← All alert recipes Alert recipes

How to get alerted when a payment fails

Most failed payments are a customer who wanted to buy and could not, which is a different problem from fraud.

Alert on payment failures in batches rather than individually, and treat a cluster within a short window as a possible gateway or configuration problem rather than a run of unrelated declines. Individual failures are usually recoverable revenue with a short half-life; a sudden cluster is an incident.

Why this one is worth an interruption

The single failure and the cluster need completely different responses, which is why alerting on each one individually is a mistake. One is a customer to help. Twenty in an hour is something broken on your side or your provider's.

Individual failures are also more recoverable than most stores assume. A card declined at checkout is frequently a limit, an expiry or an address mismatch, and a prompt, plain message recovers a meaningful share of them.

The threshold

The interesting signal is rate, not the individual event.

ConditionSet it toWhy that number
Cluster triggerfailures above your normal rate within an hourThis is the incident case
Individual triggeronly above a value floorBelow it, recovery costs more than it returns
Baselineestablish your normal rate firstEvery store has a nonzero failure rate
Repeat customer flagexisting customer failsMore likely a fixable card problem than fraud
Quiet hoursnone for the cluster triggerA broken gateway does not wait

Who should receive it

Split by which case fired. They go to different people.

  • Whoever handles payments or the storefront, for the cluster trigger.
  • Whoever contacts customers, for individual high-value failures.
  • Not everybody. A steady trickle of declines is normal and broadcasting it is noise.

What to do when it fires

  • On a cluster, check your payment provider's status page before investigating your own store.
  • Check whether anything changed recently: a theme deploy, an app install, a payment setting.
  • On an individual high-value failure, contact the customer with a plain, non-accusatory message and an easy way to retry.
  • Record the reason codes. A pattern in them usually points at one fixable configuration issue.

When not to set this alert

Do not alert on every individual failure. Every store has a background rate, and a notification per decline trains the team to ignore the channel within a week.

Do not set the cluster trigger before you know your normal rate. Without a baseline you cannot tell a spike from a busy afternoon.

Common questions

Why do payments fail at checkout?

Most commonly card limits, expiry, address verification mismatches or bank-side risk rules. Genuine fraud attempts are a smaller share than most stores assume.

Should I alert on every failed payment?

No. Every store has a background failure rate, and alerting on each one makes the channel unreadable. Alert on clusters above your normal rate, and individually only above a value floor.

What does a sudden spike in failures mean?

Usually a configuration or gateway problem rather than a run of bad luck. Check your payment provider's status and anything that changed on the storefront recently.

Related