A large set of hreflang annotations is not evidence that a multilingual setup is correct. The right pages must be related, and the published output must preserve those relationships. One inspected URL cannot establish that every template works. Start with a content pair, then expand to examples produced by the same system to identify both the error and its source.
Write the expected relationship first
List each content identity and its published language URLs. Keep unpublished translations, removed pages and resources with different purposes distinct. This becomes the reference against which a crawl or manual inspection can be assessed.
A hypothetical service guide pointing to an English category page may have syntactically valid annotations but the wrong editorial relationship. Content equivalence therefore needs review alongside automated checks.
Validate the basic conditions on a sample pair
Google's guidance describes self and alternate references, reciprocal relationships and fully qualified URLs. Language and regional codes must also use supported forms. Check these conditions on the sample, then inspect how the output is generated.
The HTML standard defines hreflang as language information for a link. Editorial review of the actual page remains necessary. An annotation cannot correct body copy published in the wrong language or an inaccurate translation.
Make each finding reproducible
Record the affected URL, expected counterpart, observed value and component producing it. The implementation team then receives a concrete problem rather than an unexplained warning.
| Finding | Investigation question | Likely owner |
|---|---|---|
| Wrong content counterpart | Is the editorial mapping correct? | Editorial team |
| Missing relationship on one version | Do templates use the same source? | Development team |
| Old destination URL | Did the change reach all versions? | SEO and development |
| Annotation and content disagree | Is the correct translation published? | Language editor |
Test templates and content changes
Homepages, services, categories and articles may use different templates. Fixing one does not automatically fix the others. Test how relationships update when content is created, a translation is published, a slug changes or a page is removed.
Reinspect sample pairs after the change and compare them with the original findings. Search-result movement should not replace implementation checks, and correct annotations do not guarantee a particular version in every search. A durable solution connects editorial mapping and technical output to a shared source of truth.