Keyword Research For A New Language Market

Translating a keyword list translates your assumptions about the market with it. The word for the same product often differs in register, in whether the English loan word is used at all, and in whether local buyers approach the category the same way. Machine translation returns a grammatically correct phrase nobody types, so the page looks finished on the plan and ranks for nothing.

The skill file

---
name: keyword-set-for-a-new-language-market
description: Use when expanding into another language or country and the plan is to translate the keyword list you already have.
---

# Keyword Research For A New Language Market

Translating a keyword list translates your assumptions about the market with it. The word for the same product often differs in register, in whether the English loan word is used at all, and in whether local buyers approach the category the same way. Machine translation returns a grammatically correct phrase nobody types, so the page looks finished on the plan and ranks for nothing.

## What you need first

- country and language treated as separate variables - Spanish for Mexico is not Spanish for Spain
- a native speaker who buys in this category, separate from whoever writes the translation
- a competitor set taken from a target-country SERP rather than your home competitor list
- the search engine share for that market, since Google is not dominant everywhere

## Method

1. Establish which engine matters before anything else. Yandex, Naver and Baidu hold enough share in their markets that Google-only research is simply the wrong dataset, and their result behaviour differs.
2. Search two or three core products natively and collect the terms the local top-ranking pages use in their titles and headings. That is the market's own vocabulary and it costs about an hour to gather.
3. Test the loan word against the local word for the same concept and pull volume for both. Both usually exist with very different volumes, and in technical categories the English term frequently wins outright.
4. Run autocomplete in-country and in-language rather than from your own location, because suggestions are localised and register shows up there before it shows up in any volume column.
5. Have the native reviewer mark every term natural, awkward or wrong. Awkward terms with volume are usually fine, since people type awkwardly; a wrong term with volume normally means the phrase means something else locally and the page would pull the wrong visitor.
6. Size total demand for the market before committing production. Expansions often fail because the whole category there is a fraction of the home market, which is a business decision better made before the pages exist.
7. Map terms to pages independently of the source site. Mirroring the home structure one to one guarantees empty pages where local demand does not exist and misses demand the home market never had.

## What this produces

A per-market keyword set in native phrasing, with a page list that deliberately differs from the source site and a demand estimate for the market.

## Where this goes wrong

- Reusing one language plan across every country that speaks it, so terminology, competitors and volume are wrong in most of them.
- Letting the translation supplier pick the target phrases, since they are optimising for correctness while you need what people actually type.
- Checking volume with the location left on your home country, which measures diaspora demand rather than market demand.

---

From the QuQi skill library - https://www.quqi.io/skills/keyword-set-for-a-new-language-market

What you need first

• country and language treated as separate variables - Spanish for Mexico is not Spanish for Spain
• a native speaker who buys in this category, separate from whoever writes the translation
• a competitor set taken from a target-country SERP rather than your home competitor list
• the search engine share for that market, since Google is not dominant everywhere

Method

1. 01 Establish which engine matters before anything else. Yandex, Naver and Baidu hold enough share in their markets that Google-only research is simply the wrong dataset, and their result behaviour differs.
2. 02 Search two or three core products natively and collect the terms the local top-ranking pages use in their titles and headings. That is the market's own vocabulary and it costs about an hour to gather.
3. 03 Test the loan word against the local word for the same concept and pull volume for both. Both usually exist with very different volumes, and in technical categories the English term frequently wins outright.
4. 04 Run autocomplete in-country and in-language rather than from your own location, because suggestions are localised and register shows up there before it shows up in any volume column.
5. 05 Have the native reviewer mark every term natural, awkward or wrong. Awkward terms with volume are usually fine, since people type awkwardly; a wrong term with volume normally means the phrase means something else locally and the page would pull the wrong visitor.
6. 06 Size total demand for the market before committing production. Expansions often fail because the whole category there is a fraction of the home market, which is a business decision better made before the pages exist.
7. 07 Map terms to pages independently of the source site. Mirroring the home structure one to one guarantees empty pages where local demand does not exist and misses demand the home market never had.

What this produces

A per-market keyword set in native phrasing, with a page list that deliberately differs from the source site and a demand estimate for the market.

Where this goes wrong

• Reusing one language plan across every country that speaks it, so terminology, competitors and volume are wrong in most of them.
• Letting the translation supplier pick the target phrases, since they are optimising for correctness while you need what people actually type.
• Checking volume with the location left on your home country, which measures diaspora demand rather than market demand.

Use this skill in your own AI

The download is a plain markdown file with the name and trigger in its frontmatter. Where an assistant supports skills it can load itself, that frontmatter is what it reads to decide this one applies.

Questions about this skill

More in Keyword research