Multilingual SEO: hreflang, Duplicate Content, and What Breaks

International SEO has a reputation for being complicated. In practice, the same handful of implementation mistakes account for most of the traffic that never arrives.

By Pierre Subeh, published June 4, 2026, 6 minute read

International SEO has a reputation for complexity that it only partly deserves. The strategy is genuinely hard. The implementation is a short list of things that are easy to get wrong and straightforward to fix once you can see them.

After a decade of untangling these setups, I can say that most lost international traffic traces back to four causes: the wrong URL structure, broken hreflang, treating translated pages as duplicates, and localizing the words while ignoring the query.

Pick a URL Structure and Stop Relitigating It

Three options, and the tradeoffs are well understood.

Country code domains such as example.de give the clearest geographic signal and the cleanest legal separation. They also mean building authority from scratch for every market, which is a serious cost that teams routinely underestimate.

Subdirectories such as example.com/de/ inherit the authority of the main domain, which is usually decisive for a company entering its second or third market. This is the right default for most organizations.

Subdomains such as de.example.com sit between the two and, in my experience, capture the disadvantages of both more often than the advantages.

Unless you have a legal or infrastructure reason that forces your hand, use subdirectories. The important thing is to choose deliberately and then leave it alone, because migrating structures later costs more than the original decision was ever worth.

One rule that matters more than the choice: separate language from country when they diverge. Spanish for Spain and Spanish for Mexico are different markets with different queries, different currencies, and different competitors. If both matter to you, they need separate URLs, not one page hedging between them.

hreflang, and the Four Ways It Breaks

The hreflang annotation tells search engines that several pages are the same content for different audiences. The concept is simple. The implementation fails in four predictable ways.

Missing return links. Every page in a set must reference every other page in the set, including itself. If the English page points to the German page but the German page does not point back, the annotation is ignored. This is the single most common failure and it is invisible without a crawl that checks reciprocity.

Wrong codes. The language value is an ISO 639-1 code and the optional region is an ISO 3166-1 Alpha 2 code. The United Kingdom is en-gb, not en-uk. Chinese written in simplified script for mainland China is zh-Hans-cn rather than an invented shorthand. A malformed value is discarded silently.

Pointing at redirects or missing pages. Every URL in the set has to return a 200. Annotations that point at a redirect chain, a 404, or a page carrying a canonical to somewhere else are dropped.

Fighting your own canonical. A translated page must canonicalize to itself. Teams sometimes point every language version at the English original to avoid a duplicate content worry, which removes the translations from the index entirely. The canonical says this is the authoritative version of this page. The hreflang says here are the other language versions. They are different statements and they should not contradict each other.

Add a self referencing annotation on every page. It is required, it is frequently forgotten, and its absence invalidates the set.

Translated Pages Are Not Duplicate Content

This worry causes more damage than the problem it imagines. Content translated into another language is not duplicate content. Search engines handle this case explicitly, which is what hreflang is for.

The real duplication risk lives elsewhere: multiple regional variants of the same language with near identical text. A Spanish page for Spain, Mexico, Argentina, and Colombia that differ only in currency symbol are competing with each other for the same queries.

The answer is not to consolidate them. It is to make them genuinely different, because if there is no real difference then there was no reason to build four pages. Local pricing, local examples, local proof, local contact routes, and local search terms. If you cannot articulate what makes the Mexican page different from the Spanish one beyond the currency, build one Spanish page and target it well.

Localize the Query, Not Only the Copy

This is the strategic failure that survives a technically perfect implementation.

Keyword research does not translate. Running your English keyword list through a translation tool produces a list of phrases that are linguistically correct and commercially irrelevant, because people in the target market search using different words, different structures, and often a different language entirely.

In many markets, professional audiences search in English for technical topics and in the local language for everything else, and the split is specific to the vertical. In others, the dominant phrasing is a loan word rather than the formal local term, and the formal term you carefully translated to has almost no volume behind it.

Run native keyword research per market. Start from the local language, look at what actually ranks, read the pages that win, and note the vocabulary they use. Then write to that vocabulary. This is the difference between a translated site and a site that ranks.

Search intent also shifts. A query that is commercial in one market can be informational in another, depending on how mature the category is locally. A page built for a buyer will underperform against a query from someone still learning the category exists.

Signals Beyond the Markup

A few things quietly shape how well an international setup performs.

Server location and delivery speed still matter, and a site served entirely from one region will feel slow on the other side of the world. A content delivery network handles most of this.

Internal linking should generally stay within a language. Linking from a German page to an English page passes authority in a direction that does not help the German market and can confuse the language grouping.

Local backlinks are the strongest geographic signal available, and they are the part teams neglect because they are the hardest. A handful of links from real sites in the target market will move rankings more than another round of tag fixes.

Language switchers should be crawlable links rather than a script driven dropdown. If a crawler cannot see the path to your other language versions, it has to rely entirely on the annotation to discover them.

How to Audit This in an Afternoon

Crawl the site and confirm that every page in a language set returns 200, canonicalizes to itself, and carries a reciprocal set of annotations including a self reference. Validate every language and region code against the standards rather than against memory. Search Console will report hreflang errors directly, and its international targeting data is worth reading before you form a theory. Then check the strategy layer: pull the actual queries each market ranks for and ask whether they are the queries you intended.

Most of what you find will be in the first paragraph of that audit. The fixes are usually small. The traffic they unlock is not.

About the author

Pierre Subeh is Co-President of AMA Orlando and the founder and CEO of X Network, an SEO and paid marketing firm whose work spans Apple Music, Häagen-Dazs, and Pepsi. He is a Forbes 30 Under 30 honoree in marketing and advertising, a TEDx speaker, and a member of the PR and Media Council at the Council of Global Change, an independent international policy council that convenes at the United Nations. He writes about search, brand, and global growth at pierresubeh.com.

Topics: multilingual seo, hreflang, technical seo, international seo, duplicate content