A subscriber who clicks "cancel" and a subscriber whose card just stopped working look identical on most churn dashboards: one fewer active subscription this month. They're not the same event, and they don't have the same fix. The first is a decision. The second is a Tuesday afternoon where a bank's fraud filter had a bad day, or a card expired and nobody updated it, or a network hiccup returned a decline that had nothing to do with the balance on the account. Subscription and billing tools call this second category involuntary churn, and it's the one category of churn that isn't really about whether the customer wants to keep paying you.
We went looking for the actual scale of the problem, because the number that circulates most — that half of all subscription churn comes from failed card payments, and 80% of that isn't the customer's fault — sounded too clean. It traces to a real study, but not an independent one. What we found instead was a smaller, disclosed-methodology figure from an actual billing platform's own transaction data, a mechanical fact about card networks that almost nothing else written on this topic mentions, and a genuinely useful distinction between the kind of decline worth retrying and the kind that just burns money on every attempt.
Key takeaways
- "50% of subscription churn is failed card payments, 80% of which isn't the customer's fault" traces to a PYMNTS study of 200 subscription businesses commissioned with FlexPay — a payment-recovery vendor with a direct commercial interest in that number being large. Treat it as vendor-sponsored, not independent.
- The more conservative, disclosed-methodology figure comes from Recurly's own billing-platform data: an average total churn rate of 3.60%, split 2.34% voluntary and 1.25% involuntary — involuntary churn accounts for roughly 35% of total churn, landing inside the independently-repeated 20–40% range even though the flashier 50%/80% figures don't hold up to sourcing.
- Not every decline is the same kind of failure. A soft decline — insufficient funds, a temporary issuer outage, a generic "do not honor" — often clears on a later attempt. A hard decline — an expired card, one reported lost or stolen, an invalid card number — will not, no matter how many times you retry it.
- Mastercard's Transaction Processing Excellence program charges merchants for retrying a card past 10 declined attempts in 24 hours (or 35 in a rolling 30 days) — a fee that rose from $0.30 to $0.50 per attempt in January 2025. Visa runs a comparable penalty for repeated reattempts on a card that will never approve, though industry write-ups don't agree on its exact name or threshold. A dunning sequence that blindly retries every decline the same way isn't just annoying a customer — on a hard decline, it's actively paying a card network to do it.
- The two recovery mechanisms with the clearest, most verifiable track record are also the least dramatic: Visa's Account Updater and Mastercard's Automatic Billing Updater silently push a cardholder's new number or expiry date to enrolled merchants when their bank reissues a card — recovering revenue before a single dunning email ever goes out.
What's actually the difference between a customer who cancels and a card that just stops working?
Voluntary churn is a subscriber actively deciding to leave — they click cancel, or they let a plan lapse on purpose. It's a signal about the product, the price, or a competitor: something about the relationship changed and they acted on it.
Involuntary churn is a subscriber losing access because a scheduled charge didn't go through — a card expired, a bank's fraud system flagged the transaction, an issuer's system was briefly unreachable, or a dozen other reasons that have nothing to do with whether the person still wants what they're paying for. Most billing systems still record both outcomes as the same event — one fewer active subscription — which is exactly why the two get conflated in casual reporting on "why customers leave."
The distinction matters because the fixes don't overlap. A voluntary-churn problem is a product, pricing, or retention-offer problem. An involuntary-churn problem is a payments-infrastructure problem — and no amount of improving the product fixes a card that silently expired three renewals ago.
How big is involuntary churn, really — and where does the "50% of churn" number come from?
The number that gets repeated most often in dunning and payment-recovery content is specific and alarming: roughly half of all subscription churn comes from failed card payments, and about 80% of those failures aren't anything the customer did or could control. It traces to a real, named study — PYMNTS ran it in partnership with FlexPay, surveying roughly 200 subscription businesses. It's worth knowing that before treating it as independent evidence: FlexPay sells third-party payment-recovery software, and a headline number this large is also the number that makes their product look most necessary. Real data, real vendor interest — the same pattern as a lot of the statistics that circulate in this space.
The more conservative figure comes from a source with a different kind of authority: Recurly, a subscription-billing platform, publishing benchmarks drawn from its own network's transaction data rather than a survey of opinions. As of its July 2026 update, the all-industry average sits at a 3.60% total churn rate, split 2.34% voluntary and 1.25% involuntary — involuntary churn works out to roughly 35% of total churn. That's a meaningfully smaller claim than "50% of churn," but it lands squarely inside the 20–40% range that shows up independently across multiple payment-recovery vendors' own materials — which is the sign of a number that's at least directionally real, unlike the more specific 50%/80% figures that only trace back to one commercially-interested source.
The industry breakdown is where the picture gets more useful than either headline number: involuntary churn isn't a flat tax across every business. It scales with how the business collects payment in the first place.
SaaS
B2B & professional services
Travel & hospitality
Digital media
Ecommerce
Education
Education and ecommerce carry the heaviest involuntary-churn load in this data, likely because both skew toward higher-turnover payment methods (student cards, one-off debit cards) compared to a B2B SaaS renewal running on a company card with an account owner tracking it.
What's a realistic industry-wide figure for the cost of failed payments?
In January 2024, Recurly published its own forecast — built from its billing platform's transaction data, not a survey — projecting that failed recurring payments would cost subscription businesses more than $129 billion globally in 2025. It's a forecast from a company that processes real recurring charges at scale, which is a meaningfully different kind of number than a percentage pulled from a 200-company survey — but it's still a projection, not an audited outcome, and it's still published by a vendor whose product addresses exactly this problem. Read it as the most credible large-scale estimate available, not as a measured fact.
$129B
Recurly's January 2024 forecast for what involuntary churn would cost subscription businesses worldwide in 2025 — a projection built from the billing platform's own transaction data, not an audited industry-wide result.
Source: Recurly, "Failed payments could cost subscription companies more than $129 billion in 2025" (press release, January 2024)
Why does a card actually stop working — and does retrying it help?
Every decline a card issuer sends back falls into one of two structurally different categories, and the categorization is standard across the payment industry's own decline-code documentation (Visa, Mastercard and every major processor publish some version of the same list) — not a distinction one vendor invented to sell a product.
| Category | What's actually happening | Worth retrying the same card? |
|---|---|---|
| Soft decline | Insufficient funds at the moment of the charge, a temporary issuer-system outage, a generic "do not honor" response, or a processing error on the issuer's side | Often, yes — funds land, the issuer's system recovers, or the temporary condition clears within a day or two |
| Hard decline | The card is reported lost or stolen, has expired, the card number is invalid, or the account is closed | No — nothing about the card changes between attempt one and attempt ten, and the issuer will keep returning the same answer |
Treating every decline the same way — retry on a fixed schedule regardless of the reason code — recovers the soft declines you'd have recovered anyway and wastes every attempt on the hard declines, which brings up the part almost nothing written about dunning mentions.
The retry penalty nobody mentions: card networks now fine you for retrying too much
Mastercard runs a program called Transaction Processing Excellence, or TPE, aimed at exactly this behavior. Once a merchant sends more than 10 declined authorization attempts on the same card, at the same card-acceptor ID, within a 24-hour window — or more than 35 within a rolling 30-day period — every additional attempt is billed as an excessive-authorization fee. That fee rose from $0.30 to $0.50 per attempt effective January 2025. Mastercard's own guidance to acquirers is blunt about the fix: stop sending authorizations on the same card once ten declines have come back in 24 hours, and use an Account Status Inquiry instead of a live charge attempt to check whether a card is still valid.
Visa runs a comparable penalty for the same underlying behavior — repeatedly resubmitting a charge on a card that has already returned a decline code meaning it will never approve — though payments-industry write-ups don't agree with each other on what to call this specific fee or its exact threshold, and card-network fee schedules are proprietary and revised annually. The mechanism is the part worth taking away even where the exact dollar figure isn't: retrying a hard decline isn't a free extra attempt at revenue. On a card that was never going to approve, every retry is a small, real cost with zero chance of the charge succeeding.
If retry logic runs through your payment processor rather than something you built yourself, check what retry cap and cadence it applies by default — "smart retries" is a specific, bounded strategy, not a synonym for "try again until it works."
What card networks' own tools already fix, before a single email goes out
The two recovery mechanisms with the most verifiable track record aren't dunning emails at all — they're services the card networks themselves operate. Visa's Account Updater and Mastercard's Automatic Billing Updater sit between a merchant's stored card token and the issuing bank: when a cardholder's bank reissues a card — a new expiry date, a new number after a reported-lost replacement — the updater pushes the new details to every enrolled merchant with that card on file, automatically, with no customer contact required.
Independent payments-consultancy estimates put the incremental recurring revenue recovered this way at roughly 3–5% — a vendor-adjacent estimate rather than a figure either card network publishes directly, so treat the precise percentage with the same caution as any other unaudited number in this space. What's not in dispute is the mechanism: a subscriber whose card was quietly reissued by their bank never has to notice anything went wrong, because the charge that would have failed simply doesn't.
So what does an actual recovery sequence look like?
None of the above argues for ignoring failed payments, or for assuming a card-network service alone will cover the gap. It argues for building a sequence that matches the actual mechanics above, instead of one generic retry-and-email loop applied to every decline the same way.
- 1Enable Visa Account Updater and Mastercard Automatic Billing Updater (or confirm your payment processor already runs them) — this is the recovery that happens before a customer ever sees a failed-charge notice, and it costs nothing per attempt the way a retry does.
- 2Split retry logic by decline category, not by a flat schedule. A soft decline earns a small number of spaced retries; a hard decline should stop immediately and move straight to a customer-facing message — retrying it further only risks the network penalty above for zero chance of success.
- 3Cap retries to what the card networks actually enforce — under Mastercard's threshold of 10 declines per card per 24 hours — regardless of what your own dunning tool's default schedule assumes.
- 4Give a failed payment exactly the same visible ownership a routing rule gives a lead: a clear status, a first-contact target, and an escalation path if it's still unresolved after a set number of days — not a silent retry loop nobody is watching.
- 5Make the customer-facing ask easy to act on the same day it lands: a hosted payment page or a one-tap pay link reaches someone faster than a form asking them to re-enter a card number from scratch.
Where oToK fits into this directly: a recurring charge you collect through your own Cardcom or Sumit account runs on auto-generated cycles with smart retries built in when a charge fails, and a saved card means the customer never re-enters details for the retry to work. A payment request goes out by email with automatic reminders, and you can send the same pay link in a WhatsApp message when email alone isn't landing fast enough. Every payment — successful or not — lands on the contact's own record next to their full LTV, so a failed charge shows up next to the exact customer it affects instead of in a separate billing tool nobody on the team checks.
Frequently asked questions
- Is it true that failed payments cause 50% of subscription churn?
- That specific figure traces to a PYMNTS study of about 200 subscription businesses, commissioned with FlexPay — a payment-recovery vendor with a direct commercial interest in the number being large. A more conservative, disclosed-methodology figure comes from Recurly's own billing-platform data: involuntary churn (1.25%) works out to roughly 35% of total churn (3.60%) on average across industries, which lands inside the independently-repeated 20–40% range even though the specific 50%/80% figures don't trace to independent sourcing.
- What's the real difference between voluntary and involuntary churn?
- Voluntary churn is a subscriber actively choosing to leave — they cancel, or let a plan lapse on purpose. Involuntary churn is a subscriber losing access because a scheduled charge failed — an expired card, a fraud flag, a temporary issuer outage — with no decision to leave behind it. Most billing dashboards record both as the same event, which is why they get conflated in casual reporting on churn.
- Should I keep retrying a declined card until it goes through?
- Only if it's a soft decline — insufficient funds, a temporary issuer outage, a generic "do not honor" response — where a later attempt can genuinely succeed. A hard decline (expired card, reported lost or stolen, invalid card number) will return the same result every time, and retrying it repeatedly can trigger real per-attempt penalty fees from the card networks, on top of having zero chance of recovering the charge.
- What is a card network "excessive retry" fee?
- Mastercard's Transaction Processing Excellence (TPE) program charges merchants an excessive-authorization fee — which rose from $0.30 to $0.50 per attempt effective January 2025 — once more than 10 declined attempts hit the same card and card-acceptor ID within 24 hours (or 35 within a rolling 30 days). Visa charges a comparable penalty for repeatedly resubmitting a charge on a card that has already returned a decline meaning it will never approve, though the exact name and threshold aren't consistently reported across payments-industry sources — check your acquirer's current fee schedule rather than relying on a fixed figure.
- What actually recovers a failed payment, besides emailing the customer?
- The most verifiable mechanism is the card networks' own account-updater services — Visa Account Updater and Mastercard Automatic Billing Updater — which silently push a cardholder's new card number or expiry date to enrolled merchants whenever their bank reissues a card, recovering the charge with no customer contact at all. Beyond that, splitting retry logic by decline reason (retry soft declines, stop immediately on hard declines) and making the customer-facing ask a one-tap pay link rather than a re-entry form both address specific, documented parts of the problem — as opposed to a single generic retry-and-email sequence applied the same way to every failure.
