Bank or PayNow Outage on a Peak Trading Day: What Is Your Backup Way to Get Paid?

If your bank, your card acquirer or PayNow goes down during peak trading, your sales stop only for as long as you have a single way to get paid. The fix is not a better bank. It is a second payment rail that is already live and tested, one named person who can authorise the switch without calling you, and a short script so counter and chat staff do not freeze. Plan for an outage of one to four hours, on a Saturday or a campaign day, with no warning. If that plan exists on paper before November, an outage costs you an hour of awkwardness instead of a day of revenue.

What actually goes down when staff say the payment system is down?

Four separate things get reported to you with the same sentence, and they need different responses.

• The bank itself. Internet banking and the app are unreachable. You cannot see balances or release payments, but your card takings may still be settling normally behind the scenes.
• PayNow and instant transfers. These ride on the shared FAST rails between banks. When that layer is degraded, customer transfers either fail outright or sit pending, which is worse, because the customer believes they have paid.
• Card acquiring. Your terminal or online checkout cannot authorise. This is your acquirer or payment gateway, a different supplier from your bank in most cases, and it fails independently.
• Your own internet or power. The most common cause by far in a shophouse or a small office, and the only one you control.

Write those four lines on one page. Half of the panic in an outage is twenty minutes spent arguing about which one it is.

How much does one hour of payment downtime actually cost you?

Do this arithmetic once, in October, because it decides how much effort the plan deserves. Take your takings on your busiest recent Saturday, divide by trading hours, and that is your hourly exposure at normal load. For a retail or food business turning over 6,000 dollars on a good Saturday across ten hours, one lost hour is roughly 600 dollars. On a campaign day at three times normal volume, the same hour is closer to 1,800 dollars.

Then add the part most owners miss: the orders that never come back. A customer who cannot pay at a counter usually waits. A customer who cannot pay in an online checkout leaves, and you paid for that click. If your marketplace or search advertising costs 12 dollars to land an order, an hour of failed checkouts burns the advertising spend as well as the margin.

Against that, a second rail costs almost nothing to hold open. The asymmetry is the whole argument.

What is the minimum backup for a Singapore SME?

The rule is that your backup must not share a supplier or a rail with your primary. Two card terminals from the same acquirer is not a backup. Realistically:

• For a counter: a card terminal from one provider, plus a PayNow QR tied to an account at a different bank from your main operating account. Print the QR on card stock and keep it in the drawer. Screens die; paper does not.
• For online: a second payment method already enabled in the checkout, even if it is a minority of volume on a normal day. A method you can switch on in ten minutes is useful. A method you have never processed a live transaction through is not.
• For invoiced B2B work: bank details on the invoice from two banks, and a stated policy that a payment date is met by either.

Open the second bank account now if you do not have one. Account opening for a company takes days to weeks, and no amount of urgency on 11 November compresses that.

Who is allowed to switch to the backup, and how fast?

This is the part that fails, not the technology. Name two people who may declare an outage and switch methods without asking the owner, and give them a trigger that needs no judgement: three consecutive failed transactions, or five minutes of failures, whichever comes first. Then they act, and they tell you afterwards.

If the only person who can authorise a workaround is the boss, and the boss is in a meeting or on a plane, your real recovery time is however long that meeting lasts. Write the two names and the trigger into the same one-page sheet.

What should staff say to a customer who cannot pay?

Give them the words. Staff improvising an explanation is how an hour of supplier downtime becomes a review that says your shop is disorganised.

Something as plain as: the bank transfer network is having a problem across Singapore right now, it is not just us, here is our QR code from another bank, or we can take a card. For online, a banner at checkout and a pinned reply for chat and social messages, written in advance. Say the word network or bank, never the word our system, because one is a national event and the other sounds like negligence.

One more instruction matters: never accept a screenshot of a pending transfer as proof of payment during an outage. Pending transfers in a degraded rail are exactly where opportunistic loss happens. Release goods on a confirmed credit, a card approval, or a deliberate decision by a named manager who writes it down.

What do you reconcile the next morning?

Outages leave a mess that is cheap to clear on day one and expensive at year end. The morning after, check three things: duplicate charges from customers who retried, transfers that landed hours late against orders you already treated as unpaid, and anything released on a promise. Then reconcile your card settlement against your orders for the affected window rather than for the whole day, so a gap is visible instead of averaged away.

If you accepted cash more than usual, bank it and record it the same week. An unrecorded cash spike is a GST and audit problem you will have forgotten the reason for by March.

What should be in place before 11.11 and December?

A realistic October checklist, achievable in an afternoon:

1. One page listing the four failure types, the two authorised deciders, and the switch trigger.
2. A second bank account open, with a printed PayNow QR in the drawer and the same details on your invoice template.
3. A live test: process one real transaction through every backup method this month, then refund it. Untested backups are decorations.
4. Pre-written customer wording for counter, checkout banner and chat.
5. Your acquirer and bank status page bookmarked, plus the business support number saved in two staff phones, not only yours.
6. A float of cash and a plan for change, because a card and transfer outage together is not hypothetical during a power or internet failure.

The banks themselves are held to tight availability standards by the regulator, and major outages are rare. Rare is not never, and the ones that happen cluster on high-volume days because that is when systems are under the most load. Your protection is not prediction. It is holding a second way to get paid, and having decided in advance who says go.

Frequently asked questions

Should we keep a second bank account just for outages, even if it adds fees and admin?
Yes for most trading businesses, because the account earns its keep beyond outages. It separates operating cash from reserves, gives you a second set of transfer limits and approvers, and removes the single point of failure that is one bank relationship. Account fees for an SME account are typically a small monthly cost against an hourly revenue exposure in the hundreds or thousands. The honest cost is not the fee, it is the extra reconciliation each month, so decide who owns that before you open it.

Can we claim lost sales from the bank or payment provider?
Almost certainly not. Standard payment and banking terms exclude consequential loss such as lost profit, and service credits where they exist are small and tied to fees paid, not to your revenue. Treat recovery from the supplier as unlikely, and treat the backup rail as your actual insurance. If an outage caused a material loss, document the window, the failed transaction count and your takings for the equivalent period the week before, because that record is also what your cyber or business interruption insurer will ask for if a policy does respond.

How is this different from our cyber or ransomware plan?
Different cause, overlapping response. Ransomware takes out your own systems and your data, and recovery is measured in days from backups. A payment outage is a supplier failure, your data is intact, and recovery is measured in minutes by routing around it. The overlap is the human layer: the named decider, the staff script, the customer communication. Build that layer once and both plans use it, which is why October is the sensible month to write the single page rather than two separate documents nobody reads.