Petal & Pine: A Win-Back Campaign That Is Burning the List
Petal & Pine ships a $45 monthly plant-subscription box. Its list has 210,000 addresses and 38% have not opened in 90 days. Inbox placement fell from 96% to 89% after a win-back campaign mailed the full list three times in two weeks; spam complaints reached 0.14% and one-click unsubscribe 0.9%. The welcome series converts 6.1% of new subscribers to a paid plan, and about 3,200 people enter it monthly. The cart-abandonment flow converts 1.2% across 8,000 monthly entries. Contribution margin is 45% and average subscription life is five months. The lifecycle manager proposes a fourth win-back send to the lapsed segment plus a redesign of the cart email. Engineering can ship only one flow change this quarter.
1. Rank the three options - fourth win-back send, cart redesign, welcome-series rebuild - by expected incremental profit, and justify overriding the manager's proposal.
Read the worked answer
Size each option in monthly contribution, not open rates.
Fourth win-back send. 79,800 lapsed addresses x 89% placement = 71,022 delivered. At an optimistic 1.5% reactivation that is ~1,065 subscribers x $45 x 5 months x 45% margin = $107,000 of lifetime contribution, front-loaded at roughly $21,500 in month one. But it is mailed to exactly the segment generating the complaints, and a further placement drop hits the 130,000 engaged subscribers who produce nearly all revenue. The option's headline number is real and its hidden cost is larger.
Cart redesign. 8,000 entries x 1.2% = 96 recovered orders/month ≈ $4,320 revenue, ~$1,944 contribution. A 33% relative lift adds about $640 of monthly contribution.
Welcome rebuild. 3,200 entrants x 6.1% = 195 paid starts/month x $45 x 5 months x 45% = $19,700 of monthly contribution. A one-point lift (6.1% to 7.1%) adds 32 starts, worth roughly $3,240 of contribution per month - five times the cart redesign and with no reputation risk.
Recommendation: rebuild the welcome series and ship suppression this quarter, not a fourth win-back send. The win-back campaign is treating a deliverability problem as an engagement problem; sending a fourth time to the same non-responsive segment will push complaints further past the 0.1% comfort threshold and suppress the mail that actually converts. Welcome mail goes to people who just raised their hand, so it lifts revenue and reputation together. Revisit the lapsed segment only after suppression and authentication are fixed.
2. The lifecycle manager wants to A/B test the cart email. Is the test worth running, and how would you design the welcome-series test instead?
Read the worked answer
Run the sample-size math before arguing about creative. For a two-proportion test at 95% confidence and 80% power, n per arm ≈ 16 x p(1-p) / d².
Cart flow, baseline 1.2%, detecting a 0.5-point absolute lift: 16 x 0.012 x 0.988 / 0.000025 = 7,588 per arm, so 15,176 total. At 8,000 entries a month split 50/50 that is about two months just to read the result - for an effect worth ~$640 a month. Detecting a 0.4-point lift needs ~11,856 per arm, or three months. The test is technically possible but economically absurd.
The welcome test is cheaper because the comparison is different. Do not A/B two variants of a 6.1% flow; run a true holdout: 10% of new subscribers receive no welcome emails at all. Comparing 6.1% against a genuine 0% baseline needs only a few hundred per arm to reach significance, so you get a clean incremental read within weeks - and it answers the question that matters, which is how much revenue the flow creates, not which subject line wins.
Design. Randomize at subscriber level, hold the paid-plan offer constant, measure paid conversion within 14 days as primary and 90-day retention as secondary, and pre-register the analysis. Keep the holdout to 10% and time-boxed so you do not strand revenue. If the incremental lift is confirmed, then optimize variants with sequential rollout rather than a powered A/B, and monitor complaint rate weekly as a guardrail.
3. What is the deliverability remediation plan, and where does the 38% no-open stat mislead you?
Read the worked answer
The open rate is not a reliable engagement signal. Apple Mail Privacy Protection pre-fetches images, so opens are inflated for Apple Mail users and suppressed for anyone with images blocked; a 90-day no-open segment therefore contains real non-openers, real readers whose opens were never recorded, and people who read on a client that hides activity. Suppressing on opens alone will delete good subscribers and keep bad ones.
Rebuild the lapsed definition on harder signals: no click, no site visit, no purchase in 180 days. Cross-reference the web analytics user ID and purchase history to confirm who is genuinely dormant.
Remediation sequence.
- Authenticate properly - SPF, DKIM and DMARC - and confirm one-click unsubscribe is live, since bulk senders are required to honor it.
- Stop mailing the confirmed-dormant segment in bulk. Move them to a low-volume re-permission stream: one plain-text email asking whether they still want the box, with no images and a single link.
- Sunset anyone who does not respond within 30 days; remove them from all sends. A smaller, engaged list out-earns a large, complained-on list.
- Cut cadence to the engaged base and monitor Google Postmaster Tools daily. Complaints at 0.14% are elevated; the practical ceiling is 0.1%, and bulk-sender enforcement triggers nearer 0.3%.
- Re-warm gradually and watch placement before resuming normal volume.
Reframe the goal. The win-back campaign is being measured on reactivations when it should be measured on whether the engaged 130,000 can still reach the inbox. Deliverability is the asset; reactivations are the output.