Blog

The first hour after a payment gateway outage: a checklist for subscription teams

When your payment gateway has a bad hour, renewals fail and customers notice. Here is what to check, in what order, and what to tell customers, without making it worse.

Payment gateways go down. Usually it's a partial outage, for some card types or regions, for twenty minutes. If it lands during your renewal run, a batch of good customers gets declined through no fault of their own, and some of them will email you, worried their subscription is gone.

Here is a calm first hour.

Minutes 0 to 10: confirm it's the gateway

A rise in declines can be the gateway, a bank, fraud, or a change you made. Before you do anything, narrow it down:

"What share of charges declined in each of the last six hours, by gateway?"

  • One gateway spiking, others normal: probably that gateway. Check its status page.
  • Every gateway spiking: look at what changed on your side, such as a checkout release or a new pricing page.
  • Mostly new signups from one traffic source: probably fraud, not an outage.

Minutes 10 to 20: read the decline reasons

"For declines in the last two hours, what were the reasons, with counts?"

An outage tends to show up as errors and generic declines, rather than the usual mix of insufficient funds and expired cards. If you see the usual mix, it may not be an outage at all.

Minutes 20 to 30: size the damage

"How many renewal charges declined since the problem started, and how much was due?"

"Which customers had a renewal decline in that window? Give me the list."

Keep that list. It's who to reassure, and who to check on tomorrow.

Minutes 30 to 45: let retries do their job

Renewals that fail go into Recurly's dunning schedule and are retried automatically. For a short outage, the first retry will usually succeed once the gateway recovers. Avoid manual retries in a panic: retrying many cards at once during an outage just adds more declines.

Minutes 45 to 60: talk to customers who contact you

Keep it short and calm: "There was a temporary problem processing payments this morning. Your subscription is fine, and we'll retry the payment automatically. You don't need to do anything." Don't promise a time unless you know your retry schedule.

Afterwards: make sure you hear first next time

"Alert me if the decline rate on renewals goes above 30% in any hour."

A watch like that tells you within 15 to 60 minutes, by email or Slack, instead of when customers start writing in. Retrixus shows you first how it would have behaved over your recent data, so the line is set where it means something.

And tomorrow, check the list from minute 20: "Of the customers whose renewal declined yesterday morning, how many have paid since?"

Start free on one Recurly site, or read about watches and alerts.

Try it on your own Recurly data

Start free on one Recurly site. Connect Claude or ChatGPT and ask your first question in about 15 minutes.

Start free