How do you set canonical + hreflang correctly in a multilingual Next.js site?

Conclusion

The most damaging i18n SEO mistake is cross-language canonicalization:

• /ja/... declares canonical to /en/... (or the reverse)

That often causes the “non-canonical” language to never index.

The safe approach is:

1. normalize URL format (slash, casing, params)
2. use self-referencing canonicals for each language page
3. add reciprocal hreflang alternates for every indexable page
4. verify via page source + sitemap + Search Console URL Inspection

Explanation

Canonical and hreflang are signals. Google may choose different canonicals when your signals conflict.

In Next.js i18n, conflicts usually come from:

• multiple URL variants (trailing slash, rewrites)
• canonical built from a different URL builder than sitemap/hreflang
• locale routing creating duplicates

Fixing this is less about “SEO tricks” and more about consistent URL normalization.

Practical Guide

Step 1: define your intended URL shape (one true form)

Decide:

• trailing slash on/off
• lowercase vs mixed case
• which query params are allowed

Decision rule:

• Prefer redirects for normalization so Google discovers fewer variants.

Step 2: use self-referencing canonicals by default

Most pages should canonicalize to themselves, per locale.

Only canonicalize all locales into one “master” language if you intentionally want to collapse content.

Step 3: never canonicalize across languages by accident

Bad:

• /ja/product/ → canonical /en/product/

Outcome:

• JA clusters as duplicate and may never index.

Step 4: add hreflang alternates for every indexable page

Each page lists all locale variants (and optionally x-default).

Rules:

• alternates must be reciprocal (A points to B, B points to A)
• do not include non-indexable pages

Step 5: keep URL formats identical across canonical/hreflang/sitemap

Common drift:

• canonical has no trailing slash, hreflang has a trailing slash
• hreflang uses a parameterized URL

Decision rule:

• Centralize URL building into one helper and reuse it for canonical/hreflang/sitemap.

Step 6: handle parameters intentionally

• tracking params (utm_*): canonical must point to the clean URL
• functional params (?page=2): decide whether this is indexable or not

Step 7: validate (fast)

Pick 3 representative pages (home, category, article), then:

• view source:
  • <link rel="canonical" ...>
  • <link rel="alternate" hreflang="..." ...> for all locales
• Search Console URL Inspection:
  • user-declared canonical vs Google-selected canonical
  • indexing status per locale

Pitfalls

• middleware rewrites create multiple discoverable URLs (rewrite instead of redirect)
• locale pages are near-duplicates (Google clusters them)
• trailingSlash setting differs from canonical/hreflang builder
• hreflang is non-reciprocal or incomplete

Checklist

• [ ] One true URL form is defined (slash/casing/params)
• [ ] Normalization uses redirects where possible
• [ ] Canonicals are self-referencing per locale
• [ ] No cross-language canonicals exist
• [ ] hreflang alternates exist for every indexable page
• [ ] hreflang alternates are reciprocal
• [ ] canonical/hreflang/sitemap URL formats match exactly
• [ ] Tracking params canonicalize to clean URLs
• [ ] Functional params have an explicit indexing decision
• [ ] Validation is done via source + sitemap + GSC URL Inspection

FAQ

Q1. Can I canonicalize all languages to one “master” page?

Only if you intentionally want to collapse locales. For true multilingual content, self-referencing canonicals per locale are safer.

Q2. Do I need x-default?

It’s optional but often helpful for a generic or language selector page. The key requirement is reciprocal alternates across real locales.

Q3. Why does Google choose a different canonical than mine?

Because your signals conflict: inconsistent URL formats, duplicates from rewrites, or near-duplicate locale content. Fix consistency first.

Internal links

• Parent hub: Indexing: start here
• Related:
  • Google Search Console setup (Next.js + Vercel)
  • Discovered but not indexed checklist
  • Crawled, currently not indexed fixes

References

• Google: Canonicalization and duplicate URLs: https://developers.google.com/search/docs/crawling-indexing/consolidate-duplicate-urls
• Google: Localized versions (hreflang): https://developers.google.com/search/docs/specialty/international/localized-versions

Disclaimer

Indexing is probabilistic. You can improve signals, but you cannot force Google to index every URL.