International SEO: hreflang, canonicals, and translated pages
How to keep multilingual URLs clean, prevent duplicate signals, and send users to the right language version.
International SEO starts with a boring but critical contract: every localized URL must tell search engines which language or region it serves, which URL is canonical, and which sibling versions exist. Without that contract, translated pages can compete with each other or the wrong language can rank for the wrong audience.
The hreflang contract
Canonical rules for translations
A translated page is not a duplicate of the original. Its canonical should normally point to itself, not to the English version. Canonical tags consolidate duplicate signals; hreflang connects equivalent localized choices. Mixing those jobs is one of the fastest ways to make a translated section disappear from search.
Seora audits alternate links, self-canonicals, sitemap language coverage, and locale routing. It flags broken reciprocal hreflang pairs before they turn into ranking confusion.
The content still has to be genuinely localized. Translating words is not enough if pricing, examples, support details, screenshots, or legal copy remain built for another market.
Sources
Put this into practice
Run a free Seora audit and get the exact fixes for your site — performance, AI readiness, internal links and more.
Keep reading
All articlesRobots.txt, noindex, and AI crawlers: what each control actually does
A clean guide to crawl blocks, index controls, and crawler allowlists so important SEO pages stay reachable.
Programmatic SEO without scaled-content spam
How to build template-driven pages that deserve to rank: unique data, editorial controls, and quality gates before publishing.