sarojkjha Email Lifecycle Sequence
Build the automated emails that carry a user from signup to habit to expansion — and remove the ones quietly costing you. Lifecycle email is the highest-leverage owned channel most startups half-build: a welcome sequence written once in year one, an onboarding flow nobody has opened since, and no win-back at all.
Two failures dominate. The first is gaps — no win-back means lapsed users are simply lost, and no expansion sequence means NRR depends entirely on sales. The second is stacking — three sequences that each look reasonable alone firing at the same new user, producing eleven emails in two weeks and a spike in unsubscribes.
Both are visible in about a minute. Run the audit before writing anything.
When to use
The user is building, fixing, or auditing automated email — sequences, flows, drips, or triggered campaigns. For diagnosing whether churn is the problem, use lifecycle-and-retention first; this skill builds the fix once the diagnosis points at communication.
Before you start
- Read the brand/product context (
.agents/product-marketing.md) for voice, audience, and the offer sequences should route toward. - Find the activation moment. The specific action that predicts a user sticks — the second project created, the first teammate invited, the first report run. Onboarding email that doesn't drive toward a named activation moment is just a product tour.
- List what already fires, including anything transactional. Users don't distinguish between your marketing automation and your product notifications; the inbox is one place.
- Get volume and value. Monthly signups and value per conversion decide which sequences are worth building at all.
Method
Six sequences cover the lifecycle. Most teams have two.
| Sequence | Trigger | Job |
|---|---|---|
| Welcome | Signup | Confirm the decision, set expectations, drive the first action |
| Onboarding / activation | Signup, then behaviour | Reach first value, then habit |
| Engagement / nurture | Ongoing | Stay useful between purchases |
| Expansion | Usage threshold | Move to higher value when usage justifies it |
| Win-back | Inactivity | Recover lapsed users before they're gone |
| Sunset | Long inactivity | Exit the disengaged cleanly to protect deliverability |
Build in the order the engine returns — onboarding and welcome first, because they act on the highest-intent audience you will ever have.
Behaviour beats time. A sequence that sends email three on day four regardless of what the user did is a newsletter with extra steps. Trigger on actions taken and not taken, and exit the moment the goal is met — nothing erodes trust faster than being nagged to do something you already did.
See resources/lifecycle-playbook.md for each sequence's structure, timing, exit conditions, subject-line patterns, and a worked example.
Run the engine
Paths assume you installed with
npx skills add. From a clone of this repo, useskills/email-lifecycle-sequence/resources/…instead.
node .agents/skills/email-lifecycle-sequence/resources/lifecycle-audit.js # demo
node .agents/skills/email-lifecycle-sequence/resources/lifecycle-audit.js --have welcome,onboarding --signups 500 --value 400
node .agents/skills/email-lifecycle-sequence/resources/lifecycle-audit.js --have welcome --build-hours 8 --rate 150 --payback 3
node .agents/skills/email-lifecycle-sequence/resources/lifecycle-audit.js --json --have welcome,winback
node .agents/skills/email-lifecycle-sequence/resources/lifecycle-audit.js --help
It returns coverage across the six sequences, the gaps in build order, a send-load check counting every sequence that overlaps a new user's first 14 and 30 days, and a breakeven per sequence.
The breakeven is deliberately framed as a question you can answer: rather than estimating lift from benchmarks that vary too widely to mean anything, it computes how many extra conversions a month a sequence needs to pay for itself. Usually that number is a fraction of a percent of signups — which settles the argument quickly.
Workflow
- Audit first. Run the engine with what already exists. The stacking problem in particular is invisible until you count across sequences.
- Confirm the activation moment before writing onboarding. Everything in that sequence points at it.
- Pick the highest-priority gap from the engine's build order.
- Draft the sequence using the structure in the playbook — trigger, email count, timing, one job per email, and the exit condition.
- Write each email to one job with one CTA. An email doing two things does neither, and lifecycle email is where that discipline slips most.
- Set exit conditions explicitly. Every sequence needs a rule for stopping when its goal is met, and a global rule so a user can't be in three sequences at once.
- Re-run the audit with the new sequence included to confirm you haven't created a fatigue problem while fixing a coverage one.
- Name what to measure — the downstream behaviour, not the open rate. Opens have been unreliable since Apple's Mail Privacy Protection began pre-fetching images.
Present the result
Lead with the coverage map and send load — that's the diagnosis. Then the drafted sequence, formatted so it can be pasted into an ESP: trigger, timing, subject, preheader, body, CTA, exit condition per email.
State the assumptions the breakeven rests on, and flag any claim in the copy that needs proof before it ships.
Guardrails & common mistakes
- Count sends across sequences, not within them. Every sequence can look reasonable and the combined load still be punishing. This is the most common lifecycle failure and it's invisible from inside any single flow.
- Always set an exit condition. Continuing to ask a user to do something they've already done is the fastest way to lose them.
- Behaviour-trigger where possible. Time-based sends ignore what the user actually did.
- Don't measure on opens. Mail Privacy Protection inflates them. Measure clicks and, better, the downstream action.
- Transactional and marketing email share an inbox. Count both when assessing load.
- Deliverability is a real constraint. Sending to disengaged addresses degrades inbox placement for the engaged ones — which is the entire point of the sunset sequence.
- Honour consent and give a working unsubscribe. Beyond being legally required under CAN-SPAM, GDPR and similar regimes, a hard-to-find unsubscribe converts a quiet exit into a spam complaint, which costs far more.
- Don't write six sequences at once. Ship one, measure, then build the next. A half-working set of six is worse than one that works.
- A sequence can't fix a product that doesn't deliver value. If onboarding email is carrying activation on its own, the problem is upstream.
Related AAJ resources
- Interactive tool: https://aajconsult.com/tools/churn-nrr-calculator
- Template: https://aajconsult.com/resources/retention-nrr-workbook
Related skills
lifecycle-and-retention (diagnose whether churn is the problem before building sequences) · onboarding-activation (the in-product half of activation) · copywriting (the emails themselves) · marketing-psychology (the friction blocking the action a sequence is asking for) · cold-email-sequence (outbound, a different job entirely).
Credits
Original AAJ skill. The Agent Skills format and the marketing-skills catalog by Corey Haines (coreyhaines31/marketingskills, MIT) were references for structure and coverage; this skill is independently written. See the repository README for the full reference list.