Every cross-border commerce project we take on arrives with the same brief: traffic is fine, conversion is not. And almost every one of them has spent the previous six months optimising the product page.
The product page is where the traffic lands, so it gets the attention. But in a cross-border context the product page is rarely the constraint. The constraint is the moment a shopper who has already decided to buy discovers something about the transaction they were not expecting.
1. Show the real cost before the last screen
This is the one that matters more than the rest. If duty, tax and shipping appear for the first time on the final step, you have designed a trap. The shopper has invested effort, formed a price expectation, and is now being told the price was wrong. A meaningful share of them will leave, and some of those will not come back.
Moving landed cost forward — to the product page where possible, the basket at the latest — almost always reduces the conversion rate of the earlier step and increases completed orders. Teams find this counter-intuitive and resist it, because the earlier metric gets worse in the dashboard they look at daily.
A checkout that surprises people at the last step is not a conversion problem. It is a disclosure problem wearing a conversion problem’s clothes.
2. Offer the payment methods people actually use
Card-first checkout is a habit inherited from markets where cards dominate. Across much of Asia they do not. Wallets, bank transfer and instalment options vary significantly by market, and a shopper who does not see a familiar method reads the absence as a signal that the merchant is not really set up for them.
Put the local methods above the fold rather than behind a “more payment options” link. The link is a fine pattern for a market where cards are the default and a poor one where they are not.
3. Use the market’s address format
One address form serving nine countries will be wrong in at least seven of them. Field order differs, postcodes are optional or absent in some places, and building and floor conventions vary enough that a shopper forced into the wrong shape will either abandon or enter something that fails at fulfilment.
- Order the fields the way the market writes an address, not the way your database stores it
- Never require a postcode in a market that does not have one
- Validate on blur with a useful message, not on submit with a red block
- Let people paste an address and parse it rather than punishing them for it
4. Make the delivery promise honest
“Delivery in 5–12 working days” is not a promise, it is a hedge. Cross-border shoppers already suspect delivery will be slow; a vague range confirms the suspicion without giving them anything to plan around. A specific date, even a conservative one, converts better than an optimistic range.
5. Survive a bad connection
A significant share of cross-border checkout happens on mobile, on the move, on connections that drop. If your checkout loses everything when the connection goes, you are losing orders to the underground railway rather than to your competitors.
6. Say what happens if they send it back
Returns are the single largest unspoken objection in cross-border retail. A shopper buying from another country wants to know who pays the return shipping and how long the refund takes, and if you do not tell them they will assume the worst answer.
Put it in the checkout, not only in a policy page. One sentence is enough.
7. Ask for less
Every field is a chance to lose someone. Company name, title, date of birth, a second phone number, an account password — if fulfilment does not need it and the law does not require it, it should not be on the screen. Guest checkout should be the default path, not a link underneath the account form.
The order these should be tackled in
- Landed cost disclosure — almost always the largest single win
- Local payment methods for your top two markets by traffic
- Address formats for the same two markets
- Field reduction, which is cheap and low-risk
- Delivery date specificity
- Returns clarity inside the flow
- State persistence, which is the most engineering-heavy and the least visible
Most teams start at seven because it is the most interesting engineering problem, and reach one about four months later. If you have limited time, do them in the order above.
If you are working through this on a live storefront and want a second opinion, write to us — we will tell you which of the seven is your constraint, whether or not you engage us to fix it.