Subscription Billing Best Practices for Online Businesses

Jamal Brooks·5 min read
Subscription billing analytics with MRR chart

Key Takeaways

  • Involuntary churn from failed payments is substantially recoverable with configuration, unlike voluntary churn which requires persuasion
  • Card account updater is supported by most major processors and frequently never enabled, and it fixes failures with no customer involvement
  • Notify before card expiry, before annual renewals and before price increases: prevention is cheaper than recovery
  • Warn before the first post-trial charge, because charging without notice is the most reliable way to generate disputes
  • A customer who turned off auto-renew is still paying until the period ends, so never label them cancelled in your own reporting

Most subscription businesses spend their attention on cancellations and lose more money to failed payments. A customer who cancels made a decision. A customer whose card expired did not, and they will frequently not notice they have lost access until weeks later, by which point the habit is broken.

This is a guide to the billing mechanics that decide whether your recurring revenue actually recurs.

Involuntary churn is the bigger number

Churn splits into two kinds:

  • Voluntary churn: the customer chose to leave.

  • Involuntary churn: the payment failed and the subscription lapsed.


Involuntary churn is a meaningful share of total churn in most subscription businesses, and unlike voluntary churn it is substantially recoverable with configuration rather than persuasion.

The causes are mundane: cards expire, cards get reissued after fraud, banks decline for insufficient funds, and issuers decline recurring charges they judge suspicious.

The three controls that recover most of it

1. Card account updater. Card networks operate services that push updated card details to merchants when a card is reissued. Major processors support this and many businesses have simply never enabled it. It silently fixes a large fraction of failures with no customer involvement at all. Check whether yours is on.

2. Intelligent retries. A failed charge should not be retried on a naive fixed schedule. Retry timing matters: attempts near payday, attempts spaced to avoid triggering issuer fraud rules, and a limited number of attempts before giving up. Most processors now offer a smart retry option that outperforms any fixed schedule you would design.

3. Dunning that actually reaches people. The email sequence after a failed payment is usually an afterthought and it is doing real work.

What a functional sequence looks like:

  • Immediately: a clear notice that the payment failed, what happens next, and a one-click link to update the card. No login wall.

  • Day three: a reminder, with the date access will lapse stated explicitly.

  • Day seven: final notice before suspension.

  • After suspension: a win-back at two weeks and again at a month, because a meaningful share of lapsed customers return if reaching them is easy.


Two details that disproportionately matter: the update-card link must work without requiring a password reset, and the email must come from an address a human can reply to.

Pre-dunning: stopping the failure before it happens

Cheaper than recovering a failed payment is preventing it.

  • Notify before a card expires. You know the expiry date. A reminder two weeks before, with an update link, prevents the failure entirely.

  • Notify before annual renewals. Legally required in several jurisdictions and good practice everywhere. Customers who are surprised by a large annual charge dispute it; customers who were told do not.

  • Notify before a price increase, with enough notice to cancel. Surprise increases generate chargebacks, which are far more expensive than the churn you were trying to avoid.


Trials and the first charge

The transition from trial to paid is where a large share of subscribers are lost.

Card up front or not? Requiring a card at trial start produces fewer trials and a much higher trial-to-paid rate. Not requiring one produces more trials and more support load. Neither is universally right; what matters is that you measure paid conversions rather than trial signups, because optimising trial volume can reduce revenue.

Always warn before the first charge. Charging at the end of a trial without notice is the most reliable way to generate disputes. A reminder three days before, stating the amount and the date, costs a small number of cancellations and prevents a larger number of chargebacks.

Proration, upgrades and downgrades

Getting this wrong generates support tickets forever.

  • Upgrades should generally take effect immediately with a prorated charge for the remainder of the period. This is what customers expect.

  • Downgrades should generally take effect at the end of the current period, not immediately. Refunding the difference mid-cycle invites gaming.

  • Cancellations should retain access until the end of the paid period. Cutting access at the moment of cancellation for a period already paid is the kind of thing customers write about publicly.


Whatever you choose, state it on the billing page in plain language. Most proration complaints are about surprise rather than about the policy.

Cancellation flow

There is a real difference between reducing friction and creating a trap. Regulators in several jurisdictions now require cancellation to be as easy as signup, and the direction of travel is clear.

What is legitimate:

  • Offering a pause instead of a cancellation. Genuinely useful and frequently accepted.

  • Offering a downgrade to a cheaper plan.

  • Asking one question about why, optionally.


What is not:

  • Requiring a phone call to cancel when signup was one click.

  • Hiding the cancellation option.

  • Multiple confirmation screens designed to exhaust.


The last category generates chargebacks, regulatory attention and public complaints. It also does not work for long: a customer who cancels after being obstructed does not come back.

The metrics worth watching

MetricWhat it tells you
Payment failure rateHow much revenue is at risk before any customer decision
Recovery rateWhat share of failed payments you win back
Involuntary share of total churnWhether your problem is product or billing
Trial to paid conversionWhether the trial is qualifying or just generating volume
Cancellation reason mixWhether churn is price, product or fit

The second row is the one to improve first, because it is the cheapest revenue available to any subscription business.

Cancelled is not the same as expired

A terminology point with real consequences: in most billing systems, a subscription where the customer has turned off auto-renew is still active and still paying until the period ends. Labelling that customer "cancelled" in your own reporting will understate current revenue and, worse, may cut their access early.

Keep the distinction explicit: not renewing means still paying until a known date. Expired means the access period has ended. Conflating them produces both bad numbers and angry customers.

Where this connects

If you run an affiliate program on top of a subscription product, all of the above affects your partners as well. Commissions on a subscription that lapses involuntarily are commissions your partner loses through no fault of their own, and recovering the payment recovers their income too. The affiliate payout strategies piece covers how reversals should be communicated, and recurring revenue models covers why recurring commissions change partner behaviour so sharply.

For the underlying payment mechanics, how online payments work explains why a decline is not always a failure and what the different decline codes actually mean.

paymentssubscriptionsbillingchurn

Written by Jamal Brooks

Jamal is a product engineer at Affiliateo who writes about payments, integrations, and technical best practices.

Frequently Asked Questions

Related Articles