9.3Intermediate8 min

Hreflang: Correctly Tagging Multilingual Websites

Lucas Blochberger·7 April 2026·Updated 12 June 2026

Definition

Hreflang is an HTML attribute (link rel=alternate hreflang) that tells search engines the language version and regional assignment of different page versions to display the correct version in respective local search results.

Key Takeaways

✓Hreflang is only supported by Google and Yandex; Bing uses the content-language meta tag instead, and AI search systems ignore hreflang completely.
✓Self-reference and bidirectional links (reciprocity) are mandatory, as are fully qualified URLs including protocol; Google ignores one-sided or incomplete sets.
✓Defining x-default for the fallback is the most frequently forgotten element: according to Ahrefs, it was missing on 56.3 percent of domains.
✓Hreflang is error-prone: 67 percent of 374,756 examined domains had at least one issue, most commonly missing x-default, missing reciprocity, and links to redirects.
✓For DACH sites, hreflang solves SERP assignment, not cannibalization: de-AT, de-DE, and de-CH must be differentiated in content (vocabulary, prices/currency, legal).
✓The three implementation methods (HTML, HTTP header, XML sitemap) are equivalent from Google's perspective; for large and programmatic sites, the sitemap is the most maintainable.
✓Avoid automatic IP redirects as they prevent Googlebot from crawling all versions; instead, offer a visible language switcher.

Hreflang is technically simple, but error-prone in practice.

Implementation

Three options: HTML link tags in the head (most common), HTTP headers (for non-HTML resources like PDFs), XML sitemap (scalable for large websites). All three are equivalent — one is sufficient.

DACH Implementation

Typical setup: hreflang=de for general German, hreflang=de-AT for Austrian content, hreflang=de-CH for Swiss content, hreflang=en for English, hreflang=sk for Slovak, hreflang=x-default for fallback.

The 5 Most Common Hreflang Errors

Missing bidirectional references (page A links to B, but B doesn't link back to A). Incorrect language/country codes. Forgotten self-referencing. Hreflang pointing to non-canonical URLs. Contradiction between hreflang and canonical.

Hreflang and AI Search

AI systems ignore hreflang completely. ChatGPT selects sources independently of the language version. Consequence: content must be independently high-quality and geo-optimized in every target language.

Data & Statistics

67% von 374.756 untersuchten Domains mit hreflang hatten mindestens ein Problem; 56,3% fehlte x-default; 18% fehlten selbstreferenzierende Tags; 16,9% verwiesen auf weitergeleitete/defekte Seiten

Ahrefs - Hreflang Study (374,756 Domains) (2023)

31,02% widerspruechliche hreflang-Direktiven; 47,95% ohne x-default; 16,04% ohne selbstreferenzierende Tags; 18.786 Sites analysiert

Search Engine Land - Studie von Dan Taylor (SALT.agency) in Partnerschaft mit NerdyData (2023)

9,6% der Sites nutzen hreflang-Tags auf Desktop, 8,9% auf Mobile

Web Almanac 2022 (HTTP Archive), SEO-Kapitel (2022)

Die drei hreflang-Implementierungsmethoden (HTML-Tags, HTTP-Header, Sitemap) sind aus Sicht von Google gleichwertig; URLs muessen voll qualifiziert sein

Google Search Central - Tell Google about localized versions of your page (2025)

76% bevorzugen Produkte mit Informationen in ihrer Muttersprache; 40% kaufen nie in anderssprachigen Shops (8.709 Konsumenten, 29 Laender)

CSA Research / Kantar - Can't Read, Won't Buy - B2C (via Newswire) (2020)

8,69 Mio. Internetnutzer in Oesterreich; Internetdurchdringung 95,3% (Januar 2025)

DataReportal - Digital 2025: Austria (2025)

Ungueltige HTML-Elemente im head: 10,1% Desktop / 10,3% Mobile

Web Almanac 2025 (HTTP Archive), SEO-Kapitel (2025)

FAQ

What does de-AT mean in hreflang?

de-AT consists of the language code according to ISO 639-1 (de for German) and the country code according to ISO 3166-1 Alpha-2 (AT for Austria). The value signals to Google that the page is intended for German-speaking users in Austria. The order is fixed: first the language, then the region.

Which search engines support hreflang?

Hreflang is only evaluated by Google and Yandex. Bing does not use the attribute and relies instead on the content-language meta tag as well as server and domain signals. AI search systems do not use hreflang.

Do I need x-default with hreflang?

x-default is technically optional but strongly recommended. It defines the fallback version for users whose language or region does not match any hreflang entry. The missing x-default tag is the most common hreflang error according to studies: Ahrefs found it missing on 56.3 percent of examined domains.

What is the most common hreflang error?

The most common error is the missing x-default tag, followed by missing reciprocity and missing self-referencing tags. According to an Ahrefs study of 374,756 domains, 67 percent had at least one hreflang issue. Links to redirects or broken pages and conflicts with the canonical tag are other typical error sources.

ccTLD or subfolder for DACH sites?

A ccTLD like .at sends the strongest geo-signal and creates local trust, but requires triple effort and does not inherit domain authority between domains. A subfolder like /de-at/ consolidates all authority on one domain, is easy to maintain, and scalable. For most B2B projects in the DACH region, the subfolder is the most pragmatic choice; the weaker geo-signal can be compensated through hreflang and content.

How do I avoid keyword cannibalization between de-AT, de-DE, and de-CH?

Hreflang solves SERP assignment, not content differentiation. Make each version content-independent instead of translating one-to-one: local vocabulary (Jänner instead of Januar), correct currency (EUR for at/de, CHF for ch), country-specific imprint and legal. Hreflang signals the separation; the content must justify it.

Which hreflang implementation method should I choose?

There are three equivalent methods: HTML link tags in the head, HTTP headers, and XML sitemap. For small to medium sites, HTML tags are suitable; HTTP headers are necessary for non-HTML documents like PDFs, and the XML sitemap is the cleanest solution for large and programmatic sites. Google recommends choosing one method, as parallel implementations offer no advantage and complicate maintenance.

How does your website perform?

Get a free, AI-powered SEO report of your website by email – technical SEO, on-page, keywords & competitors. No obligation.

Get a free SEO audit →

Previous← Domain Strategy: ccTLD vs Subfolder vs Subdomain NextContent Localization: More Than Just Translation →