hreflang validator.
The hreflang validator reads the language links in pasted markup and flags the six most common mistakes.
How to use
- Paste the page’s <head> section, or just the link elements
- Press Validate
- Findings come out as a numbered list — if it is clean, it says so too
The tool runs entirely in your browser; it sends no data and writes no cookies.
The check runs through the browser’s own parser: the pasted text is read as a document and the links carrying `rel="alternate"` are collected. Parsing markup with regular expressions is fragile — quote style, attribute order and whitespace all change the outcome.
Why these six
The underscore is the most common typo: because it is habitual in file names and programming languages, people write `tr_TR`, yet a language tag wants a hyphen. The second frequent mistake is format — taking a country code for a language, for instance. The third is a relative address: hreflang wants an absolute one, because a link may point to a different domain.
The remaining three concern the integrity of the set. Declaring the same tag twice leaves the engine uncertain. A missing self-reference makes the set one-way — Google ignores declarations that are not reciprocal. And no x-default means leaving no default for a visitor who matches nothing. We have written up the full list of mistakes in a separate guide.
Why the page is not fetched
The tool does not take an address and download the page, and that is a decision rather than a shortcoming. This site’s content security policy is closed to external requests; even if it were not, sending an address the user is auditing to a server would contradict the privacy statement. The input is pasted, the check runs in the browser, and no data leaves the page.
In practice the difference is small: viewing a page’s source and copying the head section takes a few seconds. In exchange, the check also works on pages that require a login or are not published yet.
What it does not validate
The tag format check is a practical gate, not the whole of BCP 47: the standard rests on a large registry of language, script and region subtags, and full validation requires that table. The tool catches the three most frequent mistakes; a tag it does not recognise is not always wrong. Reciprocity, too, cannot be verified from a single page — whether the target pages point back can only be seen by auditing them as well. We describe the concept itself and the tag standard separately.