Translated content targets terms nobody searches and reads foreign to the people who do. Multilingual SEO that works runs native keyword discovery per language, transcreation by in-market writers under one strategic framework, and hreflang engineered as code so publishing schedules can't silently break it.
A short introduction, your site URL, and what you’re trying to achieve. If it’s a fit, we’ll book a 30-minute call.
Free £500 SEO audit included with any web dev or SEO package · no card required
Multilingual SEO fails in a predictable order: someone translates the site, rankings do not come, more translation is bought, hreflang breaks under the volume, and eventually a cleanup project unwinds the lot. The alternative costs more per page and less per outcome, because it starts where the demand actually is, inside each language.
Concepts do not map cleanly across languages: German compounds collapse English phrases into single searchable nouns, French precision splits intent English blurs, Arabic search mixes Modern Standard, dialect and transliteration. Native discovery, seeded in-language by speakers who read SERP intent correctly, produces demand maps translated research cannot approximate, then maps back to one strategic framework so markets stay comparable.
Transcreation gives native writers the intent, evidence and conversion goal, then lets them write for their audience: their examples, their references, their register (formality norms differ sharply, Sie/du, vous/tu, Gulf-Arabic courtesy conventions). Governance keeps it consistent at scale: content classification (transcreate / localise / keep source), native reviewers with rejection authority, glossaries and tone guides per locale, and machine translation banned from commercial pages.
Multilingual sites break hreflang because teams publish asynchronously in different languages: annotations drift, return tags vanish, canonicals cross locales. The durable pattern is generation from a single locale registry, emission across pages and sitemaps, CI validation blocking releases that break reciprocity, and scheduled Search Console reconciliation. Same-language variants (en-GB/US, es-ES/MX, fr-FR/CA) get designed here too, they cannibalise fastest and are missed most often.
Language priority follows data: impressions by country and language, competitor coverage, prize versus cost, then one language done excellently before the next. Non-Latin scripts add genuine technical work rather than another content cell: RTL layouts, bidirectional text with Latin brand names, font weight in Arabic script, and register decisions that shape conversion. We run bilingual EN/AR programmes for Gulf clients (see Gulf practice) and European sequences through the EU hub and country pages like Germany. Architecture-level decisions live in international SEO.
Multilingual SEO · transcreation · hreflang engineering
Two failure modes. Technically, keyword research does not translate: demand patterns, phrasing and even concept boundaries differ per language (German compounds, Arabic dialect variance, French precision), so a translated page targets terms nobody searches while missing the ones they do. Commercially, native readers detect translation instantly and trust drops. Add Google's policies on scaled low-value content, and machine-translated pages at volume become a liability rather than an asset.
Transcreation rebuilds the argument for the target market: a native writer takes the strategic intent, the proof points and the conversion goal, then writes in-market, choosing the phrasing, examples, references and register the audience expects. It costs more per page and produces pages that can actually rank and convert, which makes it cheaper per outcome, usually by a wide margin.
Through native specialists with a documented process: native keyword discovery (seeded in-language, not translated), SERP analysis by native speakers reading intent correctly, mapping back to a shared strategic framework, and validation against local search behaviour. Senior strategy stays in English; language-specific judgement is made by people who live in the language. Pretending otherwise is the industry's most common multilingual failure.
As engineering, not annotation: a single locale registry generating tags consistently across pages and sitemaps, CI validation so releases cannot silently break reciprocity, and scheduled reconciliation against Search Console. Multilingual sites break hreflang constantly because different teams publish in different languages on different schedules; automation is the only durable answer.
By evidence rather than ambition: existing Search Console impressions by country and language, competitor language coverage, and market prize versus content cost. Typical European sequences run German, French, Spanish, Italian, Dutch, but the honest sequence is your data's, and one excellent language beats four thin ones every time. Gulf clients add Arabic with its own considerations, RTL rendering, dialect and register choices, bilingual EN/AR architecture.
Distinct disciplines rather than another language cell: RTL layout correctness, bidirectional text handling with Latin brand names, font rendering, dialect and register decisions (Modern Standard versus Gulf usage), and search behaviour that mixes Arabic with transliterated terms. We run bilingual EN/AR programmes for Gulf clients, and treat script-specific technical work as first-class rather than a translation afterthought.
Architecture and governance design from £2,500. Per-language programmes from £1,200/month each (strategy, native keyword research, transcreation management, in-language optimisation), with content produced per piece by native specialists. Machine-translation cleanup projects, unwinding auto-translated pages that damaged a domain, scoped after audit, and more common than anyone would like.
Summary: Multilingual SEO succeeds through native demand discovery, transcreation governed against decay, hreflang engineered to survive asynchronous publishing, evidence-ordered language sequencing, and first-class handling of non-Latin scripts. Translation at volume is a liability; one excellent language beats four thin ones. Related: international architecture, European practice, bilingual Gulf.
Send your site and target languages. You'll get a read on hreflang health, translation exposure and sequencing, from Syed, within one working day.
A short introduction, your site URL, and what you’re trying to achieve. If it’s a fit, we’ll book a 30-minute call.
Free £500 SEO audit included with any web dev or SEO package · no card required