Translation is the cheapest part of going cross-border and the most overrated. The expensive parts are payment habits, delivery and returns, and whether a stranger in another country has any reason to trust you.
Entering a second market usually starts with a quote for translation, which is the cheapest item on the list and the one least likely to decide the outcome. People buy in a language they half understand every day. What they do not do is pay with a method they have never used, or order from a company with no visible way to return anything.
Here is what actually changes.
Payment habits are national and they are strong. In one market cards dominate; in the next it is bank transfers, or a local wallet, or invoice-after-delivery, or an instalment option that is not a credit product so much as a cultural default.
This is the single largest and least visible loss in cross-border selling, because it never appears in your analytics. There is no event for “I could not see my way to pay”. The customer simply leaves, and the drop looks like ordinary indecision. Find out what the majority in that market uses before you translate a single page.
Three separate decisions that people run together. Which currency you show. Whether prices include tax — a convention that differs by market and by whether you sell to consumers or businesses, and getting it backwards makes you look either expensive or dishonest. And rounding: a price converted precisely from your home currency reads as a conversion, not as a price, and that alone signals “foreign shop” more clearly than any language.
The registration and reporting side is a question for an accountant in that country rather than for your developers. What we will say is that it is cheaper to ask before launching than to reconstruct a year of invoices afterwards.
Delivery time and cost must be visible before checkout, in local terms. “5–7 days” means something different when a border is involved, and a customer who discovers duties on the doorstep does not become a repeat customer.
Returns are the part people postpone. If your return address is in another country, the cost and hassle of sending something back is a reason not to buy in the first place — and in consumer markets, return rights are frequently not optional. Decide who pays for the return leg before you advertise, because it is a pricing decision disguised as an operations one.
At home you have a reputation. In a new market you are a domain nobody recognises. What closes that gap is unglamorous: a local phone number or a support address that answers in a sensible timezone, a company identity that is findable, a returns policy in plain language, delivery promises with dates, and reviews from that market rather than translated ones from yours.
Local reviews take time to accumulate, which is an argument for entering one market properly rather than five thinly.
Keywords are not the same words in another language; they are the words that market actually types, which are often shorter, sometimes English, and frequently not what a translator would choose. Your competitors are different there, the comparison sites are different, and the questions people ask before buying are different.
Technically, the requirement is boring and easy to get wrong: one address per language, correct language annotations between them, and no automatic redirect that traps a visitor in a version they did not choose. Nothing exotic — just done once, properly.
Support in a language you do not speak is a commitment, not a feature. Before launching, decide how questions in that language get answered, how fast, and by whom — and if the honest answer is “in English, within a day”, say so on the site. Stating a limit costs less than discovering it in public.
Four practical faults show up in nearly every cross-border launch, and all four are cheap to prevent and irritating to find later.
You do not need the full apparatus to find out whether anybody wants your product there. Translate the pages that carry the argument, enable the one payment method that market expects, quote delivery honestly, and spend a small amount on traffic for a month.
What you are measuring is not revenue — the volumes will be too small to be meaningful. It is whether people reach the checkout and stop, which tells you the offer works and the machinery does not, or whether they never get that far, which tells you something earlier is wrong. Either answer is worth more than a full launch built on an assumption.
Take it in this order: payment methods first, because it is the largest invisible loss and usually the fastest fix; then costs and delivery shown early and in local terms; then trust signals; then translation quality; then search. That is roughly the reverse of the order most companies work in, and it is ordered by money rather than by what feels like progress.
The same rule from the funnel work applies: buy from your own shop the way a customer in that country would, on their phone, paying the way they pay. Almost nobody does, and it takes twenty minutes.