Email finding skill

Email Finding

Overview

Finding verified work emails is the highest-stakes step in B2B prospecting. Unverified emails damage sender reputation within a week. A single batch of bounces can require months of domain warmup to recover from.

This skill walks through email discovery using a multi-provider waterfall — starting with the highest-accuracy source first, falling back only when that source returns empty. Every email found must be verified before entering any outbound sequence.

When to Use

• "Find the email for John Smith at Acme Corp"
• "Get me verified emails for this list of contacts"
• "What's the email pattern for company.com?"
• "Look up Jane's work email"
• "Enrich this CSV with email addresses"

Do NOT use for:

• Phone number finding — that's a separate waterfall
• Full company research — use lead-enrichment first
• Already have emails — skip to contact-verification

Authoritative Foundations

Email finding follows waterfall enrichment principles from DAMA-DMBOK data-quality dimensions (waterfall waterfall: Apollo → ZoomInfo → PDL → Claygent → Claude normalization) and Ziellab's 3-waterfall architecture (separate company, email, and phone waterfalls run independently).

The core insight: no single provider covers more than 60-75% of B2B contacts. Chaining 3-5 providers in sequence routinely pushes coverage to 85-92%.

Prerequisites

• First name + last name + company domain (minimum)
• LinkedIn URL significantly improves match rates (15-25 percentage points)
• If using a provider like Apollo, ZoomInfo, or Clay: valid API key or active subscription

Step-by-Step Process

Phase 1: Intake

Ask the user for:

1. How many contacts need emails? (single lookup vs batch)
2. What data do you have for each? (name, company, LinkedIn URL)
3. What providers do you have access to? (Apollo, ZoomInfo, Clay, etc.)
4. What's the budget per contact? (emerging vs enterprise providers)

Phase 2: Provider Selection

Treat these vendors as evaluation candidates, not a verified ranking. Measure cost per valid result on an authorized sample of the user's segment:

Provider	Best For	Match Rate Est.	Cost Profile
LeadMagic Email Finder	Evaluate for the target segment	Measure on an authorized sample	Check current pricing
Apollo	Evaluate for the target segment	Measure on an authorized sample	Check current pricing
ZoomInfo	Evaluate for the target segment	Measure on an authorized sample	Check current pricing
People Data Labs	Evaluate for the target segment	Measure on an authorized sample	Check current pricing
Hunter.io	Evaluate for the target segment	Measure on an authorized sample	Check current pricing
Regional email provider	Evaluate for the target segment	Measure on an authorized sample	Check current pricing
Claygent (Clay)	Evaluate for the target segment	Measure on an authorized sample	Check current pricing

Build the waterfall from cheapest + highest-coverage first, falling back to more expensive providers only when earlier steps miss.

Phase 3: Run the Waterfall

For each contact:

1. Primary provider — Run the highest-accuracy, lowest-cost provider first. For most US B2B: LeadMagic Email Finder (returns verified results) or Apollo.

2. Fallback 1 — If primary returns empty, run secondary provider. ZoomInfo for enterprise, Hunter for domain-based lookup.

3. Fallback 2 — If still empty, run tertiary. People Data Labs or regional compliance provider depending on region.

4. AI-assisted (Claygent) — For remaining 10-15%, use Claygent with an explicit prompt: "Find the work email for [name] at [company]. Return the email address AND the source URL. Do NOT guess or construct emails from patterns. If no verified source found, return empty."

5. Normalization — Consolidate results. Use COALESCE logic: prefer results from earlier waterfall steps. Prefer verified over pattern-matched.

Phase 4: Verification

Every found email must be verified before use. Run verification as a separate step — never skip it. See contact-verification skill.

Acceptable verification results: valid only. Handle risky/catch-all results separately (lower volume, extra monitoring). Discard invalid results.

Output Format

For a single lookup:

Contact: John Smith, Acme Corp
Email: [email protected]
Source: LeadMagic Email Finder
Status: Verified — valid
Confidence: High

For batch enrichment, produce a CSV with columns: email, email_source, email_confidence, verification_status

Quality Check

• Every email found has a source attribution
• Every email has been run through verification
• Bounce rate target: under 2% across the batch
• No pattern-guessed emails without source confirmation
• Catch-all domains flagged separately

Common Pitfalls

1. Skipping verification. Unverified emails cause bounces. Bounces damage sender reputation. Recovery takes weeks. Always verify after finding.

2. Single-provider dependency. One provider covers 60-75% max. Running only Apollo leaves 25-40% of contacts unreachable. Always waterfall.

3. Using pattern-guessed emails. Constructing [email protected] without confirmation creates 40-60% bounce rates. Never guess. Verify.

4. Running waterfall in wrong order. An expensive provider first burns budget on contacts a cheaper provider would have found. Sort by cost-per-hit.

5. Not normalizing company names. "Acme Inc" vs "Acme, Inc." vs "Acme Corporation" create duplicate lookups. Match on domain, not name.

6. No LinkedIn URLs in input. Adding LinkedIn URLs improves match rates by 15-25 percentage points across all providers. Run a LinkedIn URL finder column before email finding when URLs are missing.

Execution Artifacts

• references/framework-notes.md — Named frameworks and reference tables
• templates/output-template.md — Deliverable shell for agent output
• scripts/check-output.py — Lightweight deliverable validator

Related Skills

• lead-finding: Finds the contacts this skill enriches with email addresses
• lead-enrichment: Full enrichment beyond email (firmographics, technographics)
• contact-verification: Run after this skill — validates found emails before sending
• waterfall-enrichment: Complete multi-field waterfall architecture