Most multilingual SEO services are translation projects wearing an SEO badge. The agency takes your best-performing pages, runs them through a translation workflow, publishes the output under /de/ or /es/, and declares the market open. Six months later the German pages rank for almost nothing, half of them compete with the English versions in the wrong search results, and the monthly report can't explain why.
If you're evaluating providers, the fastest way to separate serious from superficial is to look at the deliverables. This article breaks down what a real engagement includes, how the work is typically structured, and what actually drives the price. For the discipline itself, start with this primer on multilingual SEO. This piece is about what you should be buying.
What multilingual SEO services should include
Six deliverables form the core of any credible engagement. If a proposal is missing more than one of them, you're probably looking at a translation vendor with an SEO logo.
International site architecture
Before a single word gets translated, someone has to decide where each language lives. Subdirectories (yoursite.com/de/), subdomains (de.yoursite.com), and country-code domains (yoursite.de) distribute authority differently. Subdirectories inherit the strength of your existing domain, which is why they're the default recommendation for most companies expanding from a strong home market. Country-code domains send the clearest geographic signal but start from zero authority, each one a separate site you have to build up. Subdomains sit awkwardly in between.
This decision is cheap to make and brutally expensive to reverse, because reversing it means migrating every URL in every market. A serious provider makes it explicitly, in writing, with reasoning tied to your situation: how strong your current domain is, whether you're targeting languages or countries (Spanish for Spain is not the same project as Spanish for everyone), and what your CMS can actually support.
Architecture also covers the unglamorous plumbing: consistent URL patterns per locale, internal linking between language versions, and a language switcher that works with links rather than forced redirects. Locale detection that bounces every visitor to their "correct" version tends to bounce crawlers too, and Googlebot crawls mostly from US IP addresses. Google's own documentation on managing multi-regional sites covers why this matters.
Hreflang implementation, with QA
Hreflang annotations tell Google which URL is intended for which language and region, so the right version appears for the right searcher. It's a matching signal, not a ranking boost, and it's notoriously easy to get wrong. Every page must reference itself and all of its alternates, the annotations must be reciprocal or Google ignores them, and an x-default should catch everyone else. Pointing hreflang at redirected or noindexed URLs quietly breaks the whole cluster.
Two things distinguish a competent implementation. First, it's generated from templates in your CMS, never maintained by hand, because hand-maintained hreflang across hundreds of pages and five locales decays within months. Second, it's monitored after launch. Template changes regress hreflang all the time, and nobody notices until a market's traffic sags. Google's reference for localized versions is the spec your provider should be building against.
Per-language keyword research
A translated keyword list is not keyword research. It's the single most common shortcut in this industry, and it fails because search demand doesn't translate. People in different markets formulate queries differently, distribute their demand across the funnel differently, and face different competitors on the results page. German searchers use the English loanword for some product categories and the German term for others, and no dictionary will tell you which is which. Only the search data will.
The deliverable you should expect is a per-market keyword map: real queries in the local language, grouped by intent, mapped to a URL plan for that market. Crucially, it should also tell you what not to build, because some of your home-market pages target demand that simply doesn't exist elsewhere.
Transcreation briefs, not translation orders
Translation reproduces meaning. Transcreation reproduces intent and effect, in language a native speaker would actually type into a search box. The difference between the two lives in the brief, which is why the brief itself is a deliverable.
A proper transcreation brief gives the linguist the target query and its variants, the searcher intent behind it, what currently ranks in that market, the terminology decisions already made (loanword or local term, formal or informal address), required internal links, and guidance on structure and metadata. Hand a translator that brief and you get a page built to compete. Hand them a source URL and a deadline and you get a faithful copy of a page optimized for the wrong market. We've written more about where this line sits in our guide to SEO translation.
Local link building
Authority has a geography. A page competing in French search results benefits from citations in French-language publications that French readers actually visit, and the same logic holds in every market. This is the deliverable agencies most often quietly drop, because it's the hardest one: media relationships don't transfer across borders, outreach has to happen in the local language, and a pitch angle that works in the UK can land flat in Germany.
Expect a per-market link plan: which local publications and communities matter in each vertical, what content earns references there, and who is doing the outreach in which language. If the proposal is vague on this, the links will be too.
Per-locale reporting
A blended dashboard hides a failing market behind a winning one. Reporting should be built per locale from the start: separate Search Console views per folder or domain, rankings tracked against the actual local results pages rather than a generic global index, and conversions attributed per market. Each market also needs its own definition of success, because a mature market compounding at scale and a new market fighting for its first rankings should not be judged on the same curve.
What should be included, and usually isn't
Beyond the core six, a few deliverables separate the best engagements from the merely competent:
Demand validation before entry. Measuring whether a market actually searches for your category before you commit budget to it. Sometimes the honest finding is that it doesn't, or that it does under a completely different vocabulary. Hearing that early is worth a great deal.
A do-not-translate list. Pruning is a deliverable. Home-market content about local regulations, local events, or local pricing logic has no business being duplicated into six languages.
Post-launch technical monitoring. Hreflang, canonicals, and locale routing regress silently. Someone should be watching.
Native review inside the workflow. A native speaker checking terminology and tone before publication, not a post-hoc audit after rankings disappoint.
AI search coverage per language. AI Overviews and LLM-based assistants assemble answers per language, drawing on the sources that are visible and credible in each market's index. A brand absent from the German index is absent from German AI answers too. Your provider should be tracking this surface market by market, not just classic rankings.
How multilingual SEO engagements are structured
Most credible engagements follow the same arc, whatever the agency calls the phases.
Audit and market prioritization. Current technical state, existing international signals (you may already rank accidentally in markets you never targeted), demand and competition per candidate market, and a sequenced rollout plan. Markets are ranked, not treated as equals.
Foundation. Architecture decision, hreflang templating, locale routing, analytics and reporting split by market. This is done once, properly, before content scales.
Market waves. One or two markets launch first. The point of the first wave is to validate the pipeline: transcreation quality, indexation behavior, tracking accuracy. Then the pipeline scales to the remaining markets with the kinks already worked out. Launching eight markets simultaneously multiplies every mistake by eight.
Ongoing operation. Content production against the per-market keyword maps, local link acquisition, technical monitoring, and quarterly re-prioritization as data comes in. Markets that outperform get more investment; markets that stall get diagnosed rather than fed.
The structural question worth asking any provider: who owns native-language sign-off, and where does it sit in the workflow? If the answer is fuzzy, quality will be too.
What drives the cost of multilingual SEO services
Nobody can quote this work seriously without scoping it, but the drivers are consistent and worth understanding before you compare proposals:
Number of markets. The largest driver, though not linearly. Architecture, hreflang templating, and reporting infrastructure are built once and reused, so the first market carries setup costs the fifth doesn't. Research, content, and links, on the other hand, scale with every market you add.
Content volume and depth. Fifty product pages and a dozen guides per market is a different engagement from a five-page corporate site. Competitive verticals demand deeper content, which raises the transcreation load.
Competitive intensity per market. Entering a market where entrenched local players hold the results page costs more, mostly in content quality and link acquisition, than entering one where the category is underserved.
Site and CMS complexity. A platform with native localization support keeps implementation lean. A legacy CMS that fights hreflang and locale routing turns foundation work into an engineering project.
In-house native capacity. If your local teams can review terminology and validate copy, you're buying less from the agency. If the agency supplies native review for every market, that capacity is priced in.
One pattern to treat with suspicion: pricing that scales purely per translated word. That's how translation is sold, not SEO, and it tells you where the provider's head is.
What good execution looks like
The pattern that works is architecture plus per-market content, never a translation dump. When we ran this playbook with BMW Europe Moto, the official distributor for BMW motorcycle parts, the SEO and link-building program opened six new markets, generated 55,000+ organic visitors, and consolidated the site's Domain Rating at 70. For Gadero, a European retailer of outdoor wood products, non-brand organic traffic grew 105% and SEO-attributed revenue rose 59%.
Neither result came from clever tricks. They came from doing the deliverables above in the right order and holding each market to its own numbers.
If you're still comparing providers, this breakdown of how to choose a multilingual SEO agency covers the evaluation criteria in more depth.
Frequently asked questions
What's the difference between multilingual SEO services and translation services?
Translation services reproduce your content in another language and are priced per word. Multilingual SEO services build search visibility in other markets, which includes architecture, keyword research, hreflang, local links, and reporting, with transcreated content as one component among several. If the proposal only talks about words, it's translation.
Is multilingual SEO the same as international SEO?
They overlap but aren't identical. Multilingual SEO targets languages; international SEO targets countries. A Swiss company publishing in German, French, and Italian for one country is doing multilingual SEO. A UK brand expanding into the US and Australia in English alone is doing international SEO. Most real programs are both at once, and the architecture decision has to account for which one you're actually doing.
Should every page be translated into every language?
No. Each market gets the pages its search demand justifies, which usually means a core of shared commercial pages plus market-specific content, minus home-market pages with no local relevance. A do-not-translate decision is as valuable as a translation order, because every page you localize carries ongoing maintenance cost across every locale.
Ready to make more than one language pay its way?
Multilingual SEO rewards teams that treat every market as a first-class project and punishes everyone else with a quiet, expensive plateau. If you'd rather skip the plateau, talk to our team or see how our SEO agency approaches international growth.
Founder and CEO of Junto
Founder & CEO of Junto, Étienne has been an entrepreneur and digital marketing consultant for over 15 years. An expert in Paid Media, SEO, Data, Automation, AI, Growth and Performance, he helps ambitious companies build high-impact growth strategies — generating lasting results and helping brands move forward in a constantly evolving digital environment.
