Should you really add a canonical tag to every hreflang URL?

Why does Google emphasize this canonical + hreflang combination?

Hreflang informs Google about the relationships between language variants of the same content. Canonical, on the other hand, designates the version to index among similar or duplicate pages.

The problem arises when these two signals coexist without coherence. If a FR page has an hreflang pointing to an EN version without a clearly defined canonical, Google has to guess which URL deserves indexing. The result: indexing fluctuations, unintentional cannibalization, or even the de-indexing of entire variants.

The self-referential canonical (an URL that points to itself via rel="canonical") may seem redundant at first glance. But it removes any ambiguity: even if this FR page is linked to other language versions, it is this URL that must appear in the index for French queries.

Does this guideline apply to all multilingual scenarios?

Yes, whenever a page carries a hreflang attribute, it should also carry a canonical. Whether your architecture relies on subdomains (fr.example.com, en.example.com), subdirectories (/fr/, /en/), or distinct domains (.fr, .co.uk), the logic remains the same.

The self-referential canonical becomes your anchor: it states, "this page is the master version for this language." Without it, Google may interpret the hreflang relationships as duplication signals rather than legitimate variants.

What happens if we forget this canonical?

In most cases, Google manages. Its algorithms compensate for missing signals by analyzing content, detected language, and the server's geolocation. But this tolerance comes at a cost: extended crawl time, uncertain index consolidation, and unpredictable shifts between versions.

Some multilingual sites have functioned without an explicit canonical for years. But as soon as a better-structured competitor appears, or an algorithm update refines duplication detection, the flaws become visible: drops in rankings for certain languages, snippets mixing multiple versions, and loss of internal link signals.

• Hreflang alone is not enough to guarantee indexing of the correct language variant.
• A self-referential canonical clarifies which URL to index, even among international sibling pages.
• The absence of a canonical creates an ambiguity that Google sometimes resolves poorly, especially under crawl budget constraints.
• This recommendation applies to all types of multilingual architectures: subdomains, subdirectories, distinct domains.
• Combining both signals reduces the risks of cannibalization and accelerates consolidation into the correct index.

Is this guideline consistent with real-world observations?

Overall, yes. Audits of poorly indexed multilingual sites often reveal impeccable hreflang configurations but without a canonical, or worse, with contradictory canonicals pointing to another language. Google has publicly confirmed that it prioritizes the canonical signal over hreflang in case of conflict: if your FR page has a canonical pointing to the EN version, you explicitly tell Google to ignore the French variant.

Where it gets tricky: Google has never quantified the real impact of this absence. On sites with strong domain authority and a comfortable crawl budget, the omission often goes unnoticed. However, on average sites or poorly crawled sections (deep pages, long-tail categories), the lack of a self-referential canonical leads to documented partial de-indexing.

What nuances should be added to this rule?

The self-referential canonical is not a universal shield. If your content is genuinely duplicated across languages (low-quality machine translation, nearly identical pages), Google may still choose to index only one variant, canonical or not. The technical signal helps, but it does not replace editorial substance.

Another point: implementations via HTTP header rel="canonical" or HTML tag produce the same effect, but the HTML tag is preferred for its readability in audits. Some multilingual CMS (WPML, Polylang, Drupal i18n) automatically generate canonical + hreflang, others do not. Always check the actual code output, not just the plugin settings.

[To verify] Google has never published a detailed case outlining the exact arbitration between hreflang and canonical when both contradict each other in complex architectures (for example, a FR page with hreflang to EN, canonical to DE, and alternate media for mobile). Feedback suggests that canonical wins, but edge cases remain unclear.

In what situations does this rule not strictly apply?

Monolingual sites with regional variants (for example: en-US vs en-GB) where the content is nearly identical. Here, some prefer to canonical to a master version (en-US) and use hreflang for the variants, rather than self-referential canonicals everywhere. Risky, but defensible if crawl budget is tight.

Multilingual AMP or Web Stories: the canonical + hreflang logic applies, but with an additional layer (rel="amphtml" or rel="canonical" between AMP and standard HTML). Complexity skyrockets, and Google sometimes tolerates pragmatic simplifications. Again, insufficient Google documentation makes it hard to settle matters decisively.

What should be done concretely to align your site with this recommendation?

Start with a thorough audit of your multilingual pages. Extract all URLs carrying hreflang (via Screaming Frog crawl, Oncrawl, or Search Console export). For each, check for the presence of a self-referential canonical tag: <link rel="canonical" href="https://example.com/fr/page" /> on example.com/fr/page.

If your CMS automatically manages hreflang but forgets canonical, modify the templates or inject the tag via your tag manager (Google Tag Manager works, but prefer a server-side solution to avoid JS execution delay). For WordPress + WPML, activate the "self-referential canonical" option in advanced SEO settings.

What mistakes should be avoided during implementation?

Never point the canonical of a FR page to the EN version unless you explicitly want to de-index the French variant. Common mistake on international e-commerce sites: a FR product page carries a canonical to the .com (EN) URL because stock is managed there. Result: Google ignores the FR page, and you lose all organic French-speaking traffic.

Be careful with relative vs absolute canonicals. A relative canonical (/fr/page) can cause issues if your site serves on multiple protocols (http/https) or domains. Always use absolute URLs with HTTPS protocol: https://example.com/fr/page.

Avoid canonical chains: page A → canonical B → canonical C. Google usually follows the first redirection, but beyond that, it gives up. Each page should point directly to its final indexable version.

How to verify that my site is compliant after deployment?

Crawl post-deployment with a tool configured to extract canonical + hreflang simultaneously. Compare the two columns: each URL with hreflang should have a canonical identical to its URL (self-referential). CSV export, filter for inconsistencies, prioritize fixing pages generating organic traffic.

Monitor Search Console's "Coverage" and "Crawl" sections. The errors "Page indexed but blocked by canonical tag" or "Submitted URL not selected as canonical" often indicate hreflang/canonical conflicts. Cross-check with the "Links" reports to see if Google is properly indexing the expected variants.

• Extract all URLs carrying hreflang via crawl or Search Console export.
• Check for the presence of a self-referential canonical tag on each.
• Ensure that canonicals use absolute HTTPS URLs.
• Avoid any canonical pointing to a different language unless there's an explicit consolidation strategy.
• Test CMS templates or modify the code to automatically inject canonical.
• Re-crawl post-deployment and compare canonical vs hreflang in a cross table.