International SEO skill 06 of 13

Multilingual Keyword Research SKILL.md

The Multilingual Keyword Research skill makes Claude, Codex or another AI agent research keywords natively in each country and language instead of translating English lists. It generates seeds with native speakers, uses tools with local settings, local autocomplete, competitors and marketplaces, validates meaning and intent in local results, handles regional variants and English terms used locally, and clusters terms into one page per market.

multilingual-keyword-research/SKILL.md, version 1.0.0, 1,596 words.

Translating an English keyword list is the most common mistake in international SEO. People in each market describe the same need in their own words, with different terms, spellings, word order, slang and borrowed English. A Spaniard and a Mexican may search differently; a German might search the English term for a software category; a Brazilian searches in Portuguese with local brand names. Pages built on translated keywords target searches nobody makes.

This skill researches keywords natively in each market, with native speakers and local data, maps them to intent and pages, and compares demand across markets to guide prioritisation.

Operating principles

  1. Research in the market, in the language. Use country and language settings in every tool and real local results.
  2. Native speakers validate. Tools miss nuance; natives catch wrong meanings, formal versus informal terms and local slang.
  3. Intent can differ by market. The same term may show guides in one country and product pages in another.
  4. English terms are sometimes local terms. Many markets search English words for tech and business categories.
  5. Label every volume with tool, country and date; small markets have unreliable estimates.

Sources by market

SourceUse
Search Console filtered by countryReal queries already reaching the site
Keyword tools with country and language settingsVolumes and ideas (label estimates)
Local autocomplete, People also ask and related searchesReal phrasing
Local competitors' navigation and rankingsMarket vocabulary
Local marketplaces and forumsProduct language and questions
Native customers, sales and support teamsTerms customers actually use
Local search engines (Baidu, Yandex, Naver)Demand where Google isn't dominant

When to use, and when not to

Use it before localising content, when entering a market, when translated pages underperform and when comparing demand between candidate markets.

Don't use it for single-language keyword research (the Content SEO keyword-research skill) or for translation workflows (localisation-translation).

Inputs to ask for, in one message

  1. Markets and languages, and the products or topics to research.
  2. The English (or source) keyword set and page map, if any.
  3. Search Console data by country.
  4. Keyword tool access with country settings.
  5. Native speakers available to review.
  6. Local competitors you know of.

Raw SKILL.md file

.md

This is the exact file agents read, including the credit header and licence notice. Save it as multilingual-keyword-research/SKILL.md.

---
name: multilingual-keyword-research
description: "Research keywords natively in each country and language instead of translating English lists: native seed generation, tools with country and language settings, local autocomplete, competitors and marketplaces, native validation of meaning and intent, regional variants, English terms used locally, clustering per market and demand comparisons across markets. Use whenever someone asks for keyword research in another language, international keyword research, translating keywords, Spanish or French SEO keywords, comparing demand across countries or why translated pages don't rank."
license: CC-BY-4.0
metadata:
  author: Afeez Adebayo
  author_url: https://afeezadebayo.com
  author_linkedin: https://www.linkedin.com/in/afeez-adebayo/
  author_x: https://x.com/za_icon
  source: https://afeezadebayo.com/skills-md/marketing/international-seo/multilingual-keyword-research/
  category: marketing
  subcategory: international-seo
  version: "1.0.0"
  created: "2026-10-10"
  updated: "2026-10-10"
---

# Multilingual Keyword Research

Translating an English keyword list is the most common mistake in international SEO. People in each market describe the same need in their own words, with different terms, spellings, word order, slang and borrowed English. A Spaniard and a Mexican may search differently; a German might search the English term for a software category; a Brazilian searches in Portuguese with local brand names. Pages built on translated keywords target searches nobody makes.

This skill researches keywords natively in each market, with native speakers and local data, maps them to intent and pages, and compares demand across markets to guide prioritisation.

## Operating principles
1. **Research in the market, in the language.** Use country and language settings in every tool and real local results.
2. **Native speakers validate.** Tools miss nuance; natives catch wrong meanings, formal versus informal terms and local slang.
3. **Intent can differ by market.** The same term may show guides in one country and product pages in another.
4. **English terms are sometimes local terms.** Many markets search English words for tech and business categories.
5. **Label every volume** with tool, country and date; small markets have unreliable estimates.

## Sources by market
| Source | Use |
|---|---|
| Search Console filtered by country | Real queries already reaching the site |
| Keyword tools with country and language settings | Volumes and ideas (label estimates) |
| Local autocomplete, People also ask and related searches | Real phrasing |
| Local competitors' navigation and rankings | Market vocabulary |
| Local marketplaces and forums | Product language and questions |
| Native customers, sales and support teams | Terms customers actually use |
| Local search engines (Baidu, Yandex, Naver) | Demand where Google isn't dominant |

## When to use, and when not to
Use it before localising content, when entering a market, when translated pages underperform and when comparing demand between candidate markets.

Don't use it for single-language keyword research (the Content SEO `keyword-research` skill) or for translation workflows (`localisation-translation`).

## Inputs to ask for, in one message
1. Markets and languages, and the products or topics to research.
2. The English (or source) keyword set and page map, if any.
3. Search Console data by country.
4. Keyword tool access with country settings.
5. Native speakers available to review.
6. Local competitors you know of.

## Workflow

### Step 1. Define topics, not keywords
List the needs and products to research rather than English phrases to translate.

### Step 2. Generate seeds natively
With a native speaker, list how locals describe each topic; add terms from local competitors, marketplaces and support.

### Step 3. Expand with tools
Run seeds with country and language settings; collect autocomplete and related searches from local results.

### Step 4. Validate meaning and intent
Natives check each term's meaning; check live local results to confirm intent and page type.

### Step 5. Cluster and map
Group terms that share results; map each cluster to one page per market.

### Step 6. Compare markets
Compare demand, competition and intent across markets for the same topic.

### Step 7. Hand to content
Give translators and writers target terms per page, with notes on register and variants.

## Common market differences
| Difference | Example |
|---|---|
| Different word for the same thing | UK "current account" vs US "checking account" |
| English term used locally | German searches for "CRM Software" alongside local terms |
| Regional variants of one language | es-ES "ordenador" vs es-MX "computadora" |
| Formal versus informal phrasing | Different tone in questions and queries |
| Different intent | Guide results in one market, product results in another |
| Local brand or regulation names | Payment, tax and legal terms specific to the country |

## Native reviewer brief
Give reviewers a short brief so feedback is consistent:
- The product and audience in one paragraph.
- The list of candidate terms per topic, with estimated volumes.
- Questions: Would you search this? What would you type instead? Is this formal, informal or technical? Does it mean anything else locally? Which results would you expect?
- A column for preferred terms and notes on regional variants.

## Market comparison template
| Topic | Market | Main term | Est. volume (tool, date) | Intent | Competition | Notes |
|---|---|---|---|---|---|---|
| Invoicing software | es-ES | programa de facturaci\u00f3n | estimate | Commercial | Medium | Formal tone |
| Invoicing software | es-MX | facturaci\u00f3n electr\u00f3nica | estimate | Regulatory and commercial | High | CFDI rules drive demand |
Comparing topics across markets side by side shows where one page can serve several countries and where each market needs its own.

## Step-by-step session with a native speaker (60 minutes)
1. Explain the product and audience (5 min).
2. For each topic, ask how they would search for it, with no prompting (15 min).
3. Show tool suggestions and ask which they would use, which sound odd and which mean something else (20 min).
4. Search the top terms together in local results and discuss intent (15 min).
5. Agree preferred terms, variants and register (5 min).

## Search Console mining by country
- Filter Performance by country and export queries for 16 months.
- Separate queries in the local language from English queries.
- Find local-language queries already reaching English pages: strong candidates for localised pages.
- Note branded queries by country to gauge brand awareness per market.

## Volume reliability
| Market size | Reliability of tool estimates | What to lean on |
|---|---|---|
| Large (US, UK, DE, FR, JP) | Reasonable for head terms | Tools plus Search Console |
| Mid-size (NL, SE, PL, MX) | Patchy for long tail | Search Console, autocomplete, natives |
| Small markets | Often zero or missing | Natives, autocomplete, local competitors |
Always show estimates as ranges or labels, not precise figures.

## Handing research to content teams
For each page, give the main term, two to five supporting terms, the intent, the expected page type from local results, register notes and examples of local competitors' pages. Writers and translators then build pages around what locals search instead of converting English pages word for word.

## Edge cases and judgement calls
| Situation | What to do |
|---|---|
| Tiny markets with zero-volume estimates | Use Search Console, autocomplete and native judgement |
| One language, several countries | Research each country; decide whether one page can serve all |
| Languages without spaces (Chinese, Japanese, Thai) | Use native tools and speakers; check segmentation in tools |
| Mixed-language queries | Include them; they reflect real behaviour |
| Local search engine markets | Research in that engine's tools too |

## Handoffs to related skills
- Market prioritisation: `international-market-selection`
- Localisation: `localisation-translation`
- Content plans per market: `international-content-strategy`
- Local engines: `local-search-engines`
- Single-market research methods: the Content SEO `keyword-research` skill

## Worked example
A UK invoicing SaaS translated its keyword list into Spanish for Spain and Mexico. Native research showed Spanish users searched "programa de facturación" and "factura electrónica", while Mexican searches centred on "facturación electrónica" and CFDI, the country's electronic invoicing standard, with very different intent driven by tax rules. The company split es-ES and es-MX pages, built a CFDI guide for Mexico and rewrote Spanish pages around local terms. Impressions in both markets rose, and Mexico became the stronger market once the local regulatory angle was covered.

## Output format
```
# Multilingual keyword research: <topics>, <markets>, <date>
## Topics and native seed lists
## Clusters per market (terms, est. volume with tool and date, intent, page)
## Market comparison
## Notes for translators and writers
```

## Common mistakes
- Translating English keywords.
- Research with the wrong country setting.
- No native review.
- Assuming one Spanish, French or English keyword set serves every country.
- Ignoring English terms used locally.

## Questions to ask before you research
- Which topics and products matter in each market?
- Who can review terms as a native speaker?
- Which search engines matter in each market?

## Self-check before you answer
- Were seeds generated natively, not translated?
- Are volumes labelled with tool, country and date?
- Was intent checked in local results?
- Are clusters mapped to one page per market?

## Official documentation and tools
- Google Keyword Planner: https://support.google.com/google-ads/answer/7337243
- Google Trends (by country): https://trends.google.com/
- Google Search Console Performance report: https://support.google.com/webmasters/answer/7576553

---
Created by Afeez Adebayo, Marketing, SEO & Growth Expert
Website: https://afeezadebayo.com
LinkedIn: https://www.linkedin.com/in/afeez-adebayo/
X: https://x.com/za_icon
Source: https://afeezadebayo.com/skills-md/marketing/international-seo/multilingual-keyword-research/

Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). You may copy, share and adapt this skill, including for commercial use, as long as you credit Afeez Adebayo and link to the source URL above. Keep this notice in any copy or adapted version.

Workflow

Step 1. Define topics, not keywords

List the needs and products to research rather than English phrases to translate.

Step 2. Generate seeds natively

With a native speaker, list how locals describe each topic; add terms from local competitors, marketplaces and support.

Step 3. Expand with tools

Run seeds with country and language settings; collect autocomplete and related searches from local results.

Step 4. Validate meaning and intent

Natives check each term's meaning; check live local results to confirm intent and page type.

Step 5. Cluster and map

Group terms that share results; map each cluster to one page per market.

Step 6. Compare markets

Compare demand, competition and intent across markets for the same topic.

Step 7. Hand to content

Give translators and writers target terms per page, with notes on register and variants.

Common market differences

DifferenceExample
Different word for the same thingUK "current account" vs US "checking account"
English term used locallyGerman searches for "CRM Software" alongside local terms
Regional variants of one languagees-ES "ordenador" vs es-MX "computadora"
Formal versus informal phrasingDifferent tone in questions and queries
Different intentGuide results in one market, product results in another
Local brand or regulation namesPayment, tax and legal terms specific to the country

Native reviewer brief

Give reviewers a short brief so feedback is consistent:

  • The product and audience in one paragraph.
  • The list of candidate terms per topic, with estimated volumes.
  • Questions: Would you search this? What would you type instead? Is this formal, informal or technical? Does it mean anything else locally? Which results would you expect?
  • A column for preferred terms and notes on regional variants.

Market comparison template

TopicMarketMain termEst. volume (tool, date)IntentCompetitionNotes
Invoicing softwarees-ESprograma de facturaci\u00f3nestimateCommercialMediumFormal tone
Invoicing softwarees-MXfacturaci\u00f3n electr\u00f3nicaestimateRegulatory and commercialHighCFDI rules drive demand

Comparing topics across markets side by side shows where one page can serve several countries and where each market needs its own.

Step-by-step session with a native speaker (60 minutes)

  1. Explain the product and audience (5 min).
  2. For each topic, ask how they would search for it, with no prompting (15 min).
  3. Show tool suggestions and ask which they would use, which sound odd and which mean something else (20 min).
  4. Search the top terms together in local results and discuss intent (15 min).
  5. Agree preferred terms, variants and register (5 min).

Search Console mining by country

  • Filter Performance by country and export queries for 16 months.
  • Separate queries in the local language from English queries.
  • Find local-language queries already reaching English pages: strong candidates for localised pages.
  • Note branded queries by country to gauge brand awareness per market.

Volume reliability

Market sizeReliability of tool estimatesWhat to lean on
Large (US, UK, DE, FR, JP)Reasonable for head termsTools plus Search Console
Mid-size (NL, SE, PL, MX)Patchy for long tailSearch Console, autocomplete, natives
Small marketsOften zero or missingNatives, autocomplete, local competitors

Always show estimates as ranges or labels, not precise figures.

Handing research to content teams

For each page, give the main term, two to five supporting terms, the intent, the expected page type from local results, register notes and examples of local competitors' pages. Writers and translators then build pages around what locals search instead of converting English pages word for word.

Edge cases and judgement calls

SituationWhat to do
Tiny markets with zero-volume estimatesUse Search Console, autocomplete and native judgement
One language, several countriesResearch each country; decide whether one page can serve all
Languages without spaces (Chinese, Japanese, Thai)Use native tools and speakers; check segmentation in tools
Mixed-language queriesInclude them; they reflect real behaviour
Local search engine marketsResearch in that engine's tools too
  • Market prioritisation: international-market-selection
  • Localisation: localisation-translation
  • Content plans per market: international-content-strategy
  • Local engines: local-search-engines
  • Single-market research methods: the Content SEO keyword-research skill

Worked example

A UK invoicing SaaS translated its keyword list into Spanish for Spain and Mexico. Native research showed Spanish users searched "programa de facturación" and "factura electrónica", while Mexican searches centred on "facturación electrónica" and CFDI, the country's electronic invoicing standard, with very different intent driven by tax rules. The company split es-ES and es-MX pages, built a CFDI guide for Mexico and rewrote Spanish pages around local terms. Impressions in both markets rose, and Mexico became the stronger market once the local regulatory angle was covered.

Output format

# Multilingual keyword research: <topics>, <markets>, <date>
## Topics and native seed lists
## Clusters per market (terms, est. volume with tool and date, intent, page)
## Market comparison
## Notes for translators and writers

Common mistakes

  • Translating English keywords.
  • Research with the wrong country setting.
  • No native review.
  • Assuming one Spanish, French or English keyword set serves every country.
  • Ignoring English terms used locally.

Questions to ask before you research

  • Which topics and products matter in each market?
  • Who can review terms as a native speaker?
  • Which search engines matter in each market?

Self-check before you answer

  • Were seeds generated natively, not translated?
  • Are volumes labelled with tool, country and date?
  • Was intent checked in local results?
  • Are clusters mapped to one page per market?

Official documentation and tools

When to use this skill, and when not to

Use it before localising content, when entering a market, when translated pages underperform and when comparing demand between candidate markets.

Translating an English keyword list is the most common mistake in international SEO. People in each market describe the same need in their own words, with different terms, spellings and borrowed English, so pages built on translated keywords target searches nobody makes.

Don’t use it for single-language research (use the Content SEO Keyword Research skill) or for translation workflows (use Localisation and Translation).

Example prompts

  • “Research invoicing keywords for Spain and Mexico separately.”
  • “Do Germans search the English term for CRM software?”
  • “Mine Search Console for local-language queries by country.”
  • “Compare demand for our product across five markets.”
  • “Write a brief for our native keyword reviewers.”

Example output

The skill returns a fixed structure. A shortened example for an imaginary site:

# Multilingual keyword research: invoicing, es-ES and es-MX, 11 October 2026

| Market | Main term | Intent | Notes |
| es-ES | programa de facturaci\u00f3n | Commercial | Formal tone |
| es-MX | facturaci\u00f3n electr\u00f3nica | Regulatory + commercial | CFDI rules drive demand |

## Decision
Separate es-ES and es-MX pages; CFDI guide for Mexico
## Volumes labelled as estimates (tool, country, date)

Every volume is labelled with its tool, country and date, because estimates in mid-size and small markets are often unreliable.

What the agent needs from you

The skill asks for markets and languages, the topics or products to research, any existing keyword set and page map, Search Console data by country, keyword tool access with country settings, native speakers available to review and local competitors you know of.

How this skill hands off to related skills

Keyword research feeds planning and content. Market priorities go to Market Selection, translation briefs to Localisation, per-market plans to Content Strategy and engine-specific research to Local Search Engines.

How to run Multilingual Keyword Research step by step

  1. Install the skill as multilingual-keyword-research/SKILL.md.
  2. Define topics rather than English keywords to translate.
  3. Run a seed session with a native speaker.
  4. Expand with local tools and Search Console by country.
  5. Validate intent in local results.
  6. Cluster, map to pages and brief writers.

Situations this skill handles

Markets differ in surprising ways. The skill handles these situations:

SituationWhat the skill does
Tiny markets with no volumesSearch Console, autocomplete and natives
One language, several countriesResearches each country separately
Languages without spacesNative tools and segmentation checks
Mixed-language queriesIncludes them as real behaviour
Local engine marketsResearches in that engine’s tools too
Regulation-driven demandBuilds local guides around the rules

What makes this skill different

Most international keyword work translates a list and runs it through a tool. This skill starts from topics, generates seeds with native speakers, validates every term’s meaning and intent in local results and compares markets side by side.

The output is ready for writers and translators: target terms per page, register notes and examples of local competitors’ pages.

Practical tips from real projects

  • Topics first, terms second.
  • Natives validate every list.
  • Check intent in local results.
  • Label every estimate.

A worked example: a UK invoicing SaaS translated its keyword list into Spanish for Spain and Mexico. Native research showed Spanish users searched “programa de facturación” and “factura electrónica”, while Mexican searches centred on “facturación electrónica” and CFDI, the country’s e-invoicing standard, with intent driven by tax rules. Splitting es-ES and es-MX pages, building a CFDI guide and rewriting Spanish pages around local terms raised impressions in both markets, and Mexico became the stronger one.

Sources differ in value: Search Console by country shows real queries already arriving; tools with local settings give estimates; local autocomplete, People also ask and related searches show phrasing; local competitors and marketplaces show market vocabulary; native customers, sales and support teams supply the words customers really use; and local engines matter where Google isn’t dominant.

Common market differences include different words for the same thing (current account versus checking account), English terms used locally, regional variants of one language (ordenador versus computadora), formal versus informal phrasing, different intent behind the same term, and local brand or regulation names.

The native seed session follows a simple hour: explain the product, ask how they would search with no prompting, show tool suggestions and ask what sounds wrong, search top terms together in local results and agree preferred terms and register.

Search Console mining by country finds local-language queries already reaching English pages, which are strong candidates for the first localised pages, and branded queries by country show where awareness already exists.

Volume reliability varies by market size: reasonable for head terms in large markets, patchy for the long tail in mid-size markets and often missing in small ones, so estimates are shown as labels or ranges rather than precise figures.

Handing research to content teams is where it pays off. For each page, writers and translators receive the main term, two to five supporting terms, the intent, the page type that ranks locally, notes on register and regional variants and examples of local competitors’ pages. With that brief, they build pages around what locals search rather than converting English pages word for word, and reviewers have a clear standard to check against.

A native reviewer brief keeps feedback consistent: a paragraph on the product and audience, candidate terms with estimates, and questions such as whether they would search this, what they would type instead, whether it sounds formal or technical and whether it means anything else locally.

Repeat the research yearly as local language and demand shift.

Frequently asked questions

Translating gives you a starting list at best. Real local terms often differ, so research natively with local tools, results and speakers.

For some categories, especially tech and business, yes. Include English terms used locally alongside local-language terms.

Usually not. Spain and Mexico, for example, use different words and face different regulations, which changes intent.

Often unreliable or zero. Lean on Search Console, autocomplete, local competitors and native judgement.

They catch wrong meanings, formal versus informal phrasing, slang and terms tools miss.

Search the term with local settings and see what types of pages rank.

Changelog

  • 1.0.0, 10 October 2026: first release.

Use, adapt and share this skill under CC BY 4.0: credit Afeez Adebayo and link to this page.