International SEO: Why Translating Keywords Doesn't Work
There's a mistake we see repeated in almost every international expansion project we audit: a company decides to enter a new market, commissions a website translation, and assumes keywords translate the same way as content. This is incorrect. The consequences are severe.
At ZDS, we work with native teams in 14 languages and have managed the international SEO strategy for brands present in over 30 markets. We've learned that keyword translation isn't a linguistic problem; it's a market understanding problem. Solving it requires a completely different approach.
The Myth of Direct Translation
Let's take a real-world example. In Spain, the term "zapatillas de running" has significant search volume. The literal translation to German would be "Laufschuhe," which works. However, if you analyse the German market with keyword research tools, you'll discover that "Joggingschuhe" also has considerable volume and a different intent. And in Austria, the same product is searched for with variations that don't exist in Germany.
Another frequent case: a client in the tourism sector translated "hotel barato en Barcelona" to "cheap hotel in Barcelona" for the British market. But UK users search for "budget hotel Barcelona" or "affordable hotel Barcelona." Cheap has a negative connotation that barato doesn't have in Spanish.
These aren't minor nuances. They are the difference between appearing in the results your audience actually searches for, or spending months trying to rank for terms nobody uses.
Why Every Market Needs Its Own Keyword Research
Search behaviour is influenced by factors that go far beyond language:
- Search Culture: German users tend to use longer, more specific queries. Spanish users are often more direct. French users include more complete questions.
- Local Competition: Dominant terms in one market might be completely saturated by local competitors with unbeatable authority, while synonyms or variations offer real opportunities.
- Seasonality: Search cycles vary by market. "Winter coats" are searched for in October in Scandinavia and in December in Spain.
- Market Maturity: In markets where e-commerce is more developed, users search with stronger transactional intent. In emerging markets, informational searches predominate.
- Regional Variants: Mexican Spanish is not Argentinian Spanish. Brazilian Portuguese is not Portuguese Portuguese. US English is not UK English. Each variant has its own search vocabulary.
Hreflang: The Implementation Almost Nobody Gets Right
The hreflang attribute tells Google which version of a page to show based on the user's language and region. It's the fundamental technical piece of international SEO. It's the technical element with the highest error rate we find in audits.
The most frequent errors:
- Lack of Reciprocity: If the Spanish page points to the English version with hreflang, the English version must point back to the Spanish. Non-reciprocal hreflang tags are ignored by Google.
- Incorrect Language Codes: Using "en-UK" instead of "en-GB", or "es-LA" (which doesn't exist) instead of specifying each Latin American country.
- Absence of x-default: Not including hreflang="x-default" for users whose language or region doesn't match any specific version.
- Incorrect URLs: Pointing to URLs that redirect, return 404, or are canonicalised to another page.
- Partial Implementation: Including hreflang only on some pages or only for some languages, instead of covering all combinations.
- Conflict with Canonical: A page's canonical tag should point to itself, not to the version in another language. We have seen sites where the canonical of the French version pointed to the English version, completely nullifying the hreflang.
Implementation Methods (from best to worst):
- HTTP Headers: Ideal for PDFs and non-HTML files.
- XML Sitemap: Scalable and easier to maintain for large sites. Our preferred option for e-commerce with thousands of products.
- Tags in the <head>: The most common. Works well but can slow down loading if there are many languages.
Domain Structure: ccTLD vs. Subdirectory vs. Subdomain
This is one of the most important and difficult-to-reverse decisions in international SEO:
- ccTLD (example.de, example.fr): Clear signal of geo-targeting. But each domain starts from scratch in authority. Requires building Backlinks for each domain separately. Recommended only if you have significant resources for each market.
- Subdirectories (example.com/de/, example.com/fr/): All domain authority is consolidated. Easier to manage technically. This is our default recommendation for most projects.
- Subdomains (de.example.com, fr.example.com): Google treats them as semi-independent sites. Harder to manage than subdirectories without the geo-targeting advantages of a ccTLD. Rarely the best option.
Most of our clients with international presence use subdirectories with hreflang. It's the optimal balance between authority consolidation, ease of management, and language/region signalling.
Content Localisation vs. Literal Translation
Translation is just the first step. Localisation adapts content to the cultural, legal, and commercial context of each market. Here are the key differences:
- Currency and Formats: Not just converting prices, but displaying the local currency. Date, time, and number formats vary between markets (€1,000.50 vs. $1,000.50).
- Cultural References: Examples, success stories, and references must be relevant to the local market. A success story from a Spanish company doesn't have the same impact in the German market.
- Legal Framework: Terms of Service, Privacy Policy, shipping, and returns information must be adapted to local legislation.
- Tone and Formality: The tú vs. usted in Spanish, the du vs. Sie in German, the expected level of formality in Japanese commercial communication. These nuances directly affect the Conversion Rate.
- Local Search Intent: The same query can have different intents depending on the market. "Seguros de coche" (car insurance) in Spain has clear transactional intent; in some Latin American markets, the intent might be more informational.
The Process We Apply at ZDS for International SEO
After years of trial and error, we have developed a process that consistently works:
- Phase 1 — Market Analysis: Before touching a single Keyword, we analyse the target market's sector. Who dominates? What type of content ranks? What is the digital maturity level of that sector in that market?
- Phase 2 — Native Keyword Research: A native specialist from the target market conducts Keyword Research from scratch. We don't start from translations, but from how the local user actually searches.
- Phase 3 — Content Mapping: We cross-reference native keywords with existing content to identify which pages can be adapted, which need new content, and which have no equivalent in the target market.
- Phase 4 — Localisation: Content is created or adapted by native writers who understand both SEO and the local market.
- Phase 5 — Technical Implementation: Hreflang configuration, URL structure, Search Console per market, segmented sitemaps.
Real Errors We've Corrected
A fashion e-commerce with presence in 8 European markets had automatically translated its entire website into German, French, Italian, and Portuguese. Following our audit, we discovered that 40% of the target keywords in German did not match the actual search terms in the market. After redoing the Keyword Research with native speakers and re-localising the content, organic traffic in Germany grew by 180% in 6 months.
Another case: a B2B tech company used subdomains for each country with poorly implemented hreflang. Google indexed the incorrect versions for each market. Migrating to subdirectories with correct hreflang and localised content tripled organic traffic in secondary markets.
What AI Changes (and What It Doesn't) in International SEO
AI translation tools have vastly improved. They can produce fluid and natural translations. But they still don't solve the fundamental problem: they don't know how people search in each market. A perfect translation of a term nobody searches for is still useless for SEO.
AI adds value in scale: it can accelerate content localisation once native Keyword Research is done. But market analysis, term selection, and strategy still require human knowledge of the local market.
Are you expanding your business into new markets? At ZDS, we manage international SEO strategies with native teams in 14 languages. We don't translate: we localise. Let's discuss your international project.
