email-marketing

Email Marketing Automation

Design, build, and optimize email marketing flows for SaaS GTM. This skill covers drip sequence architecture, deliverability engineering, template design patterns, and provider API integration.

Email Marketing Architecture

Email System
  ├── Transactional Emails (account confirmations, password resets, receipts)
  │   └── Triggered by user actions, must send immediately
  ├── Marketing Emails (newsletters, announcements, product updates)
  │   └── Sent to segments on a schedule
  ├── Drip Sequences (automated multi-email flows)
  │   └── Triggered by events, sent on a delay schedule
  └── Lifecycle Emails (onboarding, activation, retention, win-back)
      └── Triggered by user state changes or inactivity

Core Sequences Every SaaS Needs

1. Welcome Sequence (Trigger: user signs up)

Day 0 (immediate): Welcome + quick-start guide
Day 1: Key feature highlight + "try this" CTA
Day 3: Social proof (testimonials, case studies)
Day 5: "Need help?" with support links + community invite
Day 7: Activation check — if not activated, branch to re-engagement

2. Activation Sequence (Trigger: signup but no key action)

Day 2: "You're 1 step away" — focus on the activation action
Day 4: Video walkthrough of the core feature
Day 7: Success story of someone who activated
Day 10: Direct offer — "Reply to this email and I'll help you set up"
Day 14: Final nudge — "Is [Product] right for you?"

3. Trial-to-Paid Sequence (Trigger: trial started)

Day 0: Trial confirmation + what to expect
Day 3: Feature highlight #1 (most valuable)
Day 5: Feature highlight #2 + comparison to alternatives
Day 7 (mid-trial): Progress check + premium feature preview
Day 10: Customer success story + ROI data
Day 12 (2 days before): Urgency — "Your trial ends in 2 days"
Day 13 (1 day before): Last chance — clear pricing, FAQ
Day 14 (expiry): "Your trial has ended" — extend offer or discount
Day 16 (grace): "We saved your data" — final conversion attempt

4. Retention/Engagement Sequence (Trigger: no login for 7+ days)

Day 7 inactive: "We miss you" + new feature announcement
Day 14 inactive: "Here's what you're missing" + activity digest
Day 21 inactive: Re-engagement offer (discount, extended trial)
Day 30 inactive: "Is everything okay?" — personal outreach feel
Day 45 inactive: Win-back offer (significant discount)
Day 60 inactive: Final email — "Should we close your account?"

5. Upsell/Expansion Sequence (Trigger: usage threshold reached)

Day 0: "You're hitting limits" — data on their usage
Day 2: Feature comparison (current plan vs next tier)
Day 5: ROI calculator or case study of upgrade
Day 7: Limited-time upgrade offer
Day 10: Follow-up if no action taken

Key Metrics

Metric	Healthy Range	Action if Below
Open Rate	25-45%	Fix subject lines, sender name, send time
Click Rate	3-8%	Fix CTA, content relevance, design
Reply Rate	1-3%	More personal tone, ask questions
Unsubscribe Rate	<0.5% per email	Reduce frequency, improve segmentation
Bounce Rate	<2%	Clean list, verify emails at signup
Spam Complaint Rate	<0.1%	Critical — fix immediately or domain gets blacklisted

Segmentation Strategy

Behavioral Segments

Segment	Criteria	Use For
Power Users	5+ logins/week, multiple features used	Upsell, referral asks, testimonial requests
Engaged	2-4 logins/week	Feature education, community building
At Risk	1 login/week, declining	Re-engagement, value reminders
Dormant	0 logins for 14+ days	Win-back sequences
Churned	Cancelled or expired	Win-back with offers

Lifecycle Segments

Segment	Stage	Primary Goal
New Signups	Onboarding	Activation (complete key action)
Activated	Using product	Engagement + habit formation
Trial Users	Pre-payment	Conversion to paid
Paying Customers	Retention	Expansion + advocacy
Churned	Lost	Win-back

Enrich Segments with the Persona Substrate + SDV Forecasts

Behavioral and lifecycle segments above tell you what a user is doing; the shared persona substrate (.gtm/personas/, schemas/persona.schema.json) and SDV forecasts (.gtm/sdv/forecasts/*.forecast.yml) tell you what they object to and what price they tolerate — build the persona once, consume it here. Cross the two axes:

• Persona objections → which email handles which objection. Each persona carries pain_points, and each SDV forecast carries structured top_objections[] (segment + objection + severity + blocker_tier B0/B1/B2 + revenue_at_risk). Map the highest-severity objection for the persona(s) in a segment to a specific email: the email that names and dissolves that exact objection. A B0 demand-blocker for a paying persona earns its own dedicated email; a B2 polish objection is a PS line. This is the routing rule — one SDV-discovered objection, one email that answers it — not generic benefit copy.
• Price tolerance → offer framing per persona. A persona's price_sensitivity (low/medium/high) and any SDV willingness_to_pay band (Van Westendorp / Gabor-Granger range) set how an upsell/trial-to-paid email frames price: lead with value and a guarantee for high-sensitivity personas (and the lower offer tier), lead with the premium outcome for low-sensitivity personas. Never send the same discount framing to a low and a high price_sensitivity segment.
• Revenue weighting. Prioritize sequence build order by the persona's revenue_pct and the objection's revenue_at_risk — handle the objection costing the most revenue first.

If .gtm/personas/ is empty, recommend /gtm-personas derive; if no SDV forecast exists yet, recommend /gtm-validate to discover objections before writing the objection-handling emails (the forecast's top_objections[] is the input, not a guess).

A/B Testing Framework

What to Test (Priority Order)

1. Subject line — 40% impact on opens
2. Send time — 20% impact on opens
3. CTA button text and placement — 30% impact on clicks
4. Email length — varies by audience
5. Personalization level — first name, company, usage data
6. From name — "Juan from Dojo" vs "Dojo Coding" vs "Dojo Team"

Statistical Requirements

• Minimum 1,000 recipients per variant
• Run for at least 24 hours (captures all time zones)
• 95% confidence level before declaring winner
• Test one variable at a time