• Duane Forrester

    I’m reaching for my tin-foil cap to stand right beside you, Ian! Great overview. I’m seeing the same kinds of uneven, incorrect implmentations as you point out. It’s a really, really simple thing to misunderstand and insert the wrong URL into the code. Seasoned SEOs will spot the issue, but maybe not in time to stop problems.

    I think the two biggest take aways are:

    1 – fix the problem, don’t bandage it
    2 – things change, so relying on any things like this create more work later

  • Ian Lurie

    Duane, couldn’t agree more (obviously). In 13+ years in this business, I’ve never seen a tag, gadget, add-on or other ‘workaround’ introduced by anyone – including search engines – that really worked as advertised.

    Remember nofollow?

    The only exceptions MIGHT be robots.txt and the meta robots tag, but I’ll bet someone will point out some issues there, too.

  • http://www.dragonsearchmarketing.com DragonSearch

    OK, from one curmudgeon to another, I beg to differ… sort of.

    Will the canonical tag be a crutch and a band-aid(tm) for many developers. Yes, no doubt. But let’s not defenestrate this baby with the bath water. There are some really legitimate situations where the tag makes a lot of sense. In comlex ecommerce sites, there are often many instances where:
    1. You want a product to show up in several places, within the context of a certain group of pages, and
    2. It isn’t practical to create unique text for each instance.

    So, c’mon, enough of the highhorse business. Canonical is a tool that makes sense.

  • http://www.sitestreamseo.com Alastair

    Thanks for a good post.

    You seem to be expecting some abuse on this one but I couldn’t agree more. The search engines providing a band-aid does not make bad practice a good idea. I have nothing against using the tag, as a backup or a last resort but it is not the ideal solution. The ideal solution is to fix the problem.

    I can just see the developers of web apps and CMS platforms in their meetings now. A lot of them never really believed or were aware that spewing out duplicate content was a problem so what are the chances that they build in rel=canonical and refuse to engage with the issue beyond that?

    Anyone who has been doing this for more than a year or two ought to have seen this before and nofollow is a perfect example. You solve your own problems, you don’t rely on individual search engines to do it for you.

  • http://www.dragonsearchmarketing.com Etela

    Thank you Ian. I couldn’t agree more. This is an ongoing, never ending discussion between my boss (who’s comment you see above under DragonSearch) and I. I’ve been saying the same things you are. This is a tag that is not in our control, which means the search engines can change the way they treat it any time. Matt Cutts himself said that they do not promise they’ll absolutely abide by this 100%. The canonical link element is a hint, not a directive/requirement. Search engines reserve the right to choose when to use or ignore it. Matt also recommends fixing duplicate issues upstream, looking at all the other alternatives to fix it before becoming rel=canonical happy.
    While it is not easy and not always feasible to redo the entire site’s information architecture, it should be set as an ultimate goal rather than relying on tools controlled by others.

  • conrad

    Thanks for the post.

    The rel=canonical tag must never be a solution to solve linking issues. If this is the case the website’s architecture should be solved.

    However I think the rel=canonical as DragonSearch also outlined is still correct to implement in certain scenarios, such as a duplicate page which is a printer friendly version.

  • Ian Lurie

    @DragonSearch I understand what you’re saying. As it stands now, if every search engine consistently applied rel=canonical, it’d be a useful tool (IF applied sensibly, which it never is).

    But it’s not applied consistently. Not even within any one search engine. It’s a total failure as an SEO tool. Even worse, it’s become one more reason for people not to build their sites right in the first place.

    Want to show one product in several places? How about aggregating the products on category pages, and then linking to a single common product page? You can use javascript to provide a contextual backlink to the category page. That’s what all that stuff was built for.

    The e-commerce scenario you’re describing happens all the time. That doesn’t mean it’s right, or that it’s the best solution.

  • http://2helixtech.com matthiaswh

    Fun post as always, Ian.

    I think saying the tag is bad because people don’t use it properly is a little bit of backward logic. Yes, it’s primarily a bandaid for emergency measures resulting from bad site architecture or poor linking. What kind of people (particularly if there is an SEO involved) make those mistakes commonly? The kind that are going to want to take the easy way out by using the canonical tag.

    That isn’t to say good SEOs and webmasters don’t make mistakes, because they do. But they’re the ones that are going to fix the problem properly (the hard way), and only slap the canonical tag in there as a last resort or temporary fix. And they’ll be careful to use it properly.

    The people that the tag is designed to help the most are the ones who already make frequent mistakes with architecture and linking. Why would you expect them to use the tag properly?


  • robhammond

    Good post, absolutely agree – also a much neglected factor is how much duplicate content screws up the accuracy and usefulness of web analytics, which is a business case in itself for fixing the problem properly.

    Obviously rel=canonical doesn’t help the analytics issue at all, and if anything risks SEOs providing advice that we have to backtrack on at a later date. Only really advisable as a last resort with lots of disclaimers IMHO.

  • http://KTurk KTurk

    Great article on rel=canonical, I was always aware of the benefits but never saw the negatives to using the tag.

    Question for Ian or anyone else (I would greatly appreciate any insight into this scenario if this IS when you want to us rel=canonical)

    On a client site, it doesn’t take too much effort to request page edits/put 301’s in place, etc. The URL structure is clean and very user friendly through out the entire site, except for the homepage. Links from the home page (on the page, not within top nav) contain a Webtrends parameter (?WT) for analytic purposes. Google and Yahoo have recognized some duplicate pages and some people have linked to the pages using the URL with the Webtrends parameter.

    Implemented properly, would this be a best case to use the rel=canonical tag? (I don’t think the Webtrends parameters would be removed)

    Reclamation will be attempted to update the inbound link URLs; I am more concerned about the value being passed from the home page to these specific pages.

    Again, thank you for providing any insight into this situation.