International SEO checklist: hreflang, canonicals and site structure

URL structure

Subdirectories such as example.com/de/ inherit the domain's authority, are cheap to run and are the right default for most teams. Subdomains isolate the markets and add operational overhead. Country-code domains give the strongest geo signal and the highest cost, and only make sense with a real local operation behind them.

Whatever you choose, apply it consistently. Mixed structures create duplicate clusters and confuse both crawlers and analytics.

Hreflang, done correctly

Hreflang tells search engines which version serves which audience. It fails in predictable ways, and the failures are all mechanical.

• Every page in the cluster lists every other page, including itself.
• The references are reciprocal: if A points to B, B must point to A.
• Use language codes, optionally with a region: de, de-AT, es-MX.
• Include an x-default for the fallback version.
• Use absolute URLs, and point at the final, indexable URL, never a redirect.

Canonicals and duplicates

Each language version canonicalises to itself. Pointing a translated page at the English original tells search engines the translation should not be indexed, which is a surprisingly common own goal.

Watch for near-duplicate variants within one language: es-ES and es-MX pages with identical text will compete. Either differentiate the content genuinely or serve one page with hreflang regional targeting.

Metadata and internal links

Titles and meta descriptions carry keywords and click-through rate, so they get written natively from the winning target-market term, not translated from English. Length limits also differ in practice: German titles run longer and truncate sooner.

Internal links must resolve within the language. A German article linking to an English pricing page hands the visitor a language switch and hands the crawler a mixed signal.

Pre-launch QA pass

Run this list once before launch and once two weeks after.

• Crawl the site and confirm no hreflang errors and no cross-language canonicals.
• Confirm html lang matches the content language on every template.
• Check that localised sitemaps include every language version.
• Confirm the language switcher links to the equivalent page, not the homepage.
• Verify structured data is present and localised on the translated templates.
• Confirm robots.txt allows both search and AI crawlers on public content.

Frequently asked questions

Keep reading