Involuntary Churn: What the Published Numbers Actually Say
Why it costs more than voluntary churn, and what the published data supports. Every figure below is attributed, with a link. Where no reliable source exists, we say so instead of estimating.
When SaaS founders talk about churn, they usually mean voluntary churn—customers who actively cancel. Pricing tiers, onboarding, product stickiness. That gets most of the attention.
But there's a quieter killer: involuntary churn. Customers who wanted to stay, but whose card expired, whose bank blocked the charge, or whose wallet ran out of funds. They never clicked "cancel." They just stopped paying.
The actual numbers
Industry benchmark
Recurly's 2024 State of Subscriptions (2,200+ merchants, 58M subscribers, 2023 data) puts the median renewal-invoice decline rate at 8.3% on the first attempt — 6.1% on credit cards, 15.6% on debit — and the median dunning recovery rate at 49.0%. Median involuntary churn itself sits at 1.0% of subscribers per month. Stripe reports its own Smart Retries recover 57% of failed recurring payments on average. Both are vendor figures: each company sells a solution to the problem it measures.
Applied to a $50,000 MRR business: roughly $4,150 of renewal invoices fail on the first attempt each month. At the median 49% recovery rate, about $2,100 comes back and $2,050 doesn't — around $24,600 a year. Run no recovery at all and the full $4,150 a month is exposed. These are medians from other people's merchants, applied to your MRR as an approximation. The only number that means anything is your own.
Why involuntary churn is underaddressed
Three reasons:
- It's silent. No "cancel" email. No angry support ticket. The subscription just lapses.
- It's not "sexy". Founders don't talk about recovering failed payments at conferences.
- It requires infrastructure. You need retry logic, email dunning, customer-facing update pages, and analytics. Most small teams don't build this until it's a problem.
What separates top-quartile SaaS
Nobody publishes a clean top-quartile breakdown. Recurly shows percentiles inside its own product but doesn't publish the values or the sample size, and the P25/P75 tables circulating online have no study behind them. Here is what is published:
1. Recovery is front-loaded. 90% of recovered transactions happen within the first 10 days of the failure (Recurly). Whatever you do, do it early.
2. Insufficient funds dominates. Across 5.4M failed payments, it accounts for 40.5% of all failures; expired cards are only 1.1% (Churnkey with Stripe, 2025). That matters: a card that will work on payday needs patience, not a new card.
3. Attention decays fast. In the same dataset, dunning email open rates fall from 55.8% on the first email to 20.6% on the fifth. A long sequence is not a free lunch.
4. There is headroom. Recurly describes moving from 53% to 71% recovery as what an optimised strategy achieves — offered as a target, not as a measured quartile. We repeat it as a target, nothing more.
The lever most people miss
It is worth being precise about what Stripe already does, because a lot of writing on this subject — including an earlier version of this page — gets it wrong. Stripe Billing retries failed payments (8 attempts over 2 weeks by default, timed by an AI model), emails your customer after each failed attempt, hosts the page where they update the card and pay, updates card numbers automatically when the network issues new ones, and reports recovery analytics in your dashboard. None of that is missing.
What Stripe does not do is let you choose the words or the cadence. Its failed-payment email fires once per retry attempt, on the retry schedule, with copy you cannot change — only the logo and colours. If you want an escalating sequence on your own timetable, in your own voice, signed with your own name, that is the piece you have to add. That is the piece PaidGuard is.
See how much you're losing in 30 seconds.
Calculate your loss →Sources & methodology
PaidGuard has no first-party data to offer. We are new, and inventing a sample size would be worse than having none. Every figure above comes from someone else's published work: Recurly, The 2024 State of Subscriptions (2,200+ merchants, 58M subscribers, 2023 data); Churnkey with Stripe, involuntary churn benchmarks (5.4M failed payments, 2025); Stripe Billing for the 57% Smart Retries figure; and Stripe's revenue recovery documentation for what Stripe does natively. Recurly and Churnkey both sell recovery software, so treat their medians as vendor data. An earlier version of this page cited a Subscription Trade Association report and a 47-company sample; neither supported what was written here, and both are gone.