International SEO Keyword Research: A Market-by-Market Process

International SEO keyword research is the step most expansion projects rush. A team has a keyword list that works at home, runs it through a translation tool or hands it to a translator, and builds pages for five new markets. Six months later the new pages have impressions for terms nobody searches and almost no clicks. The problem is rarely the site; it is the research.

Below is the market-by-market process I use, how intent shifts between countries, how to pick a site structure, and the hreflang mistakes that undo good work.

Why translating keywords fails

A translated keyword is a guess about what someone in another country types. Often it produces a phrase that is grammatically correct and commercially useless, for three reasons:

• People describe the same need with different words. The literal translation of your product category may be a formal or technical term, while buyers search a colloquial one, a loanword, or a brand name used generically.
• The same words carry different intent. A term that signals buying intent in one market can signal research, jobs or DIY in another, and Google ranks different page types as a result.
• The competitive set changes. In your home market you might compete with specialist sites. Elsewhere the first page might be marketplaces, government portals or local directories, which changes what kind of page can realistically rank.

Translation is useful for generating seed terms, not as a substitute for looking at the market. The better frame is transcreation: adapting the message, examples and vocabulary so the page reads as if it were written locally, then doing keyword research in that market to confirm the words.

English is not one market: lessons from the US, UK and ANZ

When I was an SEO Team Leader on international campaigns, my teams ran campaigns for clients in North America, the UK and Australia/New Zealand. On paper, that was one language. In practice, it was at least three keyword universes.

Spelling is the obvious difference (optimise/optimize, colour/color, centre/center), and it matters for exact phrasing in titles and headings. Terminology matters more, because it changes which concept a word points to. Intent is the subtle one, and it is where most multi-country English sites lose.

Shared words can still rank completely different pages. A search for "holiday" in the US surfaces public holiday calendars and seasonal content; in the UK and Australia it is a travel query. A UK page optimised for "pension" will never be the right answer for an American searching about a 401(k), even though both are researching retirement savings.

The practical lesson: research each English market separately, and decide deliberately which differences justify a separate page and which can be handled with one well-written version.

A market-by-market international SEO keyword research process

1. Start with business reality, not search volume. Confirm where you can actually sell, deliver, invoice and support. A high-volume market you cannot serve properly is a distraction.
2. Build a concept list, not a keyword list. From your home-market research, list the underlying topics and problems: "compare business phone plans", "hire a commercial cleaner". Concepts travel across markets; keywords do not.
3. Generate local seed terms. Use native speakers or in-market staff, local competitor sites, local marketplaces and forums. Ask sales and support teams what customers call things in that market. AI tools can speed this up (see below), but a person in the market decides what stays.
4. Research in the local database. In Semrush or Ahrefs, switch to the country database and expand each seed. Check Google Search Console filtered by country if you already have traffic there. Local autocomplete and "People also ask" boxes, viewed with the right country and language settings, fill gaps tools miss.
5. Check the local results with your own eyes. Tools report volumes; they don't show you the page. For each priority term, look at the results page in that market with the location set to the country (and language, where it matters). Note who ranks and which format wins: product pages, guides, comparison sites, marketplaces, government pages, videos or local packs. If the page type you plan to build is absent, either pick a different term or build a different page.
6. List the local competitors you didn't know about. Your home-market competitors are often not the ones ranking abroad. Write down the local brands, marketplaces and directories that keep appearing, and look at what their pages cover that yours don't. They show you the standard you need to meet in that market.
7. Cluster and map to URLs per market. Group terms by intent and assign one target URL per cluster per market. This is where you see whether a market needs its own page or can share one.
8. Transcreate and validate. Brief local writers with the cluster, the intent and the SERP findings. Have someone in-market review the final page for terminology and tone before it goes live.
9. Measure by market. Track impressions, clicks and conversions per country in Search Console and GA4, and track rankings for each market separately (more on that below). If a market's pages collect impressions for terms unrelated to your offer, the research is off and needs revisiting.

Tools for international keyword research

No single tool covers every market well, so I combine a few:

• Semrush or Ahrefs country databases for volumes, difficulty and competitor keywords in each market. Always switch the database to the target country before researching.
• Google Keyword Planner with location and language set to the market, useful for relative demand and for markets where third-party databases are thin.
• Google Search Console filtered by country, if you already get impressions there. It shows the terms Google already associates with you in that market.
• Google Trends by country to compare competing local terms and spot seasonality that differs from your home market.
• The live results page with the right location and language, plus local autocomplete and "People also ask", to see real phrasing and the page types that win.
• Local marketplaces, forums and competitor sites for the vocabulary customers actually use, which tools often miss.

Using AI to brainstorm local phrasings

AI tools are good at generating local phrasings fast: synonyms, informal terms, question forms and the way a need might be described in another market or language. I use them to widen the seed list in step 3. They are not good at knowing which of those phrasings real buyers use, and they can sound fluent while being wrong. Treat the output as a draft list. A native speaker or in-market colleague must sign the terms off before they drive pages, titles or briefs, and every term still gets checked in the local database and on the local results page.

Track rankings per country

A single rank-tracking project for "English" or for your home market tells you nothing about the others. Track each market separately, with its own location and language settings, so a gain in one country can't hide a loss in another. I use Semrush Position Tracking for this, with a separate setup per market and the keyword set from that market's research, not a translated copy of the home list. Review it alongside Search Console filtered by the same country.

Check AI answers in each market too

AI answers are covered as standard in my SEO work, and they vary by market. An assistant can describe your brand well in one language and poorly, or not at all, in another, because it draws on different sources per market. Add local-language prompts to your AI visibility checks for every market you target, including the regional or mixed languages your buyers actually use, and run them logged out and set to that market's location. My guide to measuring AI search visibility covers the method.

Choosing a structure: ccTLD vs subfolder vs subdomain

Keyword research tells you how different each market's content needs to be; that, plus your resources, drives the structure decision. Google supports all three options.

For most businesses I advise a subfolder structure. It keeps one domain building authority, one codebase to maintain and one analytics setup. ccTLDs make sense when each country runs as a largely independent business with its own team, content and link building. Subdomains usually exist because of technical constraints, such as a regional store on a different platform, rather than as an SEO choice.

If you are moving an existing site into a new international structure, treat it as a migration and plan redirects properly. My website migration SEO checklist covers that process.

Hreflang pitfalls that quietly break international sites

Hreflang tells Google which language or regional version of a page to show to which users. It is a signal, not a directive, and it only works when every page in a set is annotated consistently. Google's localised versions documentation is the reference; these are the errors I find most often in audits:

• Missing return links. If the UK page points to the US page, the US page must point back. One-way annotations are ignored.
• No self-reference. Each page should list itself in its own hreflang set.
• Invalid codes. Language codes use ISO 639-1 and optional regions use ISO 3166-1 alpha-2. The UK is en-GB, not en-UK. A region on its own, such as GB, is not valid.
• Pointing at the wrong URL. Hreflang must reference the final, indexable, canonical URL, not one that redirects, returns an error or carries a noindex.
• Conflicting canonicals. If every regional page canonicalises to the US version, you have told Google they are duplicates and the regional versions will not show.
• No x-default. Without a fallback, users in unlisted countries may land on an arbitrary version. Use x-default for a global or language-selector page.
• Mixed implementation. Hreflang can live in the HTML head, HTTP headers or XML sitemaps. Pick one method per page set and keep it consistent; sitemaps are usually easiest to maintain at scale.
• Launching every language at once. Roll new languages out in stages, starting with the market where demand is clearest. Validate that version's hreflang URLs before adding the next: no double slashes, no redirects, and every URL returning 200. Errors found in one language are far easier to fix than the same errors copied across ten. My free hreflang checker checks codes, self-references and return links for any page.
• A root URL that sends visitors away. If the root URL must send visitors to a country version, annotate the root as x-default and make every version reachable through normal links and hreflang. Never send crawlers somewhere different from users; treat Googlebot exactly as you would a visitor from the same location.

When I plan a multilingual rollout, it runs as part of an international or migration engagement, so the staging, validation and redirects are handled together rather than one language at a time by different teams.

Do you need hreflang for 60 English-speaking countries?

Usually not. Creating 60 near-identical English pages multiplies maintenance and risks Google treating them as duplicates.

Ask what genuinely differs between countries. If price, currency, delivery terms, legal content or product range change, a regional version is justified. If nothing changes except the flag, it is not.

A common pattern that works well:

• One generic English version, annotated en (language only), plus x-default.
• Separate regional versions only for the markets with material differences, for example en-GB, en-AU and en-US.
• Transcreated copy on those regional pages, based on the market research above, so they are genuinely different pages, not spelling swaps.

Google will serve the generic en page to English speakers in the countries you have not listed. You can add regions later when a market earns its own version. If you need help deciding, this is a standard part of my SEO consulting work.

Can international SEO hurt your local rankings?

Done well, no. Adding international versions does not take authority away from your home market pages. Done badly, it can cause real damage. The patterns I see behind lost home-market rankings are:

• Near-duplicate regional pages without hreflang, so Google picks one as canonical, sometimes the wrong one, and your home page drops out.
• Automatic redirects by IP address. Googlebot mostly crawls from the US, so IP-based redirects can stop it seeing other versions at all. Offer a banner or selector instead of forcing a redirect.
• Home-market internal links pointing at regional pages, which confuses signals and sends home users to the wrong version.
• A structural change without a migration plan, such as moving the existing site into a /us/ folder without proper redirects.

The safest approach is to leave your strongest market's URLs untouched where possible and build new markets alongside them. If you are weighing expansion and want the research, structure and hreflang handled together, that is the core of my international and enterprise SEO consulting. You can also see how I have structured a large multi-market operation in the international SEO operations case study.

When international expansion is not the right move

Honest limit: international SEO cannot fix a business that cannot serve the market. Without local pricing, overlapping support hours or a way to handle returns and invoicing, rankings will not turn into revenue. It is also worth fixing the technical foundation of your main site first; a quick pass through an SEO audit checklist often reveals issues that would otherwise be copied into every new market.

Nobody can guarantee rankings in a new country. What good research does guarantee is that you are competing for the terms your buyers actually use, with the page type Google actually rewards, in the markets you can actually serve.