personalization-enricher

Personalization Enricher

Chains three research skills to build a "personalization packet" for each lead. This packet contains everything needed for Tier 3 hyperpersonalized outreach: company pain points, CRO findings, personal interests, and the original trigger signal.

Prerequisites

• agency.config.json populated
• Lead data: company name, website, contact name, linkedin URL
• Signal data: what triggered this lead (from signal-scanner or manual)

Capabilities Used

1. company-researcher -- business overview, tech stack, social presence, pain points
2. cro-auditor -- specific website issues with outreach hooks
3. person-researcher -- contact's recent activity, posts, interests

Phase 0: Intake

Gather for each lead:

1. Company name + website URL
2. Contact name + LinkedIn URL
3. Contact title + company size (for role-appropriate messaging)
4. Signal/trigger that initiated this lead

Batch mode: accept a list of leads (from CRM query) to process sequentially.

Phase 1: Company Research

Execute company-researcher for each lead:

• Visit their website via WebSearch
• Research their business, tech stack, social presence
• Identify pain points and growth signals
• Output: company_research JSON

Phase 2: CRO Audit

Execute cro-auditor for each lead:

• Audit homepage, product page, collection page
• Find 3 specific, actionable issues
• Each issue includes an outreach_hook for natural email reference
• Output: cro_audit JSON

Phase 3: Person Research

Execute person-researcher for each lead:

• Search for their recent LinkedIn posts, talks, articles
• Identify topics they care about
• Find personalization hooks (shared interests, recent achievements)
• Output: person_research JSON

Phase 4: Assemble Packet

Combine all three research outputs + the original signal into one personalization packet:

{
  "lead_id": "...",
  "company": {
    "name": "Brand X",
    "website": "brandx.com",
    "summary": "D2C skincare brand, 2 years old, growing fast on Instagram",
    "tech_stack": { "platform": "Shopify", "theme": "Dawn 2.0" },
    "pain_points": ["No customer reviews visible", "Slow mobile load time"],
    "social": { "instagram": "@brandx", "followers": "15K" }
  },
  "cro_findings": [
    {
      "issue": "No customer reviews on product pages",
      "impact": "Reviews increase conversion by 15-25%",
      "outreach_hook": "Noticed your product pages don't show customer reviews -- this alone could be leaving 15-25% of conversions on the table."
    },
    {
      "issue": "4-step checkout process",
      "impact": "Each step adds 10-15% abandonment",
      "outreach_hook": "Your checkout has 4 steps -- simplifying to 1-step could recover a significant chunk of abandoned carts."
    }
  ],
  "person": {
    "name": "Sarah Chen",
    "title": "Head of Ecommerce",
    "recent_posts": [
      { "topic": "D2C unit economics challenges", "date": "2 days ago", "hook": "Loved your take on D2C unit economics" }
    ],
    "career_notes": "Joined 6 months ago from Glossier",
    "personalization_hooks": ["Reference her post about unit economics", "Her Glossier background means she values CRO"]
  },
  "signal": {
    "type": "linkedin_post",
    "description": "Posted asking for Shopify CRO recommendations",
    "date": "3 days ago",
    "url": "https://linkedin.com/posts/..."
  },
  "recommended_approach": {
    "framework": "PAS",
    "primary_hook": "Their LinkedIn post about CRO + missing reviews on their site",
    "case_study_to_use": "Kibi Sports -- CRO audit, similar situation",
    "opening_line": "Sarah, your post about D2C unit economics resonated -- took a quick look at Brand X and found a few things that might be costing you conversions."
  }
}

Phase 5: Review

Present the packet for each lead:

• Company summary (1 line)
• Top CRO finding with outreach hook
• Person's key interest/post
• Recommended approach + opening line

User can approve, modify, or skip each lead.

Phase 6: Store

Save packets to CRM via crm-writer:

• Update lead stage from NEW to RESEARCHED
• Write personalization data to notes/description columns
• Or export as JSON for message-generator consumption

Batch Processing

For multiple leads:

1. Process company research for all leads first (most WebSearch-heavy)
2. Then CRO audits (visit each site)
3. Then person research
4. Assemble packets
5. Present batch summary

Expected throughput: 5-10 leads per session (limited by WebSearch rate)

Example Usage

Trigger phrases:

• "Research and personalize these leads"
• "Build personalization packets for today's HOT leads"
• "Enrich [company name] for outreach"
• "Deep research [contact name] at [company]"