How to Handle Subscription Cancellations (and Failed Payments)

Published July 25, 2026

Part of: Digital Products — our full guide on this topic.

Disclosure: Some links below are affiliate links. If you sign up through them we may earn a commission at no extra cost to you. We only recommend tools we'd genuinely suggest to a friend. See our full disclosure.

There is a good guide on this site for handling refund requests, and almost none of it helps you when a subscriber leaves.

A refund request is a conversation. Someone writes to you, you read it, you reply within a day, and you decide. Every piece of refund advice — respond fast, stay kind, read what they actually want — assumes a message arrived.

A cancellation is silence. Nobody writes to you. A member clicks a button in an account page you may never look at, and the first you know is a number that is smaller than last month’s. And the second way recurring revenue leaves is stranger still: a card expires, a charge fails, and a person who still wants your product stops paying for it without ever deciding to.

Those are three different events, and the refund playbook only covers one of them. This guide covers the other two.

Honest disclosure: some links below are affiliate links. If you sign up through one I may earn a commission at no extra cost to you. Everything here is my genuine assessment.

The quick version

• Split your churn in two before you react to it. Someone leaving and a card failing are different problems.
• Fix the plumbing first. Failed payments are the only revenue you can recover without changing your product.
• Let access run to the end of the paid period. Cutting it off instantly manufactures refund requests.
• Ask one question on the way out, and be careful how you read the answer.
• Offer a pause, not a discount. Most cancellations are a busy month, not a pricing objection.
• Measure when people leave, not just how many. The month tells you which problem you have.

Two ways the money leaves, and only one of them is about you

Every lost subscriber falls into one of two buckets, and lumping them together is how people end up fixing the wrong thing for months.

Voluntary churn is a decision. Someone looked at the charge, or looked at their calendar, or looked at what they got last month, and chose to stop. This is genuine feedback about your product, your price, your onboarding or your fit — and improving it means changing something real.

Involuntary churn is not a decision at all. A card expired. A card was replaced after a fraud alert. A bank declined a recurring foreign charge it did not recognise. The member has made no judgement about your membership whatsoever; they may not even know it has stopped. This is a billing failure wearing the costume of a cancellation.

On your revenue chart these look identical. In your business they are opposites. If you read a run of expired cards as a value problem, you will spend a month redesigning something that was working, and the people you actually lost were never unhappy — they just needed a link.

So before you interpret a bad month, split the number. Every billing tool distinguishes a cancellation from a failed charge, and that split is one of the few business metrics genuinely worth watching.

Failed payments: the cheapest revenue on the table

Start here, because this is the only part of churn you can improve without improving your product.

What actually happens when a charge fails

Most billing platforms handle the first line of defence for you with dunning: when a charge fails, the system retries it over the following days and emails the customer to update their card. Some of those retries succeed on their own — a card that failed on Tuesday because the account was briefly empty may well go through on Friday.

The part people skip is finding out what their own tool does by default. Retry schedules, how many attempts, whether the customer is emailed at all, and what happens to their access when the attempts run out are all configurable, and the defaults vary enormously between platforms. Go and read yours today rather than discovering it during the first failure. This is the same category of thing as knowing what your payment processor takes — unglamorous plumbing that quietly decides how much money you keep.

The decision you have to make in advance

When a charge fails, does the member keep access while the retries run, or lose it immediately?

For a subscription, the answer is nearly always keep it, for a defined grace period. The person has not left. They are mid-way through something they were happy to pay for, and locking them out over a card their bank replaced is a punishment for something that is not their fault. A grace period of a few days, ending when the retry sequence ends, costs you almost nothing and prevents the reaction you least want, which is a loyal member who feels accused.

Note that this is the opposite of the right answer for a payment plan on a one-off product, where a failed installment leaves an outstanding balance on something already delivered, and revoking access until they are current is a normal, fair policy. The difference is what the payment buys: a plan installment pays off a debt, and a subscription charge buys the next month. Do not carry the payment-plan policy across; they are genuinely different situations.

Whatever you choose, write it into the terms people agreed to at checkout, alongside your other basic legal pages. A policy someone agreed to in advance is a policy; the same policy applied for the first time after their card failed is a nasty surprise.

The update-your-card email

This email does one job and should look like it. Keep it short, say plainly which subscription is affected and what it costs, give one obvious button that goes straight to the billing page, and say what happens if nothing changes and by when.

Two things worth getting right, because both cost real money:

• Make it obviously not phishing. “Your payment failed, click here to update your card” is precisely what a scam email says. Send it from the address you always send from, use the product name they recognise, and do not invent urgency. A member who suspects a scam does nothing, and doing nothing is the outcome you are trying to prevent.
• Do not moralise. Nobody’s card failed on purpose. “Just letting you know your card didn’t go through” reads completely differently from “your payment is overdue.”

Prevent the ones you can see coming

The single most avoidable failure is a card that expires on a date you already know. If your billing tool can tell you which cards expire next month, that is a short, friendly, genuinely useful email to send before anything breaks — and one nobody resents, because it saves them a problem rather than reporting one.

When someone genuinely wants to leave

Now the other half: the member who has decided.

Let them go cleanly

The cancel button should be findable by someone who is mildly annoyed, and it should work without an email to you. Beyond the growing legal expectation that cancelling is about as easy as subscribing, the business case for friction is simply bad. A member you trap for one extra month pays you once more and then tells people how hard it was to get out. As the membership launch guide puts it: making cancellation hard buys you a few weeks of revenue and costs you every referral that person would have made.

Access runs to the end of the paid period. They paid for the month; give them the month. Ending access the instant they click cancel converts a tidy exit into a refund request, which is the most irritating possible support ticket — you are now negotiating with someone who has already left. Show the exact end date on the confirmation screen and repeat it in the confirmation email.

And send a confirmation email at all. A cancellation that produces no receipt leaves the member unsure whether it worked, and an unsure member checks their bank statement next month with suspicion rather than indifference.

One case reverses the direction of this entire section: you are the one leaving. Unpublishing the sales page does not cancel a single existing subscription — in most systems those are separate objects — so a membership you have stopped selling carries on billing people on the same dates as before. Every one of those charges is somebody paying for a month you may not be around to serve, so winding a business down starts from the payment list rather than the product list. If the business carries on and it is only the membership that is ending, that is its own sequence — closing a membership is the exact inversion of everything above, because you are cancelling all of them at once and none of them asked.

It’s worth reading everything above once from the other side, too. You are also somebody’s subscriber — to a page builder, an email tool, a scheduler — and every retention move on this page will be run on you by people who reached the same conclusions. That’s not a reason to distrust it; it’s a reason to know which move you’re looking at when you land on a cancel screen. Auditing your own software subscriptions is the buyer’s-side version of this article.

One honest offer, not a maze

You get one offer on the way out. Spend it well.

The reflex is a discount. Resist it. A retention discount tells the member your price was always negotiable, which you cannot un-tell them, and it typically delays the cancellation rather than reversing it — you have bought two months at a lower rate from someone who was leaving anyway, while every member who never threatened to go pays full price. That is a strange thing to reward.

A pause is the better offer, because it matches the most common real reason people cancel, which is not price and not disappointment but a month where they know they will not use it. “Pause for two months instead?” is a genuine kindness to someone in that position, and it keeps a member you would otherwise have to win back from scratch. A downgrade to a cheaper tier is the second-best version of the same idea.

What the cancel screen must not become is an obstacle course — three screens, a survey you cannot skip, a “are you sure?” that reappears. Offer once, accept the answer, cancel it.

The exit question

Ask one question. One.

A leaving member owes you nothing and their patience is measured in seconds, so a five-question survey gets you nothing but abandoned cancellations. A single optional question — what made you decide to stop? — with a free text box and a handful of pre-set options gets answered surprisingly often.

Then read the answers carefully, because the raw counts mislead:

• “Too expensive” is the polite default. It is the socially easy answer, it requires no explanation, and it does not blame anyone. Some of the people who tick it mean it. Many mean “not worth it to me”, which is a value problem wearing a price label. If you cut your price in response to this box, you may find the same people leave anyway, and now you earn less from the ones who stayed.
• “I wasn’t using it” is the most useful answer you will get, and it usually points at onboarding or at a rhythm that stopped being visible, not at the product’s quality.
• “I got what I needed” is not failure. Some memberships are legitimately finite for some members, and a graceful exit from a satisfied person is a future testimonial and a future re-join.
• Look at the free text, not the buttons. The pre-set options tell you what you already suspected. The typed sentences tell you what you did not.

Pair this with asking your audience directly while they are still members, because the exit question can only ever tell you about people who already left.

One answer deserves its own note because it is easy to misread: “it stopped being the thing I signed up for.” That is not a churn problem to be fixed with a better offer or a save email. It is feedback about a change you made — a format, a platform, a tier, a schedule — and the fix belongs in how you make changes to something people are already paying for, several months earlier, rather than in this conversation.

When a cancellation is really a refund request

The two estates collide in one very specific, very common ticket: “I forgot to cancel and you charged me again.”

This is not a refund request about your product. It is an admin accident, and the person is embarrassed and irritated in roughly equal measure. In most cases the right answer is to refund the charge, cancel the subscription, and say something warm — particularly when it is plain they have not logged in since the charge. The amount is small, the alternative is a chargeback that costs you the money plus a fee and counts against your payment account, and the refund guide’s core logic applies exactly: a fast, fair refund is cheap insurance against an expensive dispute.

Then fix the cause instead of getting better at the case:

• Send a renewal reminder before an annual charge. A year is long enough that people genuinely forget, and a reminder a week ahead turns a furious surprise into a considered decision. Some payment rules require this anyway; it is good practice regardless.
• Send a receipt for every charge, with the product name they will recognise. A large share of “I don’t recognise this charge” disputes are simply a business trading under a name the customer has never seen on a statement.
• Say the renewal date out loud at signup. “You will be charged $X every month until you cancel, and you can cancel any time from your account” costs you almost no conversions and removes an entire category of complaint.

If the relationship survives — and handled this way it usually does — that person is a perfectly good candidate for a win-back email later, and sometimes a review too. People remember how you behaved when they were leaving far more vividly than how you behaved when they joined.

What to actually track

Three numbers, and none of them is a benchmark you can borrow from someone else’s business.

1. Voluntary versus involuntary, separately. If a third of your losses are failed cards, your product is not the problem this month.
2. Which month people leave in. Early exits are an expectations or onboarding problem; later exits are usually a value or a library-exhaustion problem. The membership pricing guide works through what each pattern means for the price. If you are about to change that price, note that the cancellations worth reading are the ones in the first billing cycle after the change and the ones at the next renewal — raising a recurring price explains why one checkpoint is not enough.
3. The trend of your own rate, over several months. Published churn benchmarks vary so widely by price, audience and product type that comparing yourself to one tells you almost nothing.

These three sit inside a slightly wider set of numbers, because churn is only the losing half of the picture — what to track when revenue repeats covers the rest of it, including why a flat revenue line can mean either a stable membership or a treadmill, and how to tell which one you are on.

The recurring revenue projector is useful here in a way that a spreadsheet of last month’s losses is not: change the churn percentage and watch what happens to the total a year out. It makes the case for spending an afternoon on your dunning settings far better than any argument can, because the compounding is genuinely counter-intuitive until you see it move.

Doing this without a stack of paid tools

Most of the above is a billing setting and three emails, and you do not need expensive software for any of it. What you do need is for the checkout, the membership access and the email to be aware of each other — because the whole failure mode of a homemade setup is a cancelled member who keeps their access, or a recovered payment that never restores it.

That is the practical argument for keeping billing and access on one platform rather than stitching them together. Systeme.io is the one I point people to for this: the membership area, the recurring checkout and the email list live on one plan, with a free tier you can run a small membership on while you find out whether it works at all. (That is an affiliate link — if you upgrade to a paid plan through it I may earn a commission, at no extra cost to you; see the full affiliate disclosure.) If you are choosing a processor rather than a platform, Stripe vs PayPal covers the recurring-billing differences, and what a merchant of record is explains who is actually responsible for the charge on the statement. Features and free tiers change, so confirm the current details on the provider’s own site. The free membership site guide covers the whole build if you have not set one up yet.

The bottom line

Refund advice starts when a message arrives. Recurring revenue mostly leaves without one — quietly, through a cancel button you never see, or through a card that simply stopped working.

So handle the two separately. Fix the plumbing first, because failed payments are money you can recover without changing anything about your product: know your dunning schedule, keep access through the retries, send a short human email that makes updating a card a single click, and get ahead of the expiries you can already see. Then make leaving clean — a findable button, access to the end of the paid month, one honest offer that is a pause rather than a discount, one optional question, and a warm goodbye.

None of it is dramatic, and that is the point. A membership is not won or lost at the cancel screen. But the difference between a business that leaks members quietly and one that keeps the ones who never meant to leave is mostly this: a few settings nobody enjoys configuring, and the decision to treat someone walking out of the door as well as you treated them walking in.