Brand Scout Agent Skill

Purpose: Autonomous overnight research to identify new shipping leads from target industries.

Core Reference

→ See rules.md → Agent-Specific Rules → Brand Scout Agent

Execution Workflow

Phase 1: Research Planning (10 PM - 11 PM)

1. Load target industry lists from .claude/brand_scout/targets/
2. Identify companies not already in HubSpot (search first)
3. Queue research jobs for overnight processing

Phase 2: Data Collection (11 PM - 5 AM)

1. Extract company info: industry, size, location, website
2. Identify key decision makers (VP Supply Chain, Director Logistics, COO)
3. Estimate shipping volume from:
   • Company size + industry averages
   • Public shipping data (if available)
   • Job postings mentioning fulfillment/logistics
4. Flag pain points: shipping costs, slow delivery, multi-carrier complexity

Phase 3: Report Generation (5 AM - 6 AM)

1. Generate markdown summaries in .claude/brand_scout/output/YYYY-MM-DD/
2. Calculate confidence scores (High/Medium/Low)
3. Prioritize leads by estimated annual shipping spend
4. STOP → Human reviews during 9AM sync

Output Template

# Brand Scout Report: {Company Name}
**Date**: {YYYY-MM-DD}
**Confidence**: {High/Medium/Low}
**Estimated Annual Shipping Spend**: ${amount}
**Industry**: {industry}
**Employee Count**: {count}

## Company Profile
- **Website**: {URL}
- **Headquarters**: {City, State}
- **Business Model**: {B2B/B2C/Hybrid}
- **eCommerce Platforms**: {Shopify/WooCommerce/Custom/etc.}

## Key Contacts
| Name | Title | LinkedIn | Email (if found) |
|------|-------|----------|------------------|
| [Name] | [Title] | [URL] | [Email] |

## Shipping Profile
**Estimated Volume**: {parcels/month}
**Service Needs**: {Ground/Expedited/Priority mix}
**Current Carriers**: {inferred from data}
**Pain Points Identified**:
- [Pain point 1]
- [Pain point 2]

## Geographic Distribution
**Top Destination States**: {states based on customer base}
**Warehouse/Fulfillment Locations**: {if found}

## Recommendation
**Action**: {Pursue/Hold/Archive}
**Reasoning**: {why this lead is/isn't a good fit}
**Next Steps**: {if pursuing, what's the approach?}

## Sources
- [Source 1 URL]
- [Source 2 URL]

Strict Compliance Rules

✅ MUST DO

• Generate research summaries with confidence scores
• Include source URLs and data freshness timestamps
• Flag high-priority leads (>$500K annual shipping spend)
• Preserve raw research data in .claude/brand_scout/data/
• Log all research queries and results

❌ NEVER DO

• Auto-create HubSpot records without human approval
• Run brand scout during business hours (overnight only)
• Overwrite existing lead folders
• Skip confidence scoring or source attribution
• Research companies already in HubSpot pipeline

Quality Gates

Before marking research complete:

• All target companies researched (or flagged as "insufficient data")
• Confidence scores calculated for each lead
• Output files generated in .claude/brand_scout/output/YYYY-MM-DD/
• High-priority leads (>$500K) flagged in summary report
• Source URLs documented for verification

Human Approval Workflow

During 9AM Sync:

1. Prioritization Agent loads brand scout output
2. Human reviews each lead's recommendation
3. Approved leads → Create HubSpot contact + company
4. Held leads → Move to .claude/brand_scout/hold/ for later review
5. Archived leads → Move to .claude/brand_scout/archive/ (not a fit)

HubSpot Creation (only after approval):

from hubspot_sync_core import HubSpotSyncManager

# After human says "approve Brand X"
sync_manager.create_contact(
    first_name="John",
    last_name="Smith",
    email="john@brandx.com",
    company="Brand X",
    lifecycle_stage="lead"
)

# Move approved lead to [00-LEAD]_Brand_X/ folder

Research Source Priorities

Primary Sources (most reliable):

1. Company website (about, team, contact pages)
2. LinkedIn company page (employee count, locations)
3. Public shipping/logistics data (if available)
4. Job postings (mentions of "fulfillment", "shipping volume")

Secondary Sources (use with caution):

1. Industry reports (estimates, not specific to company)
2. News articles (recent shipping partnerships)
3. Glassdoor reviews (employee mentions of shipping operations)

Avoid:

• Speculative data without sources
• Outdated information (>1 year old)
• Competitor websites (not reliable for volume estimates)

Error Handling

Insufficient Data:

• Flag lead as "Low Confidence - Insufficient Data"
• Document what data is missing
• Suggest manual research approach for human

Rate Limiting:

• Respect research source rate limits (LinkedIn, Google, etc.)
• Implement exponential backoff for blocked requests
• Log all rate limit encounters for future optimization

Duplicate Detection:

• Search HubSpot before researching (avoid duplicate work)
• Check .claude/brand_scout/archive/ for previously rejected leads
• Flag potential duplicates for human review

Performance Metrics

Target Benchmarks:

• Research 10-20 companies per night

• 70% leads rated Medium or High confidence

• <10% duplicate/already-researched companies
• 100% source attribution

Monthly Review:

• Conversion rate: Brand Scout leads → Closed-Won deals
• Data quality: Accuracy of shipping spend estimates
• Time savings: Hours saved vs. manual research

Related Documentation

• rules.md - Complete agent rules and prohibitions
• .claude/docs/systems/BRAND_SCOUT_SYSTEM.md - System architecture
• .claude/brand_scout/README.md - Research templates and targets