邮件序列设计

HarborMug post-purchase sequence

Use three consent-based emails to help first-time customers care for their mug and find support. Send on delivery day, day three, and day seven. The primary outcome is a verified care-guide visit within 14 days of delivery; the secondary outcome is a support or feedback submission. Do not add discounts, product recommendations, or a repeat-purchase request.

The merchant data, destinations, address, and performance targets are synthetic planning examples; no messages have been sent and no business results are claimed.

The report is in English as requested; customer-facing copy uses the US English specified by the merchant. The complete email drafts below retain the supplied brief's factual content. Machine-readable versions are in sequence.json.

Sequence overview

Step	Purpose	Cumulative delay from delivery	Primary destination	Skip rule
1	Introduce safe care	Day 0, local 10:00	Care guide	Skip if guide already visited
2	Reinforce the care routine	Day 3, local 10:00	Care guide	Skip if guide already visited
3	Invite feedback and practical help	Day 7, local 10:00	Support form	Guide completion does not skip this message

Sender and reply-to: Mia at HarborMug, support@harbormug.example. Route incoming support to Mia; escalate unresolved cases to Alex through the merchant's existing support process. Personalization: trim the first name and use there if blank, producing “Hi there,”. Escape names for the selected email format. The unsubscribe token must be generated by the CRM; never send an unresolved merge field. The unsubscribe link is a required utility link, separate from each message's single primary CTA. Keep the support address in the copy as contact information, with no additional support button in the care emails.

Complete customer-facing drafts

Email 1: Care introduction

Send: delivery + 0 cumulative calendar days, at local 10:00, subject to the rules below.

Subject: Your HarborMug: a simple care routine

Preview: A gentle hand wash is the first step. Keep this guide nearby for everyday care of your new 500ml mug.

Complete body:

Hi {{first_name}},

Your HarborMug has arrived. Before its first use, wash it gently by hand. This 500ml stainless steel mug is hand-wash only and must not go in the microwave.

We put the essentials in one short guide so you can keep them nearby. If the mug arrived damaged, contact Mia at support@harbormug.example instead of using it.

Read more in your care guide below.

Mia, HarborMug

Read more in the care guide → https://harbormug.example/care?utm_source=lifecycle&utm_medium=email&utm_campaign=post_delivery&utm_content=e1

HarborMug | 100 Harbor Way, Boston MA 02110 Unsubscribe: https://harbormug.example/unsubscribe?token={{unsubscribe_token}}

Primary CTA: Read more in the care guide → https://harbormug.example/care?utm_source=lifecycle&utm_medium=email&utm_campaign=post_delivery&utm_content=e1

Subject alternatives (choose one per test recipient):

• Your new mug: start with gentle hand care
• A quick care guide for your HarborMug
• Keep these mug care essentials nearby

Email 2: Everyday care reminder

Send: delivery + 3 cumulative calendar days, at local 10:00, subject to the rules below.

Subject: How does your new mug fit your morning?

Preview: Make the care routine part of your day, and keep the hand-wash reminder close whenever you put your mug away.

Complete body:

Hi {{first_name}},

A small routine can make your mug easier to care for. After use, wash it by hand and let it dry before putting it away. Remember to keep it out of the microwave.

Our care guide collects these reminders in one place. You can return to it whenever you need a refresher, without searching through your order messages.

Visit the everyday care guide below.

Mia, HarborMug

Visit the care guide → https://harbormug.example/care?utm_source=lifecycle&utm_medium=email&utm_campaign=post_delivery&utm_content=e2

HarborMug | 100 Harbor Way, Boston MA 02110 Unsubscribe: https://harbormug.example/unsubscribe?token={{unsubscribe_token}}

Primary CTA: Visit the care guide → https://harbormug.example/care?utm_source=lifecycle&utm_medium=email&utm_campaign=post_delivery&utm_content=e2

Subject alternatives (choose one per test recipient):

• A small care habit for your new mug
• Your everyday HarborMug care reminder
• Keep your mug routine simple this week

Email 3: Feedback and support

Send: delivery + 7 cumulative calendar days, at local 10:00, subject to the rules below.

Subject: Is there anything we can help with today?

Preview: Tell Mia how your mug is working for you. A practical question or a problem is welcome, with no review required.

Complete body:

Hi {{first_name}},

You have had a little time with your HarborMug. How is it fitting into your day? If something is unclear, Mia can help you work through it.

We welcome questions about care as well as problems with your order. You do not need to write a public review to get help, and this message does not ask you to purchase anything else.

Visit the support form below.

Mia, HarborMug

Visit support → https://harbormug.example/support?utm_source=lifecycle&utm_medium=email&utm_campaign=post_delivery&utm_content=e3

HarborMug | 100 Harbor Way, Boston MA 02110 Unsubscribe: https://harbormug.example/unsubscribe?token={{unsubscribe_token}}

Primary CTA: Visit support → https://harbormug.example/support?utm_source=lifecycle&utm_medium=email&utm_campaign=post_delivery&utm_content=e3

Subject alternatives (choose one per test recipient):

• How is your HarborMug working for you?
• Mia is here for your mug questions
• Need a hand with your new HarborMug?

CRM implementation specification

Enrollment and required fields

Enroll once when a verified delivery event arrives for a paid, noncancelled first purchase containing HM-BLUE-001 and delivered in the United States. Require affirmative current marketing consent, a valid deliverable email, known delivery time, known US country, known customer time zone, known first-purchase status, known care-guide status, and an explicitly clear support/return status. Unknown values fail closed; route the record to data review without sending. A delivery event alone is not marketing consent. Customers with an earlier completed purchase, an already active or completed enrollment, or an ineligible order cannot enter. A first order with several qualifying units still generates one sequence.

Required implementation fields: stable customer and order references, qualifying SKU, order payment/cancellation status, first-purchase flag derived from order history, delivery timestamp with time zone, country, customer time zone, email, first name, affirmative consent and its latest status, unsubscribe/complaint/hard-bounce status, order return/refund status, customer support status, guide-visit status, and a cross-campaign marketing send ledger. These are proposed CRM fields, not connected integrations. Do not infer consent, support status, or country from missing data.

Proposed launch policy: enroll only deliveries occurring after activation; do not backfill historical orders or recreate missed steps. If an enrolled customer's required data becomes unknown, suppress sending until corrected within that step's send window. Data repair does not override an exit event.

Timing and expiry

Use the recipient's verified IANA time zone. Calculate delivery's local calendar date plus 0, 3, and 7 days, at 10:00. Delays are cumulative, not gaps of three and then seven days. Never send before confirmed delivery: if day-zero delivery is confirmed after 10:00, send email 1 at the next local 10:00. If confirmed at 10:00 before dispatch, it can send that day. Subsequent steps stay anchored to delivery, not the prior send. Do not apply a weekend exclusion.

Recommended staleness rule: email 1 expires at the scheduled day-three slot; email 2 expires at the scheduled day-seven slot; email 3 expires at delivery's local day-14 10:00. At a boundary, expire the older step before evaluating the new one. Suppressed or expired steps are never bundled or backfilled. End the flow when every step is sent, skipped, expired, or cancelled. Continue outcome observation through delivery + 14 elapsed days even if the last message sent earlier.

For the supplied sample delivered October 20, 2026 at 10:00 America/New_York, the nominal schedule is October 20, October 23, and October 27, each at 10:00. This is a schedule illustration, not an execution record.

Purchase, support, and consent exclusions

Evaluate the latest state on enrollment and immediately before every dispatch, including retries. A return request, cancellation, full or partial refund on the qualifying order, an open support case anywhere on the customer record, withdrawal of consent, unsubscribe, complaint, or hard bounce cancels all remaining messages. The return request is sufficient; do not wait for refund approval. Unknown support status blocks sends. Closed historical support cases alone do not exclude an otherwise eligible customer.

Treat support and return exits as terminal for this enrollment: closing a case does not automatically resume reminders. A new subsequent purchase also cancels remaining steps as a proposed purchase exclusion; never restart this first-purchase flow for repeat orders. If a later delivery update reports a delivery reversal or loss, cancel the remaining sequence and leave resolution to support. Service and order messages remain governed by their own transactional processes.

The supplied A100 case must receive no sequence or catch-up email: the November 3 return request takes precedence, the customer never enrolled, and there is no marketing consent. Only the allowed support reply may proceed outside this flow. Do not promise a refund or return-shipping terms; refunds require merchant approval and shipping responsibility is unknown. B200 cannot enroll because consent, delivery, country, and support status are missing. Do not fill those fields with favorable defaults.

Care-guide completion and branching

For this program, “completion” means an identified, verified visit to /care, not proof of reading. Record the visit against the customer, qualifying product, and applicable guide version. A prior known visit for that guide version also counts. An email click alone, an open, a bot request, or a security scanner must not mark completion. Link attribution requires an identified first-party session or equivalent reliable CRM association; do not add email addresses or other personal data to URLs.

On guide_viewed, mark both unsent care steps as skipped and cancel their queued retries. If a guide visit occurs between email 1 and email 2, skip email 2. If it occurs before email 1, skip both. Leave email 3 scheduled unless a higher-priority exclusion applies. Visits after the 14-day outcome window can still suppress an unsent care reminder but do not count toward the 14-day goal. Never resend a sent message after a completion event is reversed or duplicated.

A support submission should create or associate a support case and thereby exit remaining marketing steps. Count confirmed form submissions or received support replies as secondary outcomes; a support-link click is not a submission.

Frequency conflicts and deduplication

Pause newsletters and promotions during the active sequence, including conflict delays; do not release an accumulated backlog when the flow ends. Transactional order and shipping messages continue in real time and do not consume this marketing limit. This entire sequence, including its support invitation, consumes the marketing limit.

Across all marketing flows, enforce at most one email per customer per local calendar day and a minimum of 24 elapsed hours since the last marketing send. At a conflict, postpone the lower-priority message by at least 24 hours, then use the next allowed local 10:00 slot and recheck every condition. Prefer this care sequence over newsletters/promotions; give safety/service-critical communication priority. For other marketing ties, use earliest scheduled time and then a stable campaign priority. All campaigns must use the same atomic reservation ledger; already sent marketing wins the slot. Repeated conflicts can expire a step under the windows above.

Use one enrollment key per stable customer + first qualifying order + program identity. Keep a customer-level enrollment lock so duplicate order references or identity merges cannot create another first-purchase flow. Use one send key per enrollment + step; editing copy or testing a subject must not create a new send key. Deduplicate delivery events by event reference, with order/product matching as a fallback. Replayed events must not reset the delivery anchor or skipped states.

Immediately before provider handoff, atomically check suppressions and completion state, reserve the customer marketing slot, and mark the step dispatching. Reuse the same provider idempotency key on retries. If the provider outcome is uncertain, reconcile its receipt before retrying; never blindly resend. Persist sent, skipped, expired, cancelled, dispatching, and failed states with reasons. Apply exits and guide visits before evaluating queued sends; a late event cannot recall an already accepted email. Do not let a transport retry bypass the frequency limit or a new suppression.

Acceptance cases

Input or event	Expected behavior
All 100 sample customers remain eligible; no conflicts or visits	Three scheduled steps per customer; 300 potential sends, before delivery failures
Guide viewed before enrollment	Enroll if otherwise eligible; skip steps 1 and 2, keep step 3
Guide viewed after step 1	Cancel queued step 2, retain step 3
Guide click only, no verified page visit	Keep care step eligible; exclude bot clicks from outcomes
Open support or return request before a delayed dispatch	Cancel every unsent step immediately
New subsequent purchase after step 1	Cancel remaining steps; do not reenroll
Marketing already sent at the due slot	Delay at least 24 hours, select allowed 10:00, recheck; expire if stale
A delayed step reaches the next step's boundary	Expire older step; evaluate the new step with frequency and exclusions
Duplicate delivery, concurrent job, or provider timeout	One enrollment and at most one provider-accepted send per step
Missing first name	Render “Hi there,”
Missing required eligibility fields	No enrollment or send until verified; no historical backfill
A100 / B200	No marketing enrollment or catch-up send

Measurement and decision plan

Use delivered messages as the denominator for each email's unique human click rate and unique open rate. Opens are directional because privacy features and automated activity can inflate them. Track verified guide visits separately from link clicks. The primary conversion denominator is unique eligible entrants, including those who later skip care emails or exit; count unique identified visitors within 14 days of delivery. Report visitors already complete at enrollment separately from customers who first visit after entry so prior completion is not presented as email-generated progress. These metrics do not establish causal lift.

Email	Open target	Click target	Illustration if 100 delivered
1: Care introduction	40%	8%	40 opens, 8 unique clickers
2: Care reminder	38%	6%	38 opens, 6 unique clickers
3: Support check-in	35%	4%	35 opens, 4 unique clickers

The sequence planning target is 15 unique care-guide visitors per 100 eligible entrants within 14 days (15%). The per-email illustrations independently assume 100 deliveries; actual later sends shrink with skips and exits. Do not sum 8 + 6 + 4 as guide conversions: people overlap and email 3 points to support. No historical email performance is provided. Weekly orders and revenue in the merchant materials do not measure this flow and cannot substantiate a sales lift.

Measure support submissions and replies separately, with no numeric target supplied. Also report eligible entrants, exclusions by reason, care skips, conflict delays, expiry, sent/delivered totals, bounces, and duplicates prevented. Attribute clicks by the supplied utm_content=e1, e2, and e3; deduplicate guide visits at customer level. Consent and complaint events must feed the suppression checks, not just the reporting dashboard.

Stop new sends pending review when complaint rate reaches or exceeds 0.1%; review the sequence when unsubscribe rate reaches or exceeds 1%. Calculate each against delivered messages, by message and across the sequence, using unrounded rates. One complaint among 100 delivered is 1%, already above the stop threshold; one unsubscribe among 100 is 1%, reaching review. With zero delivered, rates are undefined. Individual unsubscribes, complaints, and hard bounces always suppress that customer immediately regardless of aggregate rates. Keep aggregate monitoring across a rolling 14-day send window; this window is a proposed implementation choice. Mia owns review, with unresolved issues escalated to Alex.

For subject testing, retain the supplied primary subjects initially. With only 100 example entrants, do not split into four small arms and call a winner. If volume supports testing, compare the primary with one alternative at a time, assign once at customer level, keep timing/body/CTA fixed, and predefine sample size and a measurement window using actual delivery and event rates. Optimize care emails for verified guide visits and the final email for confirmed support engagement, with complaint and unsubscribe guardrails.

The executed Skill analyzer scored the final sequence 100/100 (grade A), above its 70-point minimum. It found no pacing issues; cumulative gaps are three and four days. This is a copy heuristic, not a deliverability or business-performance prediction. It checks CTA presence rather than enforcing a single CTA, so the one-primary-CTA structure was also inspected separately.

Handoff and evidence

sequence.json contains the complete email blocks, alternatives, URLs, and cumulative delays. planning-calculations.json holds target arithmetic and denominator notes. sequence-analysis.json holds the executed Skill analyzer output. verification.json holds checks against the assembled deliverables and deterministic CRM rule scenarios. These are completed drafts and implementation specifications; CRM integrations and the fictional destination pages have not been deployed or tested live.

The evidence basis is inputs/brief.json (supplied drafts), inputs/merchant.md (product, audience, operations and edge cases), and inputs/benchmarks.json (internal planning targets). The Skill and its available playbook were read and the provided analyzer was run. Its linked copy-guidelines reference was unavailable; the available playbook supplies copy guidance. No external benchmarks are represented as evidence. Replace fictional destinations and the synthetic postal footer with approved merchant values before any real deployment, and verify event ingestion and suppression ordering in the CRM before activation.