Translate a category page for espresso machines into German and you get Espressomaschinen; into Polish, ekspresy do espresso. Both are correct, and both miss the market. Germans mostly search Siebträgermaschine — 82,000 times a month against 14,000 — and Poles search ekspres kolbowy, 20,000 against 450. Translation changes the words. Localisation changes what the page is for, and only one of them shows up in search.
The difference, in one table
| Element | Translation | Localisation |
|---|---|---|
| Target query | The source keyword, translated | The term that market searches, researched there |
| Page type | Same template as the source | Whatever ranks for that query in that market |
| Prices, delivery, payment | Converted | What that market expects: currency, VAT display, carriers, payment methods, return rules |
| Trust | Source-market reviews and logos | Reviews in the language, a local contact, locally recognised certifications |
| Examples and formats | Kept | Local references, units, dates and number formats |
| Metadata | Translated title and description | Written for the local results page, with a local slug |
One product, four markets
On 27 September 2026 I looked up the espresso-machine query in the United States, Germany, Spain and Poland in Ahrefs: how many people search each variant, and what ranks for the main one.
| Market | Term people search | Monthly | Literal translation | Monthly |
|---|---|---|---|---|
| Germany | Siebträgermaschine | 82,000 | Espressomaschine | 14,000 |
| Poland | ekspres kolbowy | 20,000 | ekspres do espresso | 450 |
| Spain | cafetera espresso / express / expreso | 4,600 / 3,500 / 1,400 | máquina de café espresso | 200 |
Spain has no single term; the demand splits across three spellings of the same thing, and a page has to cover them without reading like a keyword list. Poland’s literal translation gets a forty-fourth of the demand of the term people use.
What ranks differs just as much:
- United States: shopping results take most of the page. The organic results are Amazon, a specialist retailer, Breville’s own shop and a CoffeeGeek review, with reviews from Business Insider and Food & Wine in the news slot.
- Germany: specialist shops, Amazon, De’Longhi’s category page, a dedicated test site, a YouTube video and the price comparison site idealo. Buyers compare before they buy, and the results page says so.
- Spain: individual product pages — Leroy Merlin at number one, then MediaMarkt, Amazon and PcComponentes — plus a YouTube video and a Reddit thread asking for recommendations.
- Poland: the marketplace Allegro twice, a price comparison site and two coffee-equipment shops.
A translated category page competes with a test site in Germany, with product pages in Spain and with a marketplace in Poland. Those are three different briefs. International keyword research covers how to read the local results page; the point here is what it changes about the page you publish.
What buyers expect
CSA Research surveyed 8,709 consumers in 29 countries in 2020. 76% preferred to buy products with information in their own language, and 40% said they never buy from websites in other languages. Two further figures matter for how you localise: 65% preferred content in their language even if it was poor quality, and 73% wanted product reviews in their language, if nothing else. Reviews are the part most sites leave in English.
What to localise, in order
The query
Research the term in the market before anyone translates. It decides the title, the heading and the slug, and it is the cheapest thing to get right.
The page type
If the market ranks comparisons, a bare category page will not compete, however good the German.
The commercial details
Currency and VAT display, delivery times and carriers, the payment methods that market uses — iDEAL in the Netherlands, BLIK in Poland — and the returns terms local law requires.
The trust signals
Reviews in the language, a local address or phone number where you have one, certifications that market recognises. Google itself lists local addresses, phone numbers and currency among the signals for which country a page is for.
Examples, units and formats
Litres not ounces, commas not points, the local version of a case study.
Metadata and structured data
A title written for the local results page, a local slug, alt text in the language, and prices in structured data in the local currency.
The mechanics of handing this to a translator are in how to brief a translator, and keeping every version complete over time is content parity.
When translation is enough
Not every page needs the full treatment. Help-centre articles, documentation and product specifications usually translate well, because the question is the same everywhere. Legal pages need legal review rather than localisation. Pages with no search demand in a market need to exist for users, not to rank. Save localisation budget for pages where the query, the results page or the purchase differs — usually categories, landing pages and anything that sells.
Machine translation is a reasonable first draft for all of it. Google warns against translating only the template while the content stays in one language, and treats mass-produced pages that add nothing as scaled content abuse; a reviewed machine draft is neither.
Sources
- CSA Research, Survey of 8,709 consumers in 29 countries finds that 76% prefer purchasing products with information in their own language, 7 July 2020, from the report Can’t Read, Won’t Buy – B2C.
- Google Search Central, Managing multi-regional and multilingual sites, last updated 10 December 2025.
- Ahrefs Keywords Explorer and SERP overview, 27 September 2026: search volumes in Germany, Spain and Poland, and top-10 results for “espresso machine”, “siebträgermaschine”, “cafetera espresso” and “ekspres kolbowy” in their markets.