International SEO Keyword Research
Last updated
International SEO keyword research is the process of finding the right queries for each target market from scratch, rather than translating your home-market keyword list. Search volume, query phrasing, and user intent vary across languages and countries, often dramatically. Approaching each market with fresh research produces significantly better results than assuming existing keywords will transfer.
Why does translation not work for keyword research?
A straightforward example of why direct translation fails: the word “jumper” in British English translates to “sweater” in American English. Both refer to the same garment, but the search behaviour differs. If a UK retailer targets sweaters on their US pages by translating their jumpers keyword strategy, they may use the correct word but still miss regional phrasing, brand associations, or intent signals specific to that market.
The problem scales across every category:
- Medical terms vary between markets (“chemist” in the UK is “pharmacy” in the US)
- Legal terminology differs across jurisdictions (“solicitor” vs “attorney”)
- Financial products have different names and regulatory contexts
- Technology product names and dominant brands vary by region
- Local brand dominance shapes the queries people use when searching for a product category
How does search volume vary by market?
The same keyword can have radically different search volumes in different countries. This reflects population size, internet penetration, local search engine market share (Baidu in China, Yandex in Russia, Naver in South Korea), and whether the topic is culturally relevant in that market.
Do not assume that because a keyword has high search volume in English, its direct translation has proportional volume in French or German. Check actual volume for each language in each country. Ahrefs, Semrush, and Google Keyword Planner all support country-level and language-level filtering.
Treat the numbers themselves with more caution than you would at home. Tools model volume from clickstream panels and Keyword Planner data, and both thin out in smaller markets and less widely spoken languages, so estimates there carry wider error bars than the same tools’ figures for the UK or US. Google Trends is a useful cross-check: it gives no absolute volumes, but it compares relative interest between two phrasings within a single country, which is usually the question that actually needs answering, namely whether people in this market search for X or for Y.
How do you use research tools in unfamiliar languages?
Most keyword research tools support non-English research, but using them effectively requires preparation:
Start with Search Console where you already have impressions. For any market your existing pages already surface in, however faintly, Search Console’s country and query dimensions are the cheapest source of genuine local demand: real queries from real users in that market, rather than modelled volume. Filter by country, then read the queries. It cannot size a market you have never entered, but where it applies it beats any estimate.
Set the correct country and language filters. Running a keyword search for “assurance voiture” (car insurance in French) with the country set to the UK returns UK-based French searches, not French market data. Always match the country setting to the target market.
Use Google Search in the target country. Google’s autocomplete and People Also Ask surfaces are localised. Searching via a VPN set to the target country gives you the query suggestions Google shows to local users. This is useful for identifying natural language patterns.
Check local SERPs. The SERP for a query in Germany may look very different from the equivalent in France: different content types, different local brands, and different SERP features indicate different user intent and content format expectations.
Localising vs translating keywords
Some keywords translate cleanly. “How to bake bread” maps to the French “comment faire du pain”, and the intent is the same: instructional content about baking. Note that even here the natural phrasing is not the literal one. A word-for-word rendering (“comment cuire du pain”) is grammatical but is not what French speakers type, which is why the check has to be done by someone who searches in the language rather than by a translation tool.
Other keywords require localisation rather than translation:
- Brand-specific queries: if a particular brand dominates the German market for your product category, queries about that brand need independent research, not translation from English brand queries.
- Local compliance queries: “IR35” in the UK has no equivalent in France; French users search for “statut d’auto-entrepreneur” or “portage salarial”. These are not translations of the same query; they are different queries reflecting different regulatory environments.
- Colloquial phrasing: even within English, “estate agent” (UK) and “realtor” (US) refer to the same profession. Keyword research must use the locally natural phrasing.
Competitive conditions per market
Each market has its own established players, domain authority distribution, and content quality benchmarks. The difficulty of ranking for a keyword in France may be very different from ranking for the equivalent in Germany, even if search volumes are similar.
Before entering a new market, examine:
- Who currently ranks in positions one to ten for your target queries?
- Are they local businesses, global brands, or aggregators?
- What content format and depth does the top-ranking page use?
- Are there underserved queries (decent volume, weaker competition) where you could gain an early foothold?
This analysis must be done in the target language and country, not inferred from your home-market competitive assessment.
How does AI search change international keyword research?
Google’s AI surfaces are no longer English-first. AI Overviews reached more than 200 countries and territories and over 40 languages in May 2025,1 and AI Mode is now available in over 200 countries and territories.2 For most target markets the question is no longer whether an AI layer sits over the results page, but what it looks like in that language.
Three consequences for keyword research:
Query phrasing shifts, and it shifts differently in each language. Prompts to an AI surface run longer and more conversational than typed keywords. How that plays out is language-specific: the natural spoken form of a question, its conventions of address, and its typical length are not constant across markets. The conversational variants worth targeting in German are not translations of the ones worth targeting in English, for the same reason the keywords never were.
Read the AI answer in-market, not only the ranked results. The SERP check in the workflow below should include what the AI Overview says for that query in that country, and which sources it draws on. A market whose AI answers cite established local publishers calls for a different authority strategy from one whose answers cite international brands.
Google is not the whole picture in every market. Where Baidu, Yandex, or Naver hold significant share, the AI features those engines ship matter for the same reason the engines themselves do. Research the surfaces the market actually uses rather than assuming the Google-shaped one.
International keyword research workflow
- Identify the topic clusters you want to target in the new market.
- Generate seed keywords in the target language. Use native speakers for this where possible rather than machine translation.
- Run each seed through your keyword tool with country and language set to the target market.
- Examine SERP results for each priority keyword in the target language.
- Identify intent signals from top-ranking pages: are they guides, product pages, comparison tools, local directories?
- Map keywords to existing pages (if they can be localised) or new pages (if no equivalent exists).
- Prioritise by volume, difficulty, and business relevance.
The output of this workflow is a prioritised keyword list per market. Applying it to content production requires a multilingual content strategy: deciding which topics to create, for which market, and in what sequence.
Frequently asked questions
Can I do keyword research in a language I do not speak?
With difficulty. Tools can surface search volumes, but interpreting results, identifying natural phrasing, and evaluating SERP quality requires at least some language understanding. For markets where you have no language capability, working with a native-speaker SEO consultant or a specialist translation agency with SEO expertise is worthwhile.
How often should international keyword research be repeated?
At least annually, and whenever entering a new market. Competitive landscapes shift, new local brands emerge, and query volumes change with cultural events and regulatory changes. International markets can shift faster than home markets because they receive less monitoring.
Does search intent differ across markets?
Yes, sometimes significantly. The dominant intent behind a query may be informational in one market and transactional in another, depending on local buying behaviour and the maturity of the online market for a product category. SERP analysis is the most reliable way to identify intent differences: look at the content types that dominate positions one to three.
