Failed payment on a SaaS subscription: recovering it without losing the customer (dunning)
Published on 8 September 2026 · 8 min read
Two subscribers cancel your SaaS plan this month. The first compared options, hesitated, then clicked "cancel" fully aware of what they were doing: that's voluntary churn, and it's addressed upstream, through the perceived value of the product. The second decided nothing at all — their card expired last month, or their bank blocked a charge it flagged as unusual, and the subscription simply stopped without them noticing. That's involuntary churn, and it isn't solved with a better product: it's solved with a well-built payment recovery flow ("dunning"). Both look identical on the dashboard — one fewer subscriber — but they have neither the same cause nor the same fix, and confusing the two means treating what is, at its core, a plumbing problem with marketing content.
Why a recurring payment fails
A failed payment is almost never a deliberate refusal from the customer. Technical causes dominate by far.
- Expired card — the most common cause, and the easiest to anticipate: the expiration date is known the moment the card is saved.
- Insufficient funds or limit reached — common at the start or end of the month, often temporary: the same attempt would succeed a few days later.
- Bank anti-fraud block — some banks cautiously block an unusual recurring charge, especially after a change in amount or a long pause.
- Incomplete 3D Secure — strong authentication fails with no possible human intervention during an automatic charge, unlike a direct payment.
- Lost, stolen, or replaced card, with a new number the customer hasn't yet updated with you.
In almost all of these cases, the customer made no decision to leave. That's what sets dunning apart from classic retention: the goal isn't to convince, it's simply to make updating the card as easy and visible as possible before the subscription cuts off — the same low-friction logic covered in our article on the landing page form, applied here to a card-update page instead of a signup form.
The retry mechanics: how many attempts, and how far apart
Retrying the same charge an hour after failure changes nothing if the cause is a limit reached or insufficient funds: the same card, in the same state, produces the same result. The core of effective dunning lies in spacing out attempts, not in how many there are.
- Space attempts over several days, typically 3 to 4 attempts spread over 2 to 3 weeks: a balance that was short on a Monday can be enough the following Friday, after a paycheck lands, for instance.
- Vary the day of the week and time rather than always retrying at the same moment — some billing tools (Stripe Smart Retries, Chargebee, GoCardless) automatically schedule retries at the statistically most favorable times.
- Warn before the last attempt, not only after the final failure: a message like "your card expires this month" sent ahead of the renewal date prevents the incident rather than fixing it afterward.
- Set a clear limit (number of attempts or duration) beyond which the subscription is cleanly suspended — leaving an account in an indefinite in-between state complicates support and skews retention metrics.
The recovery email: what actually works
The simple act of sending a reminder has a measurable, well-documented effect on financial behavior: a study by Karlan, McConnell, Mullainathan and Zinman published in 2016 in Management Science, "Getting to the Top of Mind: How Reminders Increase Saving", shows across three field experiments that plain reminders — with no financial incentive attached — significantly increase follow-through on a financial commitment already made, by bringing the intention back to mind at the right moment. A payment recovery email works on the same principle: the card wasn't updated not out of refusal, but because the matter had simply slipped the customer's mind in the meantime.
Personalizing the message also matters more than one might expect. A field study by Karlan, Morten and Zinman among borrowers in the Philippines, "A Personal Touch: Text Messaging for Loan Repayment", tested several variants of loan-repayment reminder texts: only messages explicitly naming the loan officer already known to the customer meaningfully improved repayment rates, with no notable effect from loss versus gain framing or send timing. Applied to SaaS dunning, the takeaway is direct: a generic email ("Your payment failed") converts worse than a message that names the specific account involved, the likely technical reason for the failure when known (expired card rather than a vague "payment error"), and what is concretely at stake (loss of access to a given feature) instead of a vague, anxiety-inducing line.
- An explicit subject line: "Your [Product Name] subscription payment couldn't be processed" rather than something vague buried in the inbox.
- The likely cause named when the payment provider reports it (expired card, insufficient funds): it points directly to the action needed.
- A single link to a card-update page, with no requirement to log back into the full product to reach it.
- A concrete deadline ("access suspended on September 14 if the card isn't updated") rather than a vague threat, consistent with the legitimate-urgency principle covered in our article on urgency and scarcity.
- A factual tone, never guilt-tripping: the customer very likely did nothing wrong.
The card-update page: the part that gets neglected
The recovery email is useless if the page it points to reintroduces friction. Three mistakes come up often: requiring a full product password when a one-time link would suffice for this single action; showing a generic card form that mentions neither the amount, the product, nor the due date — enough to make a customer already caught off guard start doubting; and redirecting to the product's full login page rather than straight to the billing screen. The principle is the same as for any conversion-oriented form: fewer clicks between intent and action, one task per screen, context shown rather than assumed — see our guide on landing page cognitive load. The main billing providers (Stripe Billing, Chargebee) offer ready-made hosted update pages that already follow these rules, and are worth preferring over a custom-built page unless strong visual customization is specifically needed.
Where this starts, upstream, on the landing page
Dunning fixes the problem after the fact, but part of it is prevented even before the first charge. A SaaS landing page that clearly explains, right at trial signup, what will happen at renewal — a topic covered in our article on free trial length and in the one on free trial with or without a card — reduces the surprise, and with it the number of cards the customer never actually kept track of. The choice between a one-time payment and a recurring subscription, covered in one-time payment or subscription, also determines whether dunning is a one-off concern or something that comes up on every billing cycle.
If you're launching a SaaS offer and the signup page itself is still to be built, LanderKit's SaaS waitlist template ships a proven structure — value proposition, social proof, a minimal form — for €89 as a one-time payment, or in the full 10-template bundle for €229. Dunning itself is then handled on the billing tool's side; the landing page's main job is to state the rules of the game clearly from the start.
FAQ
Frequently asked questions
What is involuntary churn, and how does it differ from regular churn?
Voluntary churn is a customer actively deciding to cancel. Involuntary churn happens when a subscription stops because of a technical payment failure (expired card, insufficient funds, bank block) with no decision from the customer at all. The first is addressed through product value; the second through payment recovery (dunning) and a frictionless card-update page.
How many times should a failed payment be retried?
Generally 3 to 4 attempts spread over 2 to 3 weeks rather than clustered together, since a limit reached or insufficient funds resolves with time, not immediate repetition. Billing tools like Stripe or Chargebee automatically schedule these retries at the statistically most favorable times.
Is a simple reminder email enough to recover a failed payment?
A significant share, yes: a study by Karlan, McConnell, Mullainathan and Zinman (2016, Management Science) shows that plain reminders increase follow-through on a financial commitment already made, with no added incentive. Personalization improves the result further, as shown in a separate field study by Karlan, Morten and Zinman on loan-repayment reminders.
Should customers be forced to log back into the product to update their card?
No — that's unnecessary friction that lowers the recovery rate. A one-time link leading straight to the billing screen, without going through full product login, converts better. Stripe Billing's and Chargebee's hosted update pages are built for exactly this.
Read next
Related articles
- One-time payment or subscription: which pricing model for your landing page?"Do we sell it as a one-off or a subscription?" A question settled in five minutes inside a Stripe config actually determines the entire structure of the landing page selling the offer: the objections to overcome aren't the same, neither are the trust signals to show, nor the CTA. The simple test to decide, and what changes on the page depending on the answer.
- The SaaS subscription cancellation page: France's "3-click" law, and how to fit in an honest retention offerA SaaS team reworks its pricing page A/B test for the tenth time, while its cancellation page requires a phone call during office hours and three back-and-forth emails. The first page is optional polish. The second is, since June 2023, a precise legal requirement in France. Here's what it mandates, and how to fit in a retention offer that doesn't cross the line.
- Free trial: should you ask for a credit card or not?It's one of the most structural decisions in a SaaS offer: ask for the credit card at free-trial signup, or ask at the end? The first choice drastically filters out the curious; the second fills the pipeline. Neither is "right" in the absolute — but your situation almost always leans one way.