The standard approach to international keyword research is to take the English list, send it to a translation agency, and publish. It produces pages that are linguistically correct and commercially useless, because the translation of your keyword is frequently not the keyword.
What translation misses
The market uses the English term. Across much of Northern Europe, technical and business audiences search in English even when browsing in their own language. Germans search “cloud hosting” far more than the German equivalent. Translating it produces a page targeting a term nobody uses.
The local term is a different word entirely. Not a translation — a different concept that occupies the same commercial slot. The literal translation exists, is grammatically correct, and has no search volume.
Volume distribution is completely different. A term that is head in the UK can be long-tail in France and non-existent in Poland, because the product category has different maturity in each market.
The SERP is a different competitive set. Local competitors you have never heard of, different SERP features, and in some markets Google is not the only engine that matters.
Regulatory and cultural framing changes the query. Financial, health and legal products are searched through the local regulatory vocabulary. Buyers search the compliance term, not the marketing term.
Research each market from scratch
Treat every market as a new project, not a derived one.
- Get native seeds, not translated ones. Ask a native speaker in the market — ideally someone in sales — how they would search for the product. Do not give them the English list first; that anchors them to your framing.
- Run volume per country, not globally. Global volume hides the distribution that determines whether a market is worth entering.
- Read the actual SERP in-market. Set the country parameter and look at who ranks. This is the only reliable way to detect that a market searches in English, or that the SERP is owned by a marketplace.
- Check the term is not a false friend. Words that translate cleanly but carry different commercial connotation. This is a native-speaker check; no tool catches it.
- Validate against your own data. Search Console segmented by country will already show which markets send impressions and on what queries — usually the fastest evidence you have.
Step 5 first, always. Existing impression data from a market is worth more than any tool estimate, because it is your site being served to real searchers in that market.
English-speaking markets are not one market
The most common and most avoidable failure. UK, US, Australian and Irish English differ enough that one page cannot serve all of them well.
| Concept | UK | US |
|---|---|---|
| Recruitment | CV, recruitment agency | resume, staffing agency |
| Property | flat, estate agent, letting | apartment, realtor, rental |
| Finance | current account, VAT | checking account, sales tax |
| Transport | lorry, motorway, petrol | truck, highway, gas |
Spelling differences compound it — optimisation and optimization, analyse and analyze. Google understands these are related, but the page that uses the local spelling in its title consistently outperforms the one that does not.
This matters commercially more than it looks. If your best-performing market by position is the UK, publishing US-spelled content is quietly costing you the market where you are already strongest.
The checks that stop hreflang undoing everything
Good per-market research destroyed by bad implementation is the standard outcome. Four things to verify:
- Return tags are reciprocal. If the English page points to the German page, the German page must point back. Non-reciprocal tags are ignored entirely.
- Every page includes a self-referencing tag. Frequently omitted; invalidates the set.
- Codes are language-first, region-second, and valid.
en-GBnoten-UK.UKis not a valid region code and the tag fails silently. - An
x-defaultexists. For searchers in markets you do not target explicitly.
Silent failure is what makes this expensive. There is no error message — the tags are simply disregarded, Google picks a page itself, and you find out months later when the wrong market’s page ranks everywhere.
Do not build a market you cannot serve
The uncomfortable check that belongs in the plan: can you actually serve a customer from this market? Local-language support, local payment methods, local invoicing and VAT handling, delivery.
Ranking in a market you cannot transact in produces traffic, bounce and support load. It is worth deciding this before the research rather than after the content, because the research itself is cheap and the content is not.
Localisation is content work, not keyword work
Even with correct keywords, a translated page frequently underperforms a written one, because examples, currencies, regulatory references and proof points are all market-specific. A German page citing UK case studies and pound prices reads as an afterthought whatever its keyword targeting.
Budget for native writing on your priority markets and accept translation only on secondary ones. That trade-off is a real constraint and worth stating in the plan rather than discovering at review.
Related work
The core method still applies — keyword research guide for the process, and the keyword research checklist for verification, with one change: run the domain-strength checks per market, since your authority is not evenly distributed across country SERPs.
If a market’s SERP is owned by a marketplace, or the plan involves consolidating several country sites, that is a migration problem as much as a keyword one.
For European market entry specifically, our nearshore SEO team works in-market with native speakers and CET overlap, which is the practical answer to steps 1 and 4 above. If the international work is a replatform onto a new domain or URL structure, the site migration service covers the hreflang implementation alongside it.
