Choosing Payment Gateways for Cross-Border Sales
Settlement currency, decline rates and local payment methods matter more than published fees. What to test before launch, and why a backup gateway is not optional.
Payment gateways are chosen for the wrong reasons. Usually: whatever is easiest to activate, or whichever one a developer already integrated. Six months later the store is losing 4% of orders to declines it cannot explain, and nobody knows which gateway caused it.
This is the decision framework, ordered by what actually breaks.
Start with settlement, not checkout
The first question is not “does it accept Visa”. It is: in what currency and to what entity does the money arrive?
Three arrangements exist, and they have different consequences:
| Arrangement | What it means | Cost |
|---|---|---|
| Local entity, local acquirer | You settle in the customer’s currency, in the customer’s country | Setup effort, best rates |
| Cross-border merchant account | One account, multiple currencies, accepted as local in some markets | Higher fees, less friction |
| Single-currency processor | Everything converts to your base currency | FX spread, and often a decline-rate penalty |
The third row is the default for most small exporters, and it is where “the price was in euros but I was charged in dollars” complaints come from. Fixing that is described in the multi-currency checklist.
Decline rate is the real cost
Published fees are 1.4% to 2.9%. The cost that decides profitability is decline rate, and it is invisible until you measure it.
A gateway unfamiliar with your market declines more aggressively. If 8% of European cards are declined instead of 2%, a 2% fee advantage is irrelevant.
What raises decline rates on cross-border stores:
- No local acquiring in the customer’s country
- Address verification mismatch when your checkout accepts formats the gateway does not expect
- 3-D Secure implemented as a hard block rather than a challenge
- Currency mismatch between the displayed price and the authorisation
- First transactions from a new merchant — declines settle down after volume, which matters if you launch in December
Track decline rate per gateway, per country, per month. Without that number, gateway decisions are made on feelings.
The method mix is market-specific
Cards are not universal. Getting this wrong means a checkout that looks complete and converts badly.
- Germany — cards are a minority. Bank transfer, SEPA direct debit, and local wallets matter
- Netherlands — iDEAL dominates
- Poland — BLIK and local transfer
- Brazil — instalments (boleto and card) are expected, not optional
- Gulf markets — cards plus cash on delivery for lower-value orders
The practical approach: launch with cards plus a wallet, then add local methods for the two or three markets that produce revenue. Adding every method to every market confuses the checkout and adds failure modes.
Where the money actually goes wrong
Refunds. A refund through a different path than the original payment creates reconciliation problems and, occasionally, double refunds. Confirm the refund flow per gateway before launch, and test it.
Partial refunds and fees. Most gateways do not return their processing fee on a refund. On stores with high return rates — apparel, footwear — this is a real margin line nobody models.
Chargebacks. Cross-border chargebacks are expensive and the evidence requirements differ per scheme. A store without a documented dispute process loses these by default.
Payout timing. Weekly versus daily payouts matters for cash flow if you pay suppliers upfront. A gateway with better fees and 14-day rolling payouts may be worse for your business than one with worse fees and 2-day payouts.
The configuration that survives
- Decide settlement currency per market before choosing a provider
- One primary gateway with local acquiring where revenue is concentrated
- A backup gateway for when the primary has an outage — this is not optional at volume
- Local methods for the top three markets only
- 3-D Secure as a challenge, not a block
- Decline rate dashboard, reviewed monthly
- Refund and partial refund tested end to end
Item 3 is the one stores skip. A gateway outage during a promotion costs more than a year of fees, and switching mid-outage is impossible without a second integration already in place.
What to test before launch
- A successful payment in every enabled currency
- A declined card, and that the customer sees an actionable message
- 3-D Secure completion, including the failure path
- A full refund and a partial refund
- An order placed with a billing address in a different country from the shipping address
- The webhook that marks the order as paid — confirm it fires, and confirm what happens when it does not
The last one causes the worst launch-day problem: money captured, order still pending, and no one notices until the customer emails. If your store treats “payment received” as a webhook event, that webhook needs monitoring, not just a test.
A judgement worth stating
Most exporters do not need four gateways. They need one that works well in their main market, one backup, and honest currency handling. Complexity in payments does not increase conversion; it increases the number of places an order can fail.
If you are setting up a store for multiple markets, the gateway and currency decisions belong together — they are the same decision viewed twice. That work sits under WooCommerce development, and it is worth doing before writing any checkout code.