Search Engine Optimisation
Multilingual SEO
The trap is treating this as a translation project. A word-for-word translation of a page optimised for English search rarely targets anything anyone types in the target language.
Multilingual SEO makes a site rank in more than one language. The Nexclick researches how each market actually searches rather than translating a UK keyword list, then briefs native translation and implements the hreflang that keeps the language versions from competing with each other.
Is this you?
What usually prompts the call
- You translated the site and the translated pages get almost no search traffic.
- Machine translation was used and native speakers have commented on it.
- Your language versions rank against each other rather than in their own markets.
- You are about to translate and want the keyword research done in the target language first.
What we do
The actual deliverables
Things that appear on an invoice, not adjectives.
- Keyword research in the target language, natively
- Not a translated UK list. What people actually type differs from the correct translation surprisingly often, and the gap is where most translated sites lose.
- Translation briefing rather than translation
- Briefs supplied to native translators covering target phrases, intent and required structure, so the translated page is optimised rather than merely accurate.
- hreflang for language and region
- Language codes where content differs by language, language-plus-region where it differs by market. Getting this distinction wrong is the most common implementation error.
- URL structure for language versions
- Subdirectories, subdomains or parameters, decided against your platform and market plans, and consistent so the structure scales when the next language is added.
- Cultural and commercial adaptation
- Currency, units, contact details, legal requirements and examples. A page listing a UK phone number in German is technically translated and commercially useless.
- Search engine differences per market
- Where a market uses Yandex, Naver or Baidu, the technical requirements differ and we will say so rather than assuming Google everywhere.
- Per-language reporting
- Traffic, rankings and conversions segmented by language, because an aggregate figure hides one version working and another not.
Comparison
Why translated pages fail to rank
The failure modes we find most often on translated sites, in rough order of frequency. Most are decided before a translator is briefed, which is why translation quality is rarely the actual problem.
| Problem | What it looks like | Fix |
|---|---|---|
| Keyword research not redone in language | Page targets the correct translation of a phrase nobody searches | Research natively before translating |
| Machine translation without native review | Grammatically odd; readers bounce; trust collapses | Native translator, briefed properly |
| Missing hreflang return tags | Wrong language shown to the wrong market | Bidirectional tags, validated by crawl |
| Language code where a region code is needed | Spanish site shown in Mexico, or vice versa | Use es-ES and es-MX where markets differ |
| No x-default fallback | Unmatched visitors get an arbitrary version | Add x-default pointing at a selector or the primary market |
| Currency and contact details not localised | Technically translated, commercially useless | Localise everything transactional |
| No links from the target market | Section has no authority in that market and cannot compete | Local link acquisition per market |
| Translated URLs left in English | Weaker relevance signal in market | Translate slugs where the platform allows |
| One language version canonicalised to another | Translated pages never index at all | Each version self-canonicalises |
| Auto-redirect by browser language | Googlebot only ever sees one version | Suggest, do not force — use a banner and a selector |
How it works
Step by step, with timeframes
Timeframes are typical rather than guaranteed, and they assume we get account access and approvals when we ask.
- 01Week 1–3
Research in language
Native keyword research per target language, with SERP verification. This has to happen before translation, not after.
- 02Week 3–5
Structure and brief
URL structure decided, hreflang planned, and translation briefs written per page.
- 03Month 2–4
Translate and implement
Native translation against the briefs, then technical implementation and validation.
- 04Monthly
Measure per language
Segmented reporting, with effort concentrated on the languages showing genuine traction rather than spread evenly.
What you get
Reporting and ownership
- Native keyword research per language, not a translated UK list.
- Written translation briefs per page, so translators optimise rather than only translate.
- Validated hreflang with crawl evidence across every template.
- Per-language reporting from month one.
- A documented process so the next language can be added without repeating the analysis.
Tools and platforms
- Ahrefs or Semrush (multi-market data)
- Google Search Console (per-property)
- Screaming Frog (hreflang validation)
- Native translator network
- Local SERP checking per market
Timeline
How long this actually takes
Research takes two to three weeks per language. Translation and implementation run two to three months for a mid-sized site. Ranking follows the usual curve of three to six months, and it starts from zero — a translated section does not inherit the English version’s rankings, and it needs its own links in that market to compete properly. The honest recommendation: launch one language properly rather than four badly. Four half-implemented languages produce four underperforming sections and quadruple the maintenance, and it is a mistake we see far more often than the reverse.
Pricing model
Project, then retainer
Research and technical implementation are fixed-price per language. Translation is quoted separately, since native translators are billed at their own rates and passed through at cost.
Questions
Multilingual SEO questions
Is machine translation good enough now?
It is good enough for a first draft and not good enough to publish unreviewed. Modern engines produce fluent text that still misses idiom, register and — critically — the phrases people actually search. Machine draft plus native review against a keyword brief is the workable combination.
Do we need separate keyword research for each language?
Yes, and it is the step most often skipped. Translating a UK keyword list gives you the correct translation of an English phrase, which is frequently not what native speakers type. This single omission accounts for more failed multilingual sites than translation quality does.
What is the difference between hreflang language and region codes?
A language code targets speakers of that language anywhere. A language-region code targets that language in a specific country. Use language alone where content is identical for all speakers; add the region where currency, law or terminology differ. Getting this wrong shows the wrong version to the wrong market.
Should URLs be translated too?
Where the platform allows it, yes. A translated slug is a modest relevance signal and reads better to a native speaker. It is not worth a risky migration on an existing site, but it is worth setting up correctly on a new one.
How many languages should we launch with?
One, done properly. Four half-implemented languages produce four underperforming sections and quadruple the maintenance. Launch one, learn what the process actually costs you, then add the next. This is the most common piece of advice we give on multilingual projects and the one most often resisted.
How is this different from international SEO?
International SEO is about targeting countries — structure, hreflang, geo-targeting. Multilingual is about targeting languages, and the hard part is linguistic rather than structural. They overlap heavily, and a business targeting Germany and Austria in German needs both.
Last reviewed 28 July 2026.
Tell us what you are trying to fix
A 20-minute call, no pitch deck. The Nexclick will tell you what we would do, roughly what it costs, and whether we are the right people for it.