I've used it frequently and seen it work for canonicalizing the site's main page for "www." and for /index.#*$! or /default.#*$!, as well as for killing superfluous query strings. It's very handy for situations where setting up a 301 redirect is difficult or impractical.

I have used canonical tags on thousands of pages to pull dupes out of the SERPs, and they have worked perfectly!

The dupes were product or category pages, sometimes our fault, sometimes the fault of other webmasters linking to us, but the net result was we were able to normalize thousands and thousands of URLs this way. I prefer server-side 301s, but sometimes you cannot predict who will do what out there. ;) In fact, we even have instances of googlebot dreaming up URLs to see if they will resolve, and we can use canonical to intercept them all and make sure we don't have a big dupe content problem developing unbeknownst to us. With a canonical in place, it doesn't matter what URL is used to get to that page, it will normalize into the preferred URL. With canonicals, nothing can slip through the cracks.

Canonicals have worked EXACTLY as google promised, for us. I was a little surprised, but they really do work OK.

in the recent Matt Cutts Interview by Eric Enge [stonetemple.com] the rel=canonical tag was discussed extensively.
this probably best sums it up:

Typically, the crawl team wants to consider these things hints, and the vast majority of the time we'll take it on advisement and act on that. If you call it a directive, then you sort of feel an obligation to abide by that, but the crawling and indexing team wants to reserve the ultimate right to determine if the site owner is accidentally shooting themselves in the foot and not listen to the rel=canonical tag. The vast majority of the time, people should see the effects of the rel=canonical tag.

I have proof that Google take even longer to recognise canonical issues than they do with regular 301s ... and that's a long, long time!

One site I worked with was getting those funky backlinks that include query strings the site doesn't use - you may have seen this kind of disruption yourself. We instituted canonical tags as an attempt at a quick fix, and the oddball urls vanished from Google within a week - before we put a full solution in place on the server.

Clearly, it's a YMMV situation.

Thanks guys. I know that with anything SEO, there is no proof that anything really works, so I am trying to gauge if the effort to implement canonical is worth it. We would have to write logic for some of the tags.

Deciding whether the effort is worth it depends on how problematic your current situation is. If it's not currently "broken" then the fix is really just insurance against possible future problems.

If you do see current problems, the ideal is to fix the canonical issues on your own server - you can never be certain that Google will execute the canonical tag exactly the way that you hope, and they only take the canonical tag as a "suggestion" at any rate.

Our tech team accidently added an inadvertant Canonical Tag onto 1000+ forums-type pages...

These pages were generating about 2000 long tail Google Visits a day and after we accidently introduced a canoncial tag in there we lost all the traffic. We realized the mistake 2 days after the change went live (the traffic started dropping almost immediately) and reverted the change.

The traffic has still not come back!

Be careful with the canonical tag.

[edited by: Robert_Charlton at 6:16 pm (utc) on Mar 21, 2010]

Hello Deepesh, and welcome to the forums.

How long ago did your team make this canonical error? Did you place the same canonical tag on thousands of very different pages, or was it something more subtle than that?

Hey tedster -

We made the change on Feb 20th (the mistake with the canonical tags). How it was implemented was such that on all the long tail pages (there are 1000's in Google's Index)

The Canonical was accidentally set to "/forums/keyword" ... yes.. for all 2000+ pages :|

On Feb 25th we started seeing traffic dip and it wasn't until Feb 27th we actually removed the accidental canonical -- it took us awhile to figure out what happened - a developer had copied and pasted a different template completely accidentally.

We also proactively submitted an updated sitemap.

Unfortunately we haven't regained the traffic (amounting to about 40,000 Google visits per month).

Any suggestions?

[edited by: tedster at 3:23 am (utc) on Mar 22, 2010]
[edit reason] no urls - please see Charter [/edit]

Well, I guess that answers one question - Google does take action based on the tag. A month does seem like a long time to reverse out of the error, though. I would suggest a Reconsideration Request, if you haven't already done so.

Unfortunately, Google's resources seem preoccupied these days. As many other threads here illustrate, lots of things seem slower than usual. On top of that, the canonical tag is relatively new and we really have little data about how long it "should" take to reverse the effects of an error.

Just to clarify. The added the exact same tag in all pages?
Thanks

welcome to WebmasterWorld [webmasterworld.com], deepeshbanerji!

thanks for reporting that.
it is always interesting to know more about how the rel=canonical tag can affect a site.

Yes, exact same tag on all pages -- for about a 2-3 day time period -- after which they were removed (when we figured out the mistake).

Any ideas on how to reverse the mistake?

I would second Tedster about reconsideration request. One of reason of reverting so slowly could be that as google saw all these pages being marked as "duplicate" of one single page, that they do not crawl them that often and therefore are slower in reverting the change.

Reconsideration request explaining the mistake you made and the fact that you have rectified it could help.

One thing that this experience shows is that there is definitely some live voltage flowing in the canonical tag. As with certain other critical areas (robots.txt, .htaccess) some errors here can really mess you up!