claude-email-marketing-skill

name	claude-email-marketing-skill
description	Give Claude deep expertise in email marketing strategy, deliverability, automation, copywriting, and ESP selection
triggers	["help me with email marketing","write an email campaign","fix my email deliverability","set up email authentication","create a welcome email sequence","improve my email open rates","choose an email service provider","segment my email list"]

Claude Email Marketing Skill

Skill by ara.so — Marketing Skills collection.

This skill provides Claude with opinionated, practitioner-level expertise in email marketing. It covers authentication (SPF, DKIM, DMARC), list building, segmentation, copywriting, automation sequences, deliverability optimization, metrics analysis, and ESP selection.

What This Skill Does

Transforms Claude into an email marketing expert who can:

• Diagnose deliverability issues and provide specific fixes for spam folder problems
• Write high-converting email copy with effective subject lines and CTAs
• Design automation sequences (welcome series, abandoned cart, re-engagement)
• Configure email authentication (SPF, DKIM, DMARC records)
• Segment audiences and recommend personalization strategies
• Evaluate ESPs using a five-factor framework
• Interpret email metrics and suggest optimization tactics
• Ensure compliance with CAN-SPAM, GDPR, CASL, CCPA

Installation

For Claude Code

1. Clone the repository:

git clone https://github.com/jacquescorbytuech/email-marketing-skill.git

2. Add the skill path to your .claude/settings.json:

{
  "skills": [
    "/absolute/path/to/email-marketing-skill/email-marketing"
  ]
}

3. Restart Claude Code or reload the project.

For Claude.ai Projects

1. Navigate to the repository and open email-marketing/SKILL.md
2. Copy the entire contents
3. Create a new project at claude.ai
4. Paste the skill content into the project's custom instructions

For API Integrations

Include the contents of email-marketing/SKILL.md in your system prompt:

import anthropic
import os

with open('email-marketing/SKILL.md', 'r') as f:
    email_skill = f.read()

client = anthropic.Anthropic(api_key=os.environ.get("ANTHROPIC_API_KEY"))

message = client.messages.create(
    model="claude-3-5-sonnet-20241022",
    max_tokens=4096,
    system=email_skill,
    messages=[
        {"role": "user", "content": "Help me fix my email deliverability issues"}
    ]
)

Skill Structure

The skill is organized into focused modules:

• authentication.md — SPF, DKIM, DMARC setup and troubleshooting
• list-building.md — Lead magnets, opt-in strategies, subscriber tracking
• segmentation-and-data.md — Dynamic/static segments, data hygiene, GDPR compliance
• copywriting.md — Subject lines, email structure, CTAs, spam trigger words
• automation-and-sequences.md — Trigger-based workflows, sequence design
• deliverability.md — Inbox placement, IP warming, spam folder recovery
• esp-and-reference.md — Provider comparison, legislation reference, glossary

Common Usage Patterns

Deliverability Troubleshooting

"My emails are going to spam. I'm sending from sendgrid.net, 
my open rate dropped from 22% to 3% last week, and Gmail is 
the worst performer. What's wrong?"

Claude will diagnose:

• Shared IP reputation issues
• Authentication gaps (likely missing DMARC)
• Engagement-based filtering triggers
• Provide step-by-step remediation

Writing Email Sequences

"Create a 5-email welcome sequence for new SaaS trial users. 
Product is a project management tool, trial is 14 days, 
main friction point is CSV import complexity."

Claude generates:

• Day 0: Welcome + quick win (create first project)
• Day 2: CSV import guide (addresses friction)
• Day 5: Feature showcase (collaboration tools)
• Day 10: Social proof (case study)
• Day 13: Conversion push (upgrade CTA)

Authentication Setup

"Walk me through setting up SPF, DKIM, and DMARC for my domain 
example.com. I'm using Postmark for transactional email and 
Mailchimp for marketing."

Claude provides:

• Exact DNS records for both services
• Conflict avoidance strategies
• Testing and validation steps
• DMARC policy recommendations

ESP Selection

"I need an ESP for a 50k subscriber ecommerce list. 
Heavy personalization, abandoned cart is critical, 
budget is $500/month. Compare options."

Claude evaluates:

• Klaviyo (ecommerce-native, strong on abandoned cart)
• Drip (automation-focused, good personalization)
• ActiveCampaign (budget-friendly, solid features)
• Provides pricing, feature comparison, recommendation

Segmentation Strategy

"I have purchase history, email engagement data, and 
product category preferences. How should I segment?"

Claude recommends:

• RFM segmentation (Recency, Frequency, Monetary)
• Engagement-based tiers (active, warming, cold)
• Category affinity segments
• Automated vs. static segment uses

Configuration Examples

Setting Up DMARC Monitoring

_dmarc.example.com. IN TXT "v=DMARC1; p=none; rua=mailto:[e-mail]; ruf=mailto:[e-mail]; fo=1"

• p=none — Monitor mode (start here)
• rua — Aggregate reports
• ruf — Forensic reports
• fo=1 — Report if either SPF or DKIM fails

Klaviyo Abandoned Cart Flow (API)

// Track cart abandonment event
fetch('https://a.klaviyo.com/api/track', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    token: process.env.KLAVIYO_PUBLIC_API_KEY,
    event: 'Abandoned Cart',
    customer_properties: {
      $email: '[e-mail]'
    },
    properties: {
      $value: 129.99,
      items: [
        { product_name: 'Widget', sku: 'WID-001', price: 49.99 },
        { product_name: 'Gadget', sku: 'GAD-002', price: 80.00 }
      ]
    }
  })
});

Trigger 3-email sequence:

1. 1 hour: "Forgot something?"
2. 24 hours: 10% discount code
3. 72 hours: Last chance + free shipping

Mailchimp Merge Tags for Personalization

<!DOCTYPE html>
<html>
<body>
  <h1>Hi *|FNAME|*,</h1>
  <p>Based on your interest in *|INTEREST_CATEGORY|*, we thought you'd like:</p>
  
  *|IF:PURCHASE_COUNT > 0|*
    <p>Welcome back! Your last order was *|LAST_PRODUCT|*.</p>
  *|ELSE:|*
    <p>First-time customers get 15% off with code WELCOME15</p>
  *|END:IF|*
  
  <a href="*|PRODUCT_URL|*">Shop Now</a>
</body>
</html>

SendGrid Suppression Group Setup

curl -X POST https://api.sendgrid.com/v3/asm/groups \
  -H "Authorization: Bearer $SENDGRID_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Product Updates",
    "description": "New features and product announcements",
    "is_default": false
  }'

Users can unsubscribe from this category without leaving all emails.

Troubleshooting

"My open rates seem inflated"

Apple Mail Privacy Protection (MPP) pre-loads images:

• Opens are less reliable for engagement measurement
• Focus on clicks, conversions, and reply rates instead
• Segment out Apple Mail users for accurate baseline

"Gmail sends everything to Promotions tab"

This is normal for marketing email:

• DMARC alignment improves Primary tab odds
• Personalization and plain text help
• Don't fight it—optimize for Promotions tab success
• Use annotations (JSON-LD) for richer previews

"Spam score is high but I'm not spamming"

Common triggers:

• ALL CAPS subject lines
• Excessive exclamation marks!!!
• Spammy words: "free," "act now," "limited time"
• Large images, no text
• Broken HTML or missing plain-text version
• No physical address in footer (CAN-SPAM violation)

"Authentication records aren't validating"

# Check SPF
dig example.com TXT | grep "v=spf1"

# Check DKIM (replace 'selector' with your actual selector)
dig selector._domainkey.example.com TXT

# Check DMARC
dig _dmarc.example.com TXT

Common issues:

• SPF record exceeds 10 DNS lookups (use include sparingly)
• DKIM selector mismatch between ESP and DNS
• DMARC record has syntax errors (use validator)

Advanced Patterns

Re-engagement Workflow Logic

IF subscriber hasn't opened in 90 days:
  → Send "We miss you" email with preference center
  
  IF opens: 
    → Move to "warming" segment
  ELSE IF clicks:
    → Move to "active" segment
  ELSE (after 30 more days):
    → Send final "Stay or go?" email
    
    IF no action:
      → Suppress (don't send marketing, keep in database)

Sunset Policy Implementation

-- Example query for sunset suppression
SELECT email 
FROM subscribers 
WHERE last_open_date < DATE_SUB(NOW(), INTERVAL 180 DAY)
  AND last_click_date < DATE_SUB(NOW(), INTERVAL 180 DAY)
  AND signup_date < DATE_SUB(NOW(), INTERVAL 270 DAY);

Protects sender reputation by excluding chronically unengaged subscribers.

Best Practices from the Skill

• Always use double opt-in for cold audiences (lead magnets, ads)
• Single opt-in acceptable for warm contexts (checkout, account creation)
• Start DMARC at p=none, move to p=quarantine after monitoring
• Warm IPs gradually — double send volume every 3-5 days
• Segment on engagement first, demographics second
• Test subject lines with 10% A/B split before full send
• Include plain-text version of every HTML email
• Unsubscribe link required in every marketing email (legal + deliverability)
• Clean your list quarterly — remove hard bounces, suppress unengaged
• Track UTM parameters for every link to measure campaign ROI

Compliance Quick Reference

Regulation	Applies To	Key Requirement
CAN-SPAM	US recipients	Physical address, unsubscribe link, honor opt-outs within 10 days
GDPR	EU recipients	Explicit consent, right to erasure, data processing agreements
CASL	Canadian recipients	Express consent required, must identify sender
CCPA	California residents	Right to opt-out of data sale, privacy policy disclosure

Example Queries for Claude

"My bounce rate is 8% — is that normal?"
"Write 10 subject line variations for a webinar reminder"
"Compare Sendinblue vs Mailgun for transactional email"
"How do I set up a product recommendation sequence?"
"What's a good click-to-open rate for ecommerce?"
"Create a sunset policy for inactive subscribers"
"Why are my emails rendering badly in Outlook?"

Resources

The skill references real-world scenarios, ESP-specific quirks, and regulatory requirements. When Claude has this skill active, it will:

• Provide specific DNS records, not generic templates
• Recommend ESPs based on actual use case (not affiliate links)
• Give opinionated guidance ("do this, not that") where appropriate
• Reference actual spam filter behavior (Gmail, Outlook, Yahoo)
• Calculate realistic metrics and benchmarks by industry

---

Maintenance: This skill is based on the MIT-licensed repository. Contributions welcome at github.com/jacquescorbytuech/email-marketing-skill.