Canonical (`<link rel="canonical" href="...">`) tells search engines 'this page is a variant of that page; index that one instead.' The page is still accessible to users.
Use canonical when: (a) duplicate content exists for legitimate UX reasons (e.g., printer-friendly versions, paginated archives), (b) parameter URLs that should consolidate to a clean URL, (c) cross-domain syndication where you want credit for the original.
Do NOT use canonical to redirect users — that is what redirects are for. Canonical is a hint to crawlers, not a user-facing redirect.
Multilingual sites add a critical wrinkle: hreflang tells Google which language variant to serve. Each language variant should self-canonicalize (point its canonical to itself), and the hreflang cluster should reference all variants including itself.
Common mistake: pointing canonical from /fr/page to /en/page. This collapses your French index entirely. Self-canonicalize every language variant; let hreflang do the language-routing work.
Approximately yes — Google has stated that the dampening factor on 301s is essentially zero in modern Google. The historical ~15% loss figure is no longer accurate. Use 301 confidently for permanent moves.
Permanently when possible. At minimum 12 months — long enough for Google to fully process and for external linkers to either update their links or move on. Removing redirects too early loses external link equity.
301 routes both users and crawlers; canonical only routes crawlers (users still land on the requested URL). Use 301 for permanent moves; use canonical for legitimate duplicates that need to remain accessible to users.
Yes — domain-to-domain 301s pass equity cleanly when the destination is topically relevant. For full-site domain migrations, expect 4-12 weeks of fluctuation while Google processes the move.