How to Do Keyword Research for Multiple Languages

Keyword research in one language is already messy. Add four more languages, three regional variants, and a stakeholder who asks "can't we just use the English list in Google Translate?" — and you have the setup for an international SEO program that looks busy but targets the wrong queries everywhere.

Multilingual keyword research is not translation. It is market-by-market discovery of how real people search, what they expect to find, and which pages on your site should answer them. Done properly, it becomes the foundation for URL planning, content localization, and hreflang mapping.

If you need the broader strategic context, read The Ultimate Guide to Multilingual SEO in 2026 first. This guide walks through the workflow we use for clients entering new languages.

Why Direct Translation of Keywords Fails

English seed keyword: "project management tool"

Direct French translation: "outil de gestion de projet"

French search reality: High-volume queries include "logiciel gestion de projet," "outil gestion projet gratuit," and brand-led comparisons like "alternative à Asana." Searchers use shorter phrasing; modifiers differ; English loanwords appear in B2B SaaS.

We tracked a productivity SaaS that optimized French pages for translated English strings. Average position in France: 34. After rebuilding the keyword map with native research, average position on priority terms moved to 11 within five months — same URLs, rewritten metadata and H1s.

The lesson: every market gets its own seed list, informed by English research but not copied from it.

Step 1: Start With Business Goals and Personas Per Market

Before opening any tool, document:

• Revenue priority markets (not every language launches day one)
• Buyer personas in that market (titles, pain points, buying committees differ — German Mittelstand procurement is not US startup founder culture)
• Product positioning (are you premium, freemium, enterprise-only?)
• Existing content assets worth localizing vs. net-new pages

This scoping prevents keyword lists bloated with irrelevant volume. A million monthly searches for a generic term you cannot rank for wastes localization budget.

Step 2: Build Seed Topics in English, Then Decouple

English seed topics provide structure — not final keywords. Group seeds by funnel stage:

• Awareness: "what is CRM," "remote team challenges"
• Consideration: "CRM comparison," "best CRM for small business"
• Decision: "CRM pricing," "CRM free trial"

For each seed topic, assign a content type (blog, product page, comparison, glossary) before researching locally. This aligns with translation vs. localization decisions downstream.

Step 3: Use the Right Tools for Each Language

No single tool covers every market equally well. Our typical stack:

Tool	Strength	Limitation
Google Keyword Planner	Volume in Ads-supported countries	Groups close variants
Ahrefs / Semrush	Multi-country databases	Weaker in some APAC markets
AnswerThePublic	Question modifiers	Needs native validation
Google Trends	Seasonality	Relative, not absolute volume

For languages with thin tool data, supplement with Google Autocomplete, competitor title/H1 analysis, native speaker interviews, and Search Console queries once you have indexation in that locale.

Never reject a keyword because a tool shows zero volume — long-tail clusters add up, and tools underreport non-English queries.

Step 4: Validate With Native Speakers

Keyword tools miss nuance. A Swedish term may be technically correct but unused in B2B contexts. Brazilian Portuguese accepts different prepositions than European Portuguese for the same product search.

Validation workflow:

1. Export top 50 candidate keywords per topic from tools
2. Send to in-market reviewer (not just fluent speaker — someone who searches professionally in that vertical)
3. Tag each term: Primary, Secondary, Reject, Note colloquial variant
4. Document preferred usage in a locale glossary shared with translators

This step costs hours, not weeks, and prevents expensive rewrites after launch.

Step 5: Map Search Intent Locally

Search intent is not universal. The SERP tells the truth. For each priority keyword:

1. Search in the target country (correct Google domain or local engine)
2. Classify SERP composition: product pages, listicles, forums, video, news
3. Compare to your planned page type
4. Adjust keyword target or page format if misaligned

Example: In the US, "invoice software" returns mostly commercial homepages. In Italy, "software fatturazione" mixes comparison articles and government e-invoicing guidance. A direct translation of your US product page may lose to editorial content unless you add comparison elements or localized compliance content.

Intent mismatch is why localized pages rank poorly despite perfect hreflang implementation.

Step 6: Organize Keywords Into a Localization Matrix

Build a spreadsheet (or database) with columns:

• Market / language code
• Seed topic
• Primary keyword
• Secondary keywords
• Search intent
• Target URL (existing or planned)
• Content action (translate, localize, transcreate, create new)
• Priority (P1–P3)
• SERP notes

Share this with SEO, content, and localization teams before TMS handoff. Keywords drive title tags, H1s, URL slugs, and internal anchor text — not just body copy.

Sample Row

Market	Primary keyword	Intent	URL	Action
DE	Zeiterfassung Software	Commercial	/de/produkte/zeiterfassung/	Localize from /products/time-tracking/

One row, one owner, one ship date.

Step 7: Account for Regional Variants Within a Language

Spanish is not monolithic. Keywords for Mexico differ from Spain and Argentina. Handle this two ways:

• Separate URLs and hreflang (es-MX, es-ES) when content and keywords genuinely differ
• Single URL with language-only hreflang (es) when differences are minor (rare for commercial terms)

We prefer separate URLs when conversion elements change (currency, shipping, legal). Keyword research should run per country, not per language, for high-stakes markets.

Same logic applies to English (en-US, en-GB, en-AU), Portuguese (pt-BR, pt-PT), and French (fr-FR, fr-CA).

Step 8: Prioritize Keywords You Can Actually Win

International SEO budgets are finite. Score keywords by business value, ranking feasibility, content cost, and strategic fit. Launch P1 keywords at launch; queue P2 for the next quarter. Avoid targeting translated English head terms in every market simultaneously — you spread authority too thin.

Step 9: Monitor and Refresh Quarterly

Search behavior shifts. New competitors localize. For each market, quarterly: pull Search Console performance by country, identify striking-distance keywords (positions 8–20) for refreshes, add emerging questions from support tickets, and prune terms that never gained impressions.

Common Mistakes to Avoid

• Trusting English difficulty scores for other languages — competition differs
• Ignoring zero-click SERPs — featured snippets reduce CTR even at #1
• Keyword stuffing translated pages — unnatural density hurts readability
• Skipping internal linking keywords — anchor text in each language reinforces topical clusters

Tie Keyword Research to the Rest of Your Stack

Keywords without architecture are wishlists. Connect your matrix to:

• URL and hreflang planning
• Localization scope per page
• Content calendar and link building in-market
• Paid search validation (high CPC terms often signal valuable organic intent)

When keyword research, technical SEO, and localization share one source of truth, international programs stop guessing and start compounding.

Multilingual SEO Services builds keyword datasets and localization briefs for teams expanding into new languages. Contact us with your target markets — we will help you find what your customers actually type into the search box.