Lead Enrichment

Lead Enrichment

Enrich company/person rows with contact context. Prefer APIs when configured; otherwise infer patterns from public pages and keep confidence honest.

When to use

• Contact enrich cards / post-prospecting enrichment
• Preparing CRM export with email patterns

When not to use

• Inventing verified inboxes
• Login-walled LinkedIn scrapes

Credentials (never print secrets)

Check keys without leaking values:

python navin/skills/lead-enrichment/scripts/enrich_leads.py --keys-check /dev/null 2>/dev/null || \
  python -c "import os; print('HUNTER', 'set' if os.getenv('HUNTER_API_KEY') else 'missing'); print('APOLLO', 'set' if os.getenv('APOLLO_API_KEY') else 'missing')"

Preferred path (script)

After prospecting wrote sales/prospects-*.csv:

python navin/skills/lead-enrichment/scripts/enrich_leads.py sales/prospects-YYYYMMDD.csv \
  -o sales/prospects-YYYYMMDD-enriched.csv --verify-existing
python navin/skills/lead-qualification/scripts/score_leads.py sales/prospects-YYYYMMDD-enriched.csv

Columns written: email, email_status (verified|unverified|catch_all|invalid), email_score, enrichment_source.

Manual Hunter / Apollo (fallback)

curl -s "https://api.hunter.io/v2/email-finder?domain=example.com&first_name=Ada&last_name=Lovelace&api_key=$HUNTER_API_KEY"
curl -s "https://api.hunter.io/v2/email-verifier?email=ada@example.com&api_key=$HUNTER_API_KEY"
curl -s -X POST "https://api.apollo.io/api/v1/mixed_people/search" \
  -H "Content-Type: application/json" \
  -H "X-Api-Key: $APOLLO_API_KEY" \
  -d '{"q_organization_domains":"example.com","page":1,"per_page":5}'

Only label verified when the API status/score says so.

Public research stack (no LinkedIn login)

Use together with prospecting:

1. web_search + web_fetch for directories, registries, team pages.
2. scrape for same-domain careers/about corpora.
3. Exa / Firecrawl MCP (Settings → MCP) for deeper public crawl/search.
4. HubSpot MCP for CRM push after enrichment.

Fallback (no keys)

1. Fetch company contact/legal/press pages.
2. Infer pattern (first.last@, flast@) with confidence medium/low.
3. Leave contact_hint + email_status=unverified; never invent a full mailbox as verified.

Workflow

1. Detect keys (--keys-check).
2. Enrich CSV with enrich_leads.py.
3. Validate + score with score_leads.py.
4. Apply data-quality-agent before CRM push (crm-update-agent / HubSpot MCP).

Rules

• Never print API keys.
• Public-source or API evidence only.
• Opt-out / suppression lists honored when the user provides them.
• Prefer HubSpot MCP / crm-update-agent for CRM writes after enrichment.

Anti-patterns

• "john.doe@company.com (verified)" with no API/public proof
• Enriching personal emails scraped from private profiles
• Skipping --keys-check then claiming verified contacts