Canonical link element

Purpose

A major problem for search engines is to determine the original source for documents that are available on multiple URLs. Content duplication can happen in many ways, including:^[3]

Duplication due to GET-parameters
Duplication with multiple URLs due to CMS
Duplication due to accessibility on different hosts/protocols
Duplication due to print versions of websites

Duplicate content issues occur when the same content is accessible from multiple URLs.^[4] For example, http://www.example.com/page.html would be considered by search engines to be an entirely different page from http://www.example.com/page.html?parameter=1, even though both URLs may reference the same content.^[5]^[6]

In February 2009, Google, Yahoo and Microsoft announced support for the canonical link element, which can be inserted into the <head> section of a web page, to allow webmasters to prevent these issues.^[7] The canonical link element helps webmasters make clear to the search engines which page should be credited as the original.

How search engines handle rel=canonical

Search engines try to utilize canonical link definitions as an output filter for their search results. If multiple URLs contain the same content in the result set, the canonical link URL definitions will likely be incorporated to determine the original source of the content. "For example, when Google finds identical content instances, it decides to show one of them. Its choice of the resource to display in the search results will depend upon the search query."^[8]

According to Google, the canonical link element is not considered to be a directive, but rather a hint that the ranking algorithm will "honor strongly".^[1]^[9]

While the canonical link element has its benefits, Matt Cutts, then the head of Google's webspam team, has said that the search engine prefers the use of 301 redirects. Cutts said the preference for redirects is because Google's spiders can choose to ignore a canonical link element if they deem it more beneficial to do so.^[10]

Factors Google Considers When Choosing a Canonical for a Page

There are multiple factors Google evaluates when determining the canonical version of a page, including:

The canonical tag you set up: This is the most direct way to suggest the preferred URL to search engines.
Internal linking: Pages with strong internal links pointing to them are more likely to be treated as canonical.
Sitemap.xml: The URLs listed in the sitemap also influence Google's decision.
Redirects: Google may choose a URL redirected to from others as the canonical version.

Implementation

Semantic tag

The canonical link element can be either used in the semantic HTML <head> or sent with the HTTP header of a document. For non HTML documents, the HTTP header is an alternate way to set a canonical URL.^[3]^[11]

By the HTML 5 standard, the <link rel="canonical" href="http://example.com/"> HTML element must be within the <head> section of the document.^[12]

Self-hyperlink

Some sites such as Stack Overflow^[13] have on-page hyperlinks which link to a clean URL of themselves. Usability benefits are facilitating copying the hyperlink target URL or title if the browser or a browser extension offers a "Copy link text" context menu option for hyperlinks, the ability for the original URL to be retrieved from a saved page if not stored by the browser into a comment inside the file, as well as the ability to duplicate the opened page into a new tab right next to the currently opened one if the browser lacks such a feature.^[14]^[13]

Examples

HTML

Below is an example of HTML code that uses the rel=canonical inside the <head> tag. The code could be used on a page, such as https://example.com/page.php?parameter=1to tell search engines that the https://example.com/page.php is the preferred version of the webpage.

<!DOCTYPE html>
<html>
<head>
  <link rel="canonical" href="https://www.example.com/page.php">
</head>
<body>
...
</body>
</html>

HTTP

HTTP/1.1 200 OK
Content-Type: application/pdf
Link: <https://www.newthink.life/page.php>; rel="canonical"
Content-Length: 4223
...