International SEO: Why Translated Pages Alone Do Not Bring Traffic
Every few months a client arrives with the same story. They spent a substantial sum translating their website into four languages, the pages went live, and nothing happened. Traffic from the new markets stayed flat. The assumption underneath that disappointment is that translation and international SEO are the same project. They are not, and the gap between them is where most of the money disappears.
Search demand does not translate
The first hard lesson is that your English keyword list is not a source document. It is a record of how one market phrases its problems. Translate it literally and you get grammatically correct phrases that nobody types. A British audience searching for a solicitor is a German audience searching for a Rechtsanwalt, but the search intent behind each query, the volume, the competition and the type of page Google chooses to rank can all differ enough to change your whole content plan.
The fix is unglamorous. Run fresh keyword research in each target language, using native speakers or at minimum a tool with reliable local data, then map the results back to your existing pages. Sometimes a page that ranks well in English has no equivalent demand abroad and should not be translated at all. Sometimes a market wants an article you have never written.
Where an international SEO strategy usually breaks
Assuming the words are right, the second failure point is structural. Search engines need to be told which version of a page belongs to which audience, and that signalling has to be consistent across the whole site. Hreflang annotations are the standard mechanism, and Google Search Central documents the rules for localized versions of a page in reasonable detail. The mechanism is simple to describe and remarkably easy to implement badly.
Common faults include annotations that point to a page which does not return them, language codes that do not exist, tags that reference redirected or canonicalised URLs, and self references quietly missing on half the site. A careful write-up of how multilingual SEO tends to fail in the markup before it fails in the copy covers this ground well, and it matches what I see in audits. The copy usually gets the attention because it is visible. The tags decide whether anyone reads it.
Choosing a URL structure you can live with
Country code domains, subdomains and subfolders all work. What matters is picking one and committing, because migrating later costs more than the original build. Subfolders on a single domain are the pragmatic default for most mid sized businesses: authority stays consolidated, maintenance is simpler, and hosting stays in one place. Separate country domains make sense when a market needs genuinely different legal entities, pricing or product ranges, and when you have the resources to build authority for each domain independently.
Whichever route you take, keep the pattern predictable. A market that lives at one URL shape while another lives at a second shape produces exactly the kind of inconsistency that breaks annotations at scale. The technical background of hreflang is worth understanding before the decision is made rather than after.
Translating a page and ranking a page are two entirely different projects, which is the hard lesson behind most international rollouts. Underneath the hreflang tags and localized URLs, the fundamentals of search engine optimization still decide whether any of it earns traffic. Teams that skip that layer end up with a beautifully translated site nobody ever finds.
Automatic redirects are not the shortcut they look like
Detecting a visitor's location and forcing them to a matching version feels helpful. In practice it frustrates real users, an expatriate in Madrid who wants English, a buyer researching another market, and it can stop crawlers from ever seeing the versions you want indexed. Offer the choice instead. A clear, persistent language selector and a dismissible suggestion banner do the same job without hiding pages from the index.
Plugins solve part of the problem
Translation plugins have improved considerably, and for small sites they can be a sensible starting point, particularly where machine output is reviewed by a human. They vary widely in how they handle URLs, metadata and annotations, which is the part that actually determines search performance. PoliLingua's comparison guide to website translation plugins is a useful reference if you are choosing between them, because it looks past the translation quality claims at the plumbing.
What no plugin does is decide what deserves translating. That judgement stays with you, and it is where the return on an international SEO strategy is actually made.
What to do first if the budget is small
Pick one market rather than four. Translate and localise the twenty pages that already earn the most, not the whole site. Do proper keyword research for that language, set up hreflang correctly between just those pages and their English originals, and give it two quarters. A single market done properly produces evidence. Four markets done thinly produce a spreadsheet full of flat lines and no way to tell whether the words, the markup or the market was the problem.
Multilingual website SEO rewards patience more than budget. The businesses I have seen succeed at it were not the ones who launched in the most languages. They were the ones who fixed the technical foundation once, wrote for the market rather than translating at it, and then kept going after the first quarter looked unimpressive.