Canonical tag and 301 redirect at a glance
A canonical tag is a link element in the head of a page, or an equivalent HTTP header, naming the URL the site owner prefers among duplicates or near-duplicates. The page carrying it still loads normally; a visitor on a sorted category view or a tracked URL sees exactly what they expected. Google's documentation on consolidating duplicate URLs describes the canonical as one of several signals it weighs when choosing which URL to show, alongside redirects, sitemap entries and internal links. When Google picks a different URL from the one declared, the troubleshooting page for duplicate without user-selected canonical covers how to find out why and align the signals.
A 301 redirect is a server response that says the content has moved permanently to another address. A browser requesting the old URL is sent to the new one without the visitor having to do anything, and a crawler follows it in the same way. Google's documentation on redirects describes a permanent server-side redirect as a strong signal that the target should be canonical. The site migrations guide covers how redirects are planned and mapped when a site changes structure or domain; this page does not repeat that.
The question that separates them is simple: should anyone still land on the duplicate? If the answer is no, because the page has moved, been merged or exists only as an accidental variant, redirect it. If the answer is yes, because the variant serves a purpose for visitors, canonicalise it.
Compare table
Canonical tag vs 301 redirect
| Row label | Canonical tag | 301 redirect |
|---|---|---|
| What users see | The duplicate page, exactly as it is | The destination page; the old URL is never shown |
| What search engines do | Weigh the declared URL against other signals when choosing what to index | Follow the redirect and treat the destination as the URL to index |
| Strength of the signal | A hint: Google may choose a different URL | Strong: the old URL no longer serves content |
| Reversibility | Easy: change or remove a tag in the page | Possible, but recovery takes time once search has moved to the destination |
| Effect on links | Signals can consolidate onto the canonical when Google accepts it | Signals from links to the old URL consolidate onto the destination |
| Crawl effect | Duplicates are still crawled, because they still exist | The old URL is fetched only to discover the redirect, and less often over time |
| Implementation | A tag in page templates or an HTTP header | A rule on the server, platform or content delivery network |
| Best-fit cases | Filtered, sorted, tracked, printable and syndicated variants that must stay reachable | Moved, merged or retired pages and protocol, host or trailing-slash variants |
The rows on what users see and strength of the signal settle most cases. A canonical is the right tool only when the duplicate must remain usable; otherwise the stronger, simpler redirect is better for visitors and for search.
When to choose a canonical tag
Choose a canonical tag when the variant serves visitors. A category page sorted by price, filtered by colour or carrying campaign tracking parameters is useful to the person on it, even though search only needs one version. Redirecting those URLs would break the feature. The faceted navigation guide covers how filtered URLs are tiered and which of them deserve to be indexed; the canonical is one of the tools it uses.
Choose it for content published in more than one place on purpose. A printable version, a version for a partner site or a product sold in two categories with the same description can all keep their URLs while pointing search to the preferred one. Google's documentation describes syndication and cross-domain canonicals among the cases it supports. Setting up a partner's copy of an article, including when a noindex on the copy is the safer choice, is covered in the guide to syndicating and republishing content.
Choose it as a baseline on every indexable template. A self-referencing canonical on each page states the preferred address even when no duplicate is known, which protects against accidental variants created by parameters, capital letters or tracking. It costs little and removes an ambiguity Google would otherwise resolve for itself.
Accept that it is a hint. If the duplicate's content differs substantially from the canonical, or internal links and the sitemap point at the duplicate, Google may choose differently. Keep the signals consistent and check URL inspection when it matters.
When to choose a 301 redirect
Choose a redirect when a page has moved. A new URL structure, a change of domain or a merger of two pages leaves old URLs that people and other sites still link to. Redirecting each one to its closest new equivalent sends visitors to the right place and tells search where the content now lives.
Choose it for technical variants nobody needs. The same page served on both secure and insecure protocols, with and without the www prefix, or with and without a trailing slash should resolve to one form. A redirect enforces that for every visitor and every crawler, which a canonical tag cannot.
Choose it when two pages compete for the same purpose. Two articles answering the same question, or two near-identical service pages, split links and confuse search. Merge the content into one page and redirect the other; a canonical between two pages that both remain live leaves the duplication in place.
Redirect to a genuine equivalent. Sending retired pages to the homepage or an unrelated category does not consolidate anything; search engines can treat that as if the page were simply gone. If no equivalent exists, let the old URL return a not-found or gone status instead.
An art print shop sorts its duplicates
Oddfellow Prints moves its shop from /shop/ to /prints/ and finds four kinds of duplicate. Its 1,200 product URLs have new addresses, so each gets a 301 redirect to its new equivalent. The shop also has 340 size-variant URLs that buyers use to pick a size, so those stay live with a canonical tag pointing to the main product page. Another 85 discontinued prints have a close replacement and are redirected to it, while 40 with no replacement return a gone status rather than redirecting to the homepage. Finally, the old site answered on both the www and bare domain; every request is now redirected to the www form, and every live page carries a self-referencing canonical. Eight weeks later, URL inspection on a sample of size variants shows the main product page as Google's chosen canonical, and the old /shop/ URLs have dropped out of the index.
Decision table
| Situation | Choose | Why |
|---|---|---|
| Page moved to a new URL in a restructure or domain change | 301 redirect | Visitors and search should reach only the new address |
| Sorted, filtered or tracked versions of a page that users need | Canonical tag | The variant must stay reachable; search needs only one version |
| Protocol, host or trailing-slash variants of every page | 301 redirect | No one needs the variant, and the redirect enforces one form |
| Two pages covering the same topic | Merge the content and redirect one | A canonical between two live pages leaves the duplication in place |
| Content deliberately republished on a partner site | Canonical tag on the copy, pointing to the original | Both copies stay live; search is told which is primary |
| Retired page with no equivalent | Neither: return a not-found or gone status | A redirect to an unrelated page consolidates nothing |
Cases that need both or neither
Some cases need both tools at once. After a redirect, the destination page should carry a self-referencing canonical, so the redirect and the tag point the same way. Where variants of a page are redirected, such as an insecure version, and the surviving page also has parameter variants, the redirects handle the first and canonicals handle the second. The signals must agree: a canonical that names a URL which itself redirects elsewhere sends search a contradiction.
Other cases need neither. Two pages that look alike but serve genuinely different purposes, such as service pages for two towns, are not duplicates to consolidate; they need content distinct enough to stand on their own, or they should be merged. A page that should stay live but out of search results needs a noindex directive rather than a canonical, which the comparison of noindex and robots.txt covers. And a page with no future should return a not-found or gone status.
Watch the crawl cost. Canonicalised duplicates are still crawled, because they still exist, and on large sites that can consume Crawl BudgetHow much a search engine crawls a site. better spent elsewhere. Where variants multiply without limit, reducing the number of crawlable URLs matters as much as declaring a canonical. The comparison of subfolders and subdomains covers a related structural choice that often triggers redirects in the first place.
When a cross-domain canonical on a partner's copy is ignored and the copy is shown instead, the diagnosis for syndicated content outranking the original checks whether the copy is still a duplicate and whether its directives can be read.
Common mistakes
- Using a canonical tag where a redirect belongs, leaving moved pages live for visitors and search.
- Redirecting filtered or sorted views that users rely on, which breaks the feature.
- Redirecting retired pages to the homepage instead of to a real equivalent or returning a not-found status.
- Pointing canonicals at URLs that redirect, are noindexed or return errors, which sends contradictory signals.
- Declaring one canonical while internal links and the sitemap use another URL.
- Canonicalising pages that are not duplicates, such as distinct products or locations, and expecting Google to agree.
- Chaining redirects through several hops after repeated restructures instead of pointing each old URL at its final destination.
