What should I check first on a multilingual site?

Start with the basics that everything else depends on: each language has its own crawlable URL, pages are not blocked by robots.txt or noindex, and the content is actually translated. Then verify hreflang, canonicals and sitemaps. Fixing these first prevents wasted work on finer details.

How do I know if Google has indexed my translated pages?

Use the URL Inspection tool in Search Console on a sample of translated URLs and check the Page indexing report for statuses such as Crawled or Discovered but not indexed. A sitemap listing each language version also helps you compare submitted URLs with indexed ones.

How often should I audit a multilingual site?

Run a full audit at launch and after any major change, such as a redesign, new language, platform migration or URL change. Between those, a lighter monthly check of Search Console coverage and new pages that lack translations catches most regressions.

Is machine translation acceptable for SEO?

Google's spam policies target content produced at scale without adding value for users, however it is created. Translation that is reviewed, accurate and useful is not the problem; bulk, unreviewed output is. Review high-value pages, and avoid publishing large volumes of pages nobody has checked.

Multilingual SEO Checklist for Launch and Audits

A multilingual site is ready for search when every language version has its own crawlable URL, reciprocal hreflang annotations, a correct self-referencing canonical, a sitemap entry, genuinely translated content, and links users and crawlers can follow between versions. Use the checklist below before launch and again whenever you change the site; each item notes what to check and, where it matters, how to check it.

Before you start: is the structure decided?

[ ] You have picked a URL pattern (folders, subdomains or country-code domains) and applied it to every language. See subdirectories vs subdomains vs ccTLDs.
[ ] You are not switching languages with cookies, session state or URL parameters on a single URL. Google recommends separate URLs per language and lists parameters as not recommended.
[ ] You have decided whether slugs are translated, and applied that rule consistently. See translating URLs and slugs.

URLs and crawlability

[ ] Every translated page returns HTTP 200 on its own URL and is not behind a login or a script that only runs after interaction.
[ ] Internal links between versions are normal <a href> links. Crawlers follow those; a JavaScript-only dropdown that rewrites the page in place is easy to miss.
[ ] robots.txt does not block translated paths, and translated pages do not carry noindex by mistake (common after copying staging settings).
[ ] Letter case, trailing slashes and www versus non-www are consistent. Google's URL structure guidance notes that URLs are case sensitive.
[ ] If any URLs changed, old URLs permanently redirect (301 or 308) to the new ones in one hop.

hreflang

[ ] Each version lists itself and every other version. Google says that if two pages do not point to each other, the annotations are ignored.
[ ] Codes start with a language, optionally followed by a region: en, en-GB, de-CH, zh-Hant. Region-only codes and en-UK are wrong. See language and region codes.
[ ] The href values are absolute, canonical, indexable URLs.
[ ] A fallback x-default page is present where it makes sense, such as a language selector.
[ ] You use one implementation method per page group (HTML tags, HTTP headers or sitemap) to avoid conflicting sets. See hreflang implementation methods.

A minimal HTML example for a page available in English and German:

<link rel="alternate" hreflang="en" href="https://example.com/pricing/" />
<link rel="alternate" hreflang="de" href="https://example.com/de/preise/" />
<link rel="alternate" hreflang="x-default" href="https://example.com/pricing/" />

For a deeper walk-through, see the hreflang guide and the list of common hreflang mistakes.

Canonical tags

[ ] Each translated page has a canonical pointing to itself (its own language version), not to the default-language page. Pointing every translation at the original tells Google to treat the translations as duplicates.
[ ] Canonical URLs match the URLs in your hreflang sets and sitemap exactly, including protocol and trailing slash.
[ ] Parameter variants (tracking, sorting) canonicalize to the clean URL.

More detail is in canonical tags on multilingual sites.

XML sitemaps

[ ] Every language version is listed, and only indexable 200 URLs appear.
[ ] If you use sitemap-based hreflang, each <url> includes xhtml:link entries for all versions including itself, with the xmlns:xhtml namespace declared.
[ ] The sitemap is submitted in Search Console and referenced in robots.txt.
[ ] lastmod changes only when the page really changes.

Google's documentation notes that combining extensions such as hreflang can significantly increase sitemap file size, so watch the size limits on large sites. See multilingual XML sitemaps.

Content quality

[ ] Each page uses one language for its main content and navigation. Google advises against side-by-side translations and against translating only the boilerplate while leaving the main content in another language.
[ ] Titles, meta descriptions, headings, image alt text, structured data text and error pages are translated, not just body copy.
[ ] Key pages have been reviewed by a fluent reader. Google's spam policies cover scaled content abuse, meaning many pages produced mainly to manipulate rankings, however they are created. Quality review is how you stay clear of that. See machine vs human vs hybrid translation.
[ ] Keywords were researched per market, not translated word for word. See international keyword research.
[ ] Dates, currencies, units, phone formats and legal text match each market.

Internal links and the language switcher

[ ] The switcher links to the equivalent page in the other language, not just the homepage.
[ ] Language names appear in their own language (Deutsch, Français), and the switcher is reachable by keyboard. See language switcher UX.
[ ] You do not force-redirect visitors by IP or browser language. Google advises avoiding automatic redirection between language versions because it can prevent people and crawlers from reaching all versions. Offer a suggestion banner instead. See auto-redirect by language or location.
[ ] Body links inside translated pages point to translated pages, not back to the default language.

HTML basics

[ ] The lang attribute on <html> matches the page language. Google says it does not use the lang attribute to detect page language, but it matters for accessibility (WCAG 3.1.1) and screen readers.
[ ] Right-to-left languages set dir="rtl" where required.

Indexing checks in Search Console

Check these after launch, and again two to four weeks later:

URL Inspection. Test a sample of translated URLs per language. Compare the user-declared canonical with the Google-selected canonical. If Google chose a different canonical than you declared, find out why (often the translation is too similar to another version, or the canonical is wrong).
Page indexing report. Look for statuses such as "Duplicate without user-selected canonical", "Duplicate, Google chose different canonical than user", "Discovered - currently not indexed" and "Crawled - currently not indexed". Group the affected URLs by language folder.
Sitemaps report. Compare submitted URLs per sitemap against indexed counts.
Performance report. Filter by page path (for example /de/) and by country to see which markets get impressions.
Live test. Use it on a changed URL to confirm Google can fetch and render it, remembering that a passing test does not guarantee indexing.

Requesting indexing does not guarantee inclusion, and daily limits apply, so rely on sitemaps for large sites.

Key takeaways

Give every language version its own crawlable URL, canonical and hreflang set.
Make hreflang reciprocal and self-referencing, using valid language and optional region codes.
Translate everything users and search engines see, not just the body, and review the pages that matter most.
Do not force-redirect visitors by language or location; link between versions instead.
Verify in Search Console with URL Inspection, the Page indexing report and sitemaps, not just by eye.

Sources

Managing multi-regional and multilingual sites – separate URLs, avoiding automatic redirection, single language per page
Localized versions of your pages – hreflang rules, reciprocal links, x-default, lang attribute note
Combining sitemap extensions – hreflang in sitemaps and file-size caution
URL Inspection tool – canonical comparison, live test, request indexing
Page indexing report – duplicate and not-indexed statuses
Spam policies for Google web search – scaled content abuse