How an 11-brand subscription box operator rebuilt dunning across an entire portfolio

The operator runs 11 subscription box brands across adjacent niches — healthy meal-prep, premium pet treats, craft beer discovery, hobby boxes, teenager gift subscriptions, and several smaller lines. Aggregate portfolio MRR runs $920k, which means about $56k/mo of that revenue is at risk of being lost to involuntary churn on any given billing cycle.

Involuntary churn is the subscription-specific pain: a customer who wanted to stay, whose subscription quietly died because their card expired, their bank reissued a card, the charge failed, and the dunning workflow either didn't retry intelligently or retried so aggressively that the customer saw multiple declines and proactively cancelled out of embarrassment.

All 11 brands ran on Stripe Billing. In theory Stripe handles the card-updater feed and smart retries natively. In practice the operator had set up each brand independently over a 4-year expansion window, and the configurations had drifted. Only 4 of the 11 brands had Stripe's Account Updater active. The others had never been enrolled (it's a separate setup step for high-volume accounts). The dunning-email sequences varied from 1 email (on the oldest brand) to 5 emails (on the newest) with no consistency on cadence or copy.

Portfolio-wide involuntary churn sat at 7.2% per month. Failed-charge recovery (subscribers successfully retried after first decline) averaged 38%. On $56k/mo of at-risk revenue that meant about $35k/mo of recurring revenue was silently leaking out every cycle.

In Q3 2024 the operator hired a fractional CFO who, in her first week, asked for a portfolio-wide involuntary-churn breakdown. The operator couldn't give her one. Nobody had ever pulled the data across all 11 brands in a consolidated way — each brand's Stripe dashboard showed its own metric, but nobody had stitched them together.

The fractional CFO spent a weekend pulling the data manually from 11 Stripe exports. The aggregate number — 7.2% monthly involuntary churn, about $35k/mo in silent revenue loss — shocked the founder. She'd assumed the number was half that. She'd been optimizing the new-customer acquisition side of the funnel hard; nobody had been optimizing the recovery side because the recovery side was invisible.

The CFO's recommendation: consolidate dunning onto one platform with portfolio-wide card-updater coverage and unified smart-retry logic. She'd used multiflow on a prior engagement and introduced the founder.

The total subscriber base across 11 brands was roughly 47,000 active recurring customers. Migrating those subscriptions required coordinating the Stripe Customer / PaymentMethod transfer to the multiflow parent vault without forcing any customer to re-enter card details.

Day 1–2: Brand audit. Active subscriber count per brand, current card-updater status per brand, current dunning sequence per brand, MRR breakdown, churn rate breakdown. The CFO wanted baseline metrics before any change so post-cutover comparisons would be clean.

Day 3–4: Dunning playbook design. multiflow's implementation engineer walked through the 6-touch dunning sequence the platform recommends: Day 0 (soft decline retry with alternate bank routing), Day 3 (first dunning email with self-serve update link), Day 5 (second retry with hard decline handling), Day 7 (SMS nudge if phone on file), Day 10 (final dunning email), Day 14 (final retry + pause subscription). Each brand adapted the copy to its voice; the timing stayed consistent.

Day 5: Card-updater enrollment. All 11 brands enrolled in Visa Account Updater, Mastercard Automatic Billing Updater, and Amex Cardrefresher through the multiflow parent. This alone was going to lift recovery rates before any dunning changes took effect.

Day 6–7: Subscription migration. 47,000 active subscriptions transferred from 11 Stripe billing environments to the multiflow parent vault via Stripe's customer-transfer API. Completed in a single 40-minute batch with 93 failures that required manual cleanup (all resolved within 48 hours, mostly customers whose cards had already expired and who were in active dunning when the migration fired).

Day 8: Test retry logic. 100 simulated card-decline events across the 11 brand descriptors. Retry sequence fired correctly. Dunning emails rendered correctly per brand. Card-updater webhook events reflected correctly in the dashboard.

Day 9: Cutover. Recurring-billing logic moved to the multiflow parent. Next billing cycle hit multiflow first for retry logic, falling back to Stripe for the actual card authorization.

Day 10: First full 24-hour billing cycle ran under the new system. Smart-retry logic caught 147 charges that would have failed under the old per-brand configurations and would have gone straight to dunning.

The numbers moved within 60 days. Involuntary churn dropped from 7.2% per month to 2.8% per month at the portfolio level — a 61% reduction. The biggest driver was card-updater coverage reaching 11 of 11 brands; previously only 4 brands had it, and cards being silently updated in the background caught roughly 40% of what would have been declined charges.

Smart-retry logic was the second driver. The unified retry sequence (soft-decline retries with alternate bank routing, hard-decline respect to avoid pushing the customer into proactive-cancel territory, SMS nudges on phone-on-file) lifted failed-charge recovery from 38% to 71%. Roughly 820 subscribers per month who would have churned stayed subscribed.

Recovered monthly revenue: $41,200 at steady state. Net of the multiflow per-transaction rate delta (+$8,600/mo on the portfolio's blended rate) the operator is net-positive $32,600/mo. The fractional CFO's NPV model showed the cutover paying back inside the first billing quarter.

Dunning complaints — customers emailing support asking to reactivate their subscription after a card-decline-triggered cancel — dropped from roughly 160/mo to 35/mo. The ops team redirected the recovered hours to onboarding improvements.

The operator is clear that multiflow didn't change the subscription product itself. The 11 brands still use Stripe Billing underneath for the actual card authorization. What changed was the orchestration layer: one card-updater feed, one retry engine, one dunning sequence, one dashboard showing portfolio-wide involuntary churn.

She's honest about the cost: the per-transaction rate went up from a blended 3.1% (Stripe-native pricing on her volume band) to 5.5% on multiflow. On $920k/mo of MRR that's a material increase in processing costs. But recovered revenue of $41k/mo more than offsets, and the model says churn-reduction compounds over time — saved subscribers keep paying in subsequent months, which amplifies the LTV benefit well beyond the Month 1 recovery number.

Her quote above is the one she wanted in the hero. She added in the interview: "I should have done this three years ago. The first three years I spent all my optimization energy on acquisition because acquisition is visible. Retention failure was invisible. multiflow made it visible, and then fixable."