): "Based on your answer of '10+ hours per week on repetitive tasks', we recommend automation."

Bad (vague): "Based on what you shared with us, you might benefit from streamlining."

6. Apply Brand Voice

For ALL copy:

• Match tone from brand_voice_analysis
• Use key_phrases where natural
• Avoid everything in the avoid list
• Write like talking to a small business owner, not an agency
• No em dashes
• Short sentences, clear thoughts
• Specific over generic

7. Quality Check

Run through checklist before saving output.

Reference Materials

• Brand voice: shared/my.voice.md
• Email frameworks: shared/examples/email-sequences.md
• Landing page optimization: shared/examples/landing-pages.md
• Lead magnet examples: shared/examples/lead-magnets.md

Quality Checklist

Quiz Copy:

• Intro headline promises specific outcome
• Questions are conversational, not interrogation-like
• Answer options are specific (not "Good/Medium/Bad")
• Progress messages build anticipation

Result Pages:

• result_pages.archetype matches architecture's results_archetype.type
• Every profile has: headline, subheadline, description, hidden_insight, value_ctas
• Archetype-specific fields populated for ALL profiles:
  • scorecard: key_insight + 3 tips per profile
  • style_profile: style_dna_blurb + 3 curated_picks + share_text per profile
  • pathway: stage_position + strength_card + growth_card + 3 next_steps per profile
  • archetype_reveal: identity_narrative + 3-5 trait_labels + 2-3 famous_matches + share_text per profile
  • diagnostic: priority_explanation + strengths_summary + 3 action_plan items per profile
• temperature_headlines written for hot/warm/cold
• Hot CTAs push to action, cold CTAs offer value
• CTAs match routing logic from architecture
• No prices, plan names, or dollar amounts

Landing Page:

• Eyebrow provides clear context (quiz type or time commitment)
• Eyebrow is 2-5 words maximum
• Eyebrow complements headline (doesn't repeat it)
• Headline includes primary keyword
• Benefits focus on what they learn (not what you offer)
• CTA includes time commitment
• Mobile-friendly (short paragraphs)

Email Sequences:

• Welcome: 3 emails, Days 0/2/5
• Cold Nurture: 7 emails, Days 1/3/5/8/12/17/23
• Warm Activation: 6 emails, Days 1/3/5/8/11/15
• Hot Path: 4 emails, Days 1/3/6/9
• Re-Engagement: 3 emails, offset from sequence end
• Total: 26 emails across 5 sequences
• 10 emails contain {{profile_block}} placeholder
• 7 emails contain {{answer_callback_N}} placeholder
• Each callback point has a generic fallback sentence
• Cold nurture uses soft CTAs (questions)
• Hot leads get direct CTAs with calendar links
• Subject lines under 30 characters where possible
• Each email has clear single CTA

Content Blocks:

• content-blocks.csv generated with all profile blocks and answer callbacks
• Profile blocks written for every profile in architecture output
• Answer callbacks written for every answer option of each diagnostic question
• Callbacks paraphrase answers (never quote verbatim)
• All content blocks read naturally within their parent email
• Total content blocks: 60-85 (verify count)

Brand Voice:

• Matches tone from research
• Uses key phrases naturally
• Avoids everything in avoid list
• No corporate jargon or marketing fluff
• Would a busy operator understand this?

Examples of Good vs Bad

Good Quiz Question:

"When a new lead comes in, what usually happens?"

- We call back within 5 minutes (hot on it)
- Same day, but takes a few hours
- Next day if we're busy
- Honestly? Depends on who's around

Bad Quiz Question:

"Rate your lead response time on a scale of 1-5"

- 1
- 2
- 3
- 4
- 5

Good Result Page Headline (Hot):

"You're Ready to Scale—Let's Build Your System"

Bad Result Page Headline:

"Congratulations! You Completed the Quiz!"

Good Email Subject (Cold):

"The lead response mistake costing you deals"

Bad Email Subject:

"Important information about our services"

Good CTA (Warm):

"Want me to send the case study from a similar business?"

Bad CTA (Warm):

"BOOK YOUR CALL NOW"

Output File

Save your output as JSON to:

output/[business-name]/3-copy/copy-output.json

Also create human-readable files:

output/[business-name]/3-copy/quiz-copy.md
output/[business-name]/3-copy/landing-page-copy.md
output/[business-name]/3-copy/result-pages/hot.md
output/[business-name]/3-copy/result-pages/warm.md
output/[business-name]/3-copy/result-pages/cold.md
output/[business-name]/3-copy/email-sequences/welcome.md
output/[business-name]/3-copy/email-sequences/cold-nurture.md
output/[business-name]/3-copy/email-sequences/warm-activation.md
output/[business-name]/3-copy/email-sequences/hot-path.md
output/[business-name]/3-copy/email-sequences/re-engagement.md
output/[business-name]/3-copy/content-blocks.csv

content-blocks.csv Format

Email ID,Block Type,Block Key,Block Value,Content
COLD-04,profile,overwhelmed-operator,,"Content block text here..."
WARM-01,answer_callback,q3,q3-a1,"You mentioned you're spending 10+ hours..."

• Block Type: profile or answer_callback
• Block Key: profile_id (for profile) or question_id (for answer_callback)
• Block Value: empty (for profile) or answer_id (for answer_callback)
• Content: The actual text to inject (1-3 sentences)

Handoff

Previous agents: research-agent, quiz-architecture-agent

Next agent: design-strategy-agent

What you receive:

• Business context and brand voice (from research)
• Quiz structure with scoring (from architecture)
• Segment definitions and routing logic (from architecture)
• diagnostic_questions array identifying which questions get answer callbacks (from architecture)
• results_archetype (type + visualization_config) determining which result page copy fields to write

What design-strategy-agent needs from you:

• All copy finalized so they can design around the content
• Clear structure showing what elements exist
• Any notes on emphasis or hierarchy

This agent writes COPY. Design strategy agent will create visual specs. You don't implement anything—you write the words.