Clarify "canonical" word: Can someone please list the various uses of this word in SEO context, and point out correct or incorrect uses of it?

Question

I think the authoritative use of the word "canonical" in the SEO discipline dates back to this 2006 blog post by Matt Cuts on cononicalization:

http://www.mattcutts.com/blog/seo-advice-url-canonicalization/

Personally, I recall the first use of the word in referring to penalizing sites that were buying up many keyword stuffed URLs and pointing them all to the same site, aka "Canonical URLs."

My curiosity caused me to discover that there are math and science uses of the word, AND that the term CNAME in DNS means "canonical name."

At this point, and on this website in particular, I see that word tossed around so frequently and I suspect some of the uses of it are technically wrong, or that some of its uses could better be served with a different word or phrase. But I could be wrong about this, fueled by my less-than-complete understanding of what canonical means in all of its SEO contexts.

So, then,

• what are the correct uses for the word canonical (in SEO context)?

And, if my suspicions are correct,

• what are some of the common mis-uses of the word,
• and what should be used in its place?

Answer 1

Definition: canonical in search optimization - while other uniform resource identifiers may "point" to the same resource, the resource has only one intended name and search engines should associate all links from other uniform resource identifiers which serve the same content to that canonical URI. (See also: Specify your canonical)

An example of multiple URI's (similar to the one suggested in the blog entry you linked) which could theoretically point to the same content on a webserver which has not been configured properly:

• 1.2.3.4/example/
• 1.2.3.4/example/index.htm
• domain.com/example/
• domain.com/example/index.htm
• www.domain.com/example/
• www.domain.com/example/index.htm
• example.domain.com/
• example.domain.com/index.htm

(... and you could add any combination of miscellaneous GET params which may or may not have any effect on the contents returned by the webserver - i.e. "index.htm?a=b&c=d" ...)

As you can see, these URI's could be viewed as unique entries when people start linking to all of them and the search engines could use any one of these links to point to the content... the rel="canonical" attribute allows the publisher of the document to indicate which URI is "correct".

This is analogous to the webmaster indicating "Instead of configuring my webserver to redirect traffic from URI's which I do not want to see in the search results page to the URI which I intended, I would prefer to add markup to my documents to tell the search engines which URI should be listed in their indexes"

Misuses: Second-guessing the search engines' algorithms and using the canonical tag for anything beyond indicating which URI you want to see used for ranking is probably a misuse.

Instead of rel="canonical": Consider issuing a 301 redirect when content is requested at any hostname or path which is not unique.

Answer 2

Well, google webmaster and quality guidelines are very clear to this aspect, but let me try a free translation...

On backend you have thousands of products, and you have a 'showProducts' page, that will list 50 products each time.

You can access:

• http://domain.tld/showProducts/
• http://domain.tld/showProducts/2/
• http://domain.tld/showProducts/3/ ...

Alright. Now, let's suppose users can apply a filter, for example, by price.

• http://domain.tld/showProducts/?sort=price&sortOrder=ascending
• http://domain.tld/showProducts/2/?sort=price&sortOrder=ascending
• http://domain.tld/showProducts/5/?sort=price&sortOrder=descending

Done. All those pages have one commom purpose: list products. More than that, they can change along with time (having more or less products). So we can assume this dinamically generated list will not be the same over the time. Products displayed today, can be no more present next week. Prices can increase/decrease...

In this context we can see that treating sorted content as different content - just because it is dinamically generated - also doesn't makes sense.

Doesn't makes sense assume this need to affect SEO, as doesn't makes sense to crawlers keep indexing this kind of content over and over. For this, we can provide a canonical version. What version should crawlers follow and avoid seeing duplicate content? This is the canonical version.

Note that you can tell crawlers that the correct version to follow is anyone. May be the unsorted version, may be the increasing price order, alphabetical name order... Doesn't matter. The intention is to make crawlers don't see duplicated content (this would happen a lot if crawlers follow all sorting possibilities).

Hope this is clear enough.