Quick answer
Translating a keyword list gives you words, not demand. International keyword research means finding the terms people in each market actually type, in their spelling, script and search engine. Check local variation, research each market separately, validate with local search data, and build a keyword map per market before writing localised pages.
A CSA Research survey (2020) of 8,709 consumers in 29 countries found that 76 per cent of online shoppers prefer to buy with information in their own language, so local wording matters.
Translating your keyword list into another language gives you words, not demand. International keyword research means finding the terms people in each market actually type, in the spelling, script and search engine they actually use. Below is the process I follow, the traps that catch most teams, and a template for a keyword map you can build per market.
Why a translated keyword list misleads you
Imagine a Karachi furniture maker that ranks well at home for “sofa set” and wants to sell in the UK and Saudi Arabia. The tempting shortcut is to hand the keyword list to a translator, receive an Arabic list back, and start writing pages. It feels efficient. It usually produces pages that target phrases almost nobody searches.
The reasons are predictable:
- Translation gives the dictionary word, not the search word. A translator picks the correct term. Searchers pick the familiar one, which may be a loanword, a brand name used generically, or slang.
- Different markets think about the product differently. One market searches by material, another by room, another by price. The intent behind the query changes, not just the language.
- The same language varies by country. English in the UK, US and Pakistan differs in vocabulary and spelling. Arabic search behaviour in Egypt and the Gulf is not identical.
- Volume does not transfer. A phrase that drives your home market may have a tiny equivalent abroad while a phrase you never considered carries the demand.
So treat your home keyword list as a list of concepts to investigate, not a list of phrases to translate.
The four layers of variation to check in every market
1. Different terms for the same thing
English alone shows how much this matters. British searchers say “trousers”, “mobile phone” and “flat”, while American searchers say “pants”, “cell phone” and “apartment”. If you write one English page and use only one set of terms, you are invisible to half your English-speaking audience. Collect the real local term for each concept before anything else.
2. Spelling variants
Even when the word is the same, the spelling may not be. “Optimisation” and “optimization”, “colour” and “color”, “jewellery” and “jewelry”. Search engines handle many of these, but not all, and your page copy should still match the spelling your local reader expects. Using American spelling on a page aimed at UK buyers is a small trust signal working against you.
3. Script and transliteration
This is the layer that translation-first approaches miss completely. Many people search in their own language using Latin letters rather than their native script. Roman Urdu is the obvious example from Pakistan: a user may type a phrase like “sasta mobile” or “ghar ka khana delivery” instead of writing it in Urdu script, and often mixes in English words. The same habit appears with Hindi (often called Hinglish when mixed with English), and with Arabic written in Latin letters and numbers on chat apps.
For keyword research this means you may need three lists for one market: native script, transliterated, and the English loanwords people mix in. Transliteration is also inconsistent. The same Urdu word can be spelled several ways in Roman letters, so collect the variants rather than picking one.
4. Intent and format differences
Two markets can use the same words with different intent. A query that is mostly informational in one country may be mostly commercial in another. Look at what currently ranks for the term in that market: guides, product pages, marketplaces, videos or local directories. The result type tells you what kind of page you would need to compete.
Know which search engine you are researching for
Google is dominant in most of the markets my readers target, including Pakistan, the UK and the Gulf. It is not dominant everywhere, and researching in Google Keyword Planner for a market where people use a different engine gives you a distorted picture.
- China: Google is not generally accessible, and Baidu is the main search engine. Baidu has its own tools, ranking factors and preference for Simplified Chinese content.
- Russia: Yandex is a major search engine with its own webmaster tools and keyword data.
- South Korea: Naver is widely used alongside Google and presents results very differently, with heavy use of its own content services.
Also remember that search is not only on search engines. In many markets people research products on marketplaces, YouTube or social platforms. If your buyers search on a marketplace first, that marketplace’s search bar is part of your keyword research too.
If none of your markets fall into these exceptions, you can work mainly in Google’s data. Just confirm that assumption instead of making it.
A step-by-step research process for each market
- Turn your home keywords into concepts. Group your existing keywords into the underlying needs: “buy a sofa”, “compare sofa materials”, “sofa delivery”, and so on. You are researching each concept, not each phrase.
- Brainstorm local seed terms with a native speaker. Ask someone who lives in the market, or recently did, how they would search for each concept. Ask for the casual version, not the formal one. This is where you find loanwords, slang and transliterated forms.
- Expand with tools set to the right country and language. Use Google Keyword Planner, Search Console, and any third-party tool you trust, but always set the location and language to the target market. Data from the wrong country is worse than no data because it looks convincing.
- Study the local results page. Search each main term in the target market’s version of the search engine, or with the location set appropriately. Note who ranks, what type of page ranks, and what wording they use in titles. Competitors’ local pages are a fast source of real vocabulary.
- Mine autocomplete and related searches. Type seed terms into the local search engine and record the suggestions. These reflect what people in that market actually type, including transliterated forms that keyword tools often miss.
- Validate with a native speaker. Take your shortlist back to a native speaker and ask three questions: would you search this, does it sound natural, and does it mean what we think it means. Remove anything that fails.
- Assign keywords to pages in a market-specific map. Each market gets its own map. Do not assume the page structure from your home market applies everywhere.
Step six is the one teams skip, and it is the one that prevents embarrassing mistakes. A term can be technically correct and still sound odd, dated or even rude to a local reader.
Building a market-specific keyword map
A keyword map links each target page to its main keyword, its supporting variants and its intent. For international work, build one per market and keep them side by side so you can see where markets differ. Here is an illustrative layout for the hypothetical furniture maker above. The terms are examples of the kind of variation to look for, not researched data.
| Concept | Pakistan (English and Roman Urdu) | UK | Page type |
|---|---|---|---|
| Main product category | sofa set, sofa set design, sofa set price | sofas, corner sofa, sofa bed | Category page |
| Price-led search | sasta sofa set, sofa set price in Pakistan | cheap sofas, sofa sale | Category page with filters |
| Material comparison | wooden sofa vs fabric sofa | fabric vs leather sofa | Guide |
| Delivery | sofa home delivery Karachi | sofa delivery times, next day sofa delivery | Service or FAQ page |
A few rules make the map useful rather than decorative:
- One primary keyword per page per market. Supporting variants go in headings, body copy and FAQs.
- Record the intent and the result type you saw. That decides what you build.
- Note the source of each keyword (tool, autocomplete, native speaker, competitor) so you know how much to trust it.
- Allow markets to have pages others do not. If UK buyers search heavily for a concept your home market ignores, it deserves its own page there.
Your keyword map also feeds your site structure. If two English-speaking markets need clearly different vocabulary, they probably need separate country versions rather than one shared English page. That ties directly into the choice I cover in ccTLD vs subfolder vs subdomain, and into how you connect the versions, covered in my guide on hreflang for multi-country websites.
Turning research into localised pages
Research only pays off if the writing uses it. Brief your writers or translators with the market’s keyword map, not the home market’s copy. This is the difference between translation and localisation, and I go deeper into the process in how to do multilingual SEO.
A practical checklist for each localised page:
- The primary keyword in the title, the main heading and the opening paragraph, written naturally.
- Local spelling throughout, including in navigation and buttons.
- Transliterated or loanword variants used where native speakers confirmed they are common, without stuffing them in.
- Local units, currency, examples and references.
- A meta description written fresh for that market, not translated.
- Correct hreflang linking it to its equivalents in other markets.
If you sell business to business, the vocabulary question often shifts toward trade terms, product codes and specifications, which overseas buyers search in quite precise ways. I cover that angle in international SEO for exporters.
Questions people ask about international keyword research
Should I target Roman Urdu keywords or Urdu script keywords?
Often both, because they are different audiences and habits. Check what ranks for each version and how your audience actually writes. For many commercial searches in Pakistan, English and Roman Urdu mixes are common, so do not assume Urdu script is the only local option.
Can I rely on keyword tools for small markets?
Treat their numbers as rough. Tools often have thin data for smaller countries and less common languages, and they can miss transliterated searches entirely. Use them for direction, then confirm with autocomplete, local results and a native speaker.
Is machine translation good enough for finding keywords?
It is fine for understanding what a term means. It is not reliable for choosing which term to target, because it returns the standard word rather than the searched word. Use it as a starting point for conversations with native speakers, not as the answer.
How often should I redo international keyword research?
Revisit each market’s map when you add products, when performance in a market changes noticeably, or at least once a year. Vocabulary, competitors and search features shift, and a map built once and never checked slowly drifts out of date.
Market-specific keyword research is the foundation everything else in international search sits on, and it is where I start with every new market. If you would like help building keyword maps for the countries you are targeting, it is part of the international SEO work I do. Ask for a free audit and I will review where your current keyword targeting fits each market and where it does not.