Hreflang Tags Explained: 5 Essential Rules for Global SEO

Hey! If you run a site in more than one language or ship separate pages for different countries, you have probably typed a block of link tags into your head section and hoped for the best. This one is for you.

Picture this: your French page outranks your English page in France, except it is showing up for searchers in Canada instead, and your support inbox is full of confused messages. That mismatch is usually not a translation problem. It is a hreflang tags problem, and it is one of the most common ways international sites quietly lose traffic.

Hreflang tags are a small piece of markup that tell search engines which language and which country a page is meant for. According to Google's own documentation, the annotation exists so Google can point a searcher in Tokyo to the Japanese version of a page and a searcher in Madrid to the Spanish one, even when both pages are technically duplicates of each other.

Without hreflang tags, search engines are left guessing. They might index only one language, show the wrong country version in search results, or treat your localized pages as duplicate content competing against each other. That last outcome is the quiet killer, because it dilutes ranking signals across pages that should be working together.

Per Wikipedia's overview of the attribute, hreflang covers three real world scenarios: the same country with different languages, different countries with the same language, and different countries with different languages. Most sites only plan for one of these and miss the other two.

The scale of the problem is bigger than most teams expect. An audit of 374,756 domains found that 67 percent of sites using hreflang tags carried at least one unresolved error, meaning broken international targeting is closer to the norm than the exception. If your site has never been audited for this specifically, assume there is at least one mismatched pair somewhere in your clusters.

This is the part of international SEO that trips up even experienced teams. Google states it plainly: if page A points to page B as an alternate, page B must point back to page A, or the whole pair is ignored. One direction is not enough.

Ahrefs compares this to confirming a relationship status, a declaration only counts once both sides agree. A page that lists five language versions but is not listed back on all five is effectively invisible to the system, even though the markup looks complete on the page you are staring at.

This return tag requirement is also why mismatches are so easy to create at scale. Add a new regional page, forget to update the other four, and you have broken a cluster that was working fine yesterday.

Alongside language and country codes, hreflang tags support one special value: x-default. Google describes it as the version shown when none of your other annotations match a visitor's browser settings, think of a language picker page or a global homepage that serves everyone before they choose a locale.

x-default is not meant for your biggest market's homepage or your most heavily optimized page. It is meant for the fallback a visitor sees when nothing else fits, which is a narrower job than most teams assign it.

There are three accepted places to declare hreflang tags, and Google treats them as functionally equal. You pick based on your site's shape, not based on which one ranks better.

The first is a link element inside the HTML head of each page, the most common choice for small to mid sized sites because it is easy to template. The second is an HTTP header, which matters for files that have no head section at all, like PDFs. The third is an XML sitemap entry, which Screaming Frog recommends for large multilingual catalogs, since one sitemap file can describe hundreds of language clusters without bloating every page's markup.

If your site already maintains a sitemap for crawling, our guide on the difference between an HTML sitemap and an XML sitemap is a useful companion read before you decide where this annotation should live.

Now the part you came for. These five mistakes show up again and again in audits, and they are the reason this markup has a reputation for being fragile.

A study of 20,000 multilingual sites by Semrush found that 58 percent had hreflang conflicts somewhere in their source code, and missing self references showed up in 96 percent of the sites that had any source code conflict at all. These are not rare edge cases, they are the default state of most multilingual sites that have never been audited.

As Helen Pollitt, Head of SEO at Getty Images, put it in a Search Engine Journal interview, getting hreflang tags right matters because "search engines being confused over which version of a page to show" is exactly the outcome you are trying to avoid.

You cannot eyeball every pair across a hundred pages and trust what you see. A crawler is the only realistic way to confirm every pair actually links back correctly.

Tools built for this job flag the same handful of problems every time: broken or redirected targets, missing reciprocal links, duplicate language codes pointing at more than one URL, and annotations that contradict a noindex directive on the target page. Run that check after every content migration, not just once at launch, since a single URL change can quietly snap a whole cluster.

It also helps to confirm your overall international site structure before you fix individual tags, since this annotation layered on top of an inconsistent folder or subdomain structure tend to break again the next time someone reorganizes the site.

Hreflang tags are not a set it and forget it task. Every new locale, every page rename, and every migration is a chance to break a reciprocal link that was fine the week before.

Build a habit of treating language clusters as a single unit. When one page in the group moves or redirects, update every other member of that cluster in the same change, not as a follow up ticket that might get skipped.

Keep a simple reference sheet of your valid language and region codes so nobody on the team has to remember whether it is pt-BR or br-PT. Small clerical slips are the most common reason these tags fail, far more often than anything structurally complicated.