Google & Bing Disagree On Rel=Canonical Implementation

Google vs Bing Rel=Canonical

Bing recently posted an article:  Managing redirects – 301s, 302s and canonicals

As someone that works with Enterprise level sites on a regular basis, I deal with duplicate content & the canonical tag a lot. What really jumped out in Duanne’s article is this (emphasis added is my own):

Something else you need to keep in mind when using the rel=canonical is that it was never intended to appear across large numbers of pages.  We’re already seeing a lot of implementations where the command is being used incorrectly.  To be clear, using the rel=canonical doesn’t really hurt you.  But, it doesn’t help us trust the signal when you use it incorrectly across thousands of pages, yet correctly across a few others on your website.

A lot of websites have rel=canonicals in place as placeholders within their page code.  Its best to leave them blank rather than point them at themselves.  Pointing a rel=canonical at the page it is installed in essentially tells us “this page is a copy of itself.  Please pass any value from itself to itself.  No need for that.

This surprised me as it contradicts what we’ve been told by Google, and also contradicts my opinions on best practices for the canonical tag. I have a deep respect for Duanne, so I’m only raising this as an issue because the rel=canonical tag is supposed to be jointly supported by Google & Bing, yet we’re getting very different direction on how it should be used.

For instance, in this Google Webmaster Help video (H/T to James Constable for the link), Matt Cutts says “it doesn’t hurt to have this on every page of your site.”

Likewise, in the Google Blog, Maile Ohye of Google says “Yes, it’s absolutely okay to have a self-referential rel=”canonical”. It won’t harm the system and additionally, by including a self-reference you better ensure that your mirrors have a rel=”canonical” to you.”

Further evidence Google says its okay to have both a self-referential rel=canonical tag as well as to use the canonical tag site-wide can be found in this SEOmoz post, as well as another Google Webmaster Help video from Matt Cutts saying its okay.

I asked some of the top SEO minds on Twitter as to their opinion, and they all seem to support the notion that is okay to use the canonical tag site-wide and include self-referential canonical tags. Here are their responses:

 @Thos003: I use rel=canonical on it’s own page to deal with URL strings. I find this article disturbing

@rbucich: of course you can! Either I don’t understand the question or the person who said no doesn’t

@ajkohn: Yes. I think rel=canonical is a safeguard for HTML barnacles that might attach themselves to the canonical URL.

@james_constable: YES it needs to for all ad tracking, parameter non sense you need to avoid dups.

@dannysullivan: i wasn’t aware this was a problem. i guess i don’t see that it is. google said it wasn’t

@halvorsen I do.

So, it seems were at an impasse – what are your thoughts – is it okay to use the rel=canonical tag sitewide? Is it okay to use the rel=canonical tag on the canonical version of the page?

Will you change your recommendations to clients (or your own sites) on how they should use the rel=canonical tag based on Bing’s latest blog post?

Your comments are appreciated.

**Update @MattCutts chimes in:

@nickroshon certainly it won’t mess things up with Google to do it on every page. Can’t speak for MSFT/Bing, of course. (link)

@nickroshon self-loops on canonical pages are fine. Otherwise you’re stuck doing per-page logic for every page; not fun. (link)

Matt Cutts tweets back

**Update 2: This post has gotten some great conversation going, and also picked up by a few other sites, so I wanted to link out to that as well. It appears the SEO community finds this to be an important issue, and it’s obvious duplicate content is the thorn in everyone’s side…thanks to all that have shared my post!

21 thoughts on “Google & Bing Disagree On Rel=Canonical Implementation

  1. It seems interesting to me that Bing might disagree with Google on this. But Matt’s statement was made early on, pre-panda. Since the Panda update, sites scrambling to cope with thin content started abusing the canonical tag and using it on similar pages but not duplicate content pages. Other syndicated content sites started using it to pick the winner for a story and push juice from other sites to one main site. So perhaps this is Bing’s reaction to the abuse of canonical tags.

    1. I think you are definitely on to something that this is a reaction from Bing to deal with abuse/spam of the canonical tag.

      But as you’ll see above, Matt did chime in today (Oct 7, 2011) to clarify that self-referential canonical tags are okay, so I think what he shared in the video still holds as well.

  2. Nick – suspect Bing doesn’t want encourage more work on a site than in necessary. I have seen canonical implementations that do more harm than good. In one such case, the canonical tag mirrored the url in the address bar so if it contained tracking parameters, so did the canonical.

    That said, on an enterprise level there are so many different processes by a multitude of users, the existence of canonical tags have kept us out of trouble. A favorite is co-brand subdomains which duplicate content. They may only intend to have a page or two visible to users but any relative links spill the subdomain crawl path site wide.

    1. Great points Rick. I’m sure there are tons of sites that have implemented incorrectly, and speaking with Duanne personally a few weeks ago he confirmed they see more incorrect implementations than correct.

      I suppose that is why I also found it important to clarify: what is a correct implementation of rel=canonical? What is an incorrect one? It seems that Google & Bing differ on how exactly they define that.

  3. Hi Nick

    Your first link in the post to the Bing article is a 404. Also no matter what Bing says, in the enterprise level, it is canonical or toast. Have a good day.

    1. Thanks Mert, I fixed the link – not sure what happened there.

      I agree, on the enterprise level, my advice is still canonical tag on every single page, even if its self-referential. This practice makes the most sense, is the easiest to implement, and is concurrent with what Google recommends – I’m just surprised Bing doesn’t agree with that approach.

  4. Great post, Nick! I’m a little perplexed at Duane’s response. One very valuable aspect of having rel=”canonical” tags deployed on every page is that it enables site marketers to test click path behavior with Coremetrics and Omniture tags (e.g. adding tracking codes to URLs used with internal links) freely without creating duplicate content. I would be interested to hear Duane’s response for this.

    1. Thanks Richard! I agree, the rel=canonical tag is critical in cases like that, as a 301 redirect isn’t really feasible when dealing with tracking codes and A/B testing…

  5. I really cant give much merit to that article. The idea of canonical makes sense to reference the original source and if it is not reference who it is. How can you think your going to be “penalized” by doing so?

    1. I don’t think Bing is implying you’ll be penalized, I think they’re just saying if you do a self-referential canonical tag, or deploy the canonical tag site-wide, then you’re using the canonical tag incorrectly, and Bing will ignore it due to improper implementation…Regardless, I agree with your sentiments that logically it makes most sense to deploy on every page.

  6. The rel canonical tag is here to stay as it helps way toomany site owners deal with very important issues. Bing will just have to adjust how they treat this feature.

  7. Canonical tag in most case is the best way to get rid of duplicate pages, handle query strings and other parameters from tracking codes and ad ops, but also makes it easier to have both a good user experience as well as a good SEO strategy for landing pages. I can kind of see what Bing is talking about and it almost makes sense in that you don’t need a tag for the page you are on, especially if there is nothing to append the url, but I’m also surprised that Bing is also saying that it could maybe hurt a site because they look at it as a duplicate page of itself. This is kind of strange and shouldn’t be the case. I can see where they would recommend not putting one there because it’s not needed in some cases, but to say that the tag would tell Bing that the page is a copy cannot be good practice. I think Google interpretation of the tag is much better than Bing and I would hope that they will eventually correct this and make it so that the engine doesn’t look at a self-pointing canonical tag as a copy or duplicate page of the one it’s pointing too. However, I can’t really see how Bing could count two pages with the exact same url as duplicates of themselves, but that may indicate a bug with the system or lack of understanding to program the right solution. To me, either reason would be disappointing for Bing.

Comments are closed.