Root cause, exit-survey reasons, customer voice, and what we can still win back. Prepared for Maya Okafor, Head of Growth.
On October 3, the billing provider was migrated. The migration itself went through — but one setting did not follow it: the card updater was not re-enabled. Subscriptions with expired cards then failed to charge, with no retry, and on October 7 and 8 the billing batch cancelled them in a single pass: 84 cancellations in two days against a baseline of about 10 a day.
This is an operations failure, not a demand failure. New subscriptions held at 6–10 a day through the whole spike, and daily revenue stayed in its normal 7.2–10.1k band. The product did not change; the billing rail did.
Read this first: nineteen days at 5–12 a day, then 45 on Oct 7 and 39 on Oct 8 — the two billing-batch days after the Oct 3 migration — then back to 5–7 within 48 hours. A step, not a trend.
| Reason | Count | Share | Note |
|---|---|---|---|
| Payment failed (expired card) | 61 | 54% | Card updater not re-enabled after the Oct 3 migration; failed with no retry |
| Too much coffee | 17 | 15% | Chronic, pre-existing — a case for pause options, not this incident |
| Price | 12 | 11% | |
| Moving or travel | 9 | 8% | |
| Other | 9 | 8% | |
| Switched to café | 6 | 5% |
Dominant reason: Payment failed (expired card) — 61 of 114 answers, and every one of them is fixable with a card update.
“My card expired and my sub just cancelled?? I didn’t even get a warning.”— customer, email reply, Oct 8
“got cancelled out of nowhere, want to keep it going”— customer, email reply, Oct 7
Both are the same story: nobody wanted to leave, nobody got a heads-up, and the second quote tells you exactly what the win-back email should promise. In the customer’s mind, the subscription was never cancelled.
All 61 payment-failure cancellations were involuntary — the customer did nothing wrong; their card simply expired and we failed to ask for an update. That makes them the warmest pool we have, and the second quote above says they want to stay. Active count on Oct 11 is 2,140; the Q4 goal is 2,600 by Dec 31, a gap of 460. Winning back all 61 is 13% of that gap, for the cost of one well-timed email.
The existing “Win-back 60d” flow is too slow for this cohort — 60-day trigger, 31% open, 1.6% click. These 61 need a dedicated, immediate update-payment send, before the 60-day flow even starts counting.
Synthetic demo data — all figures from the Fernhill demo dataset, week of Oct 5–11, 2026.