Selling abroad with Shopify: translating the catalogue is not enough

When a store decides to sell abroad, the first thing it does is translate the catalogue. It is the obvious move, it can be bought as an app in half a day, and on its own it returns very little. Not because the translation is wrong, but because the problem was never the language.

What a translation does not translate

A translated page says the same things in German. The point is that in Germany buyers are not asking the same things.

Standards. Plugs, voltages, fittings, sizes, certifications. These are the details that decide whether a product is buyable in that country, and a page written for Italy often does not mention them at all: at home they are obvious, so nobody ever wrote them down.

The words people search with. Not the correct word, the used one. Every market has its own terms for the same category, and they are rarely the literal translation of ours. A translator gives you the right word. Being found needs the common one, which is sometimes technical and sometimes slang.

Service expectations. Who installs it, how long the warranty runs, who pays for the return, what happens if it breaks in August. These change by country and they are the part that blocks the purchase on considered products. A page that does not address them leaves the doubt standing, in any language.

One feed per country, not one feed for everywhere

The second move almost everyone skips is the feed. Prices, availability, shipping rules and costs change by market: a single feed for all of Europe forces a compromise that suits none of the countries in it.

The result is bad in both directions. Where the compromise is too high you lose competitiveness; where it is too low you sell at a loss, or show terms that checkout then fails to honour. And a mismatch between what the feed says and what the page says is one of the most common reasons products quietly vanish from ads with nobody understanding why.

Why price alone does not open a market

There is a belief that you enter a foreign market on price. On impulse products it sometimes works. On everything else, a buyer compares and then asks who you are: a price below a local competitor, on a page that does not answer that market's questions, produces visits and not orders. You end up paying for traffic you cannot convert.

What changed for a heating and cooling retailer

For an Italian heating and cooling retailer we rewrote the catalogue in native German and French, written for each market rather than translated, with one Merchant Center feed per country instead of a single shared one, plus technical content and trust signals on the products people buy after thinking about them.

The share of revenue generated outside Italy went from 4% to 52%. Over May to August 2026, against the same months a year earlier, Google cost per click fell 29.4% while click-through rate rose 20.2%, and on Meta cost per click more than halved (-50.7%). The full case, with the month-by-month chart.

The order worth doing it in

One market at a time, not all at once. The reason is practical: open four countries in a month and you will never know which part of the work paid off, and you will end up with four mediocre catalogues instead of one good one.

The order that makes sense is this. First choose the products that actually have a market in that country, which are not automatically your Italian bestsellers. Then rewrite those pages natively, with local standards. Then split the feed. And only then turn on the ad spend, because before that you are buying traffic for pages that are not ready to receive it.

Frequently asked questions

Do I need a separate domain per country?

Not necessarily, and for most stores it is not worth it. Shopify handles markets with subfolders or subdomains, and what matters for ranking is that each version correctly declares who it is for. A domain per country multiplies both the technical work and the authority you have to build from scratch.

Translate the whole catalogue, or a few products properly?

A few properly, if those are the only two options. But it is a false choice once rewriting a page costs close to nothing: then the question becomes which products have a market in that country, not how many you can afford.

How much does language matter in AI answers?

A lot, because an answer engine cites sources in the language of the question. A page that exists only in Italian will not be quoted in a German conversation, even if the product is the best fit. The same rule as search, only stricter.

Is machine translation enough as a starting point?

As a first draft, yes, for prose. Not for standards and units, which need checking, and not for search terms, which are not translated but chosen by looking at how people search in that country.

How many markets should I start with?

One. If it works, the second costs less than the first because the method is already proven, and you will have a real number to compare instead of an impression.