| 11:26 pm on Feb 24, 2009 (gmt 0)|
I used the tag last week to solve a parameter appending issue we had developed. It was basically 2,000 products that could be accessed with ?id=1232&cat1=something&cat2=somethingelse as opposed to ?id=1234. Google has already started using the correct version, updated urls are showing up in the index.
| 10:03 pm on Feb 25, 2009 (gmt 0)|
Google's Matt Cutts posted a 20 minute video about the new canonical tag. I haven't had the chance to view it yet, but it might clear up some gray areas for webmasters.
Some specific points addressed in the video:
1. The canonical tag will work across the https: and http: protocols
2. The canonical tag will work across hostnames (subdomains).
3. Avoid creating infinite loops where urlA says urlB is the canonical and then urlB says urlA is the canonical version.
4. Avoid a canonical url that returns a 404 or 410 status.
5. Avoid canonical chains, just as you should avoid 301 redirect chains.
6. The canonical url you declare does not need to get an exact content match from the server, but its content should be very similar. This means you can apply it to various "sort" versions of the same material.
| 5:15 pm on Feb 27, 2009 (gmt 0)|
Well this tagged has already proven successful for me on Yahoo at least...
i set this tag last week, and Y had me indexed as example.com, and now finally has me as www.example.com
| 10:20 pm on Feb 27, 2009 (gmt 0)|
The only area where I feel this is acceptable is if you have no control of the site from a server perspective. I have clients that have no access to their own server to make these modifications.
If you can fix them in the beginning then it is the way to go.
| 12:34 am on Feb 28, 2009 (gmt 0)|
In principle I agree with you - however in practice I have yet to see a site that actually deals with all the potential canonical problems [webmasterworld.com]. And that goes double for sites on an IIS server.
| 2:48 am on Feb 28, 2009 (gmt 0)|
|The canonical url you declare does not need to get an exact content match from the server, but its content should be very similar. This means you can apply it to various "sort" versions of the same material. |
This is where the utility of tag differs from having redirections in place. In other cases I agree, fix them in the beginning where possible.
Having different "versions of the same material" seems to be bending the meaning of canonical.
| 5:38 pm on Feb 28, 2009 (gmt 0)|
I am a little bit confused about this tag. I have read about this on Matt Cutts blog and have watched the video above. He gives the example of an online shop where a page with a product description can show up several times. In his example it's a website for a certain gummy candy called swedish fish.
The customer can access the product page with various links:
The simple product page:
The product page in the category: gummy candy
I'll come up with a third:
The product page in the category: gelatine free sweets
And a forth:
The product page in the category: salty sweets
Note, that in the example Matt Cutts gives, the category and product name is displayed in big bold letters, so the visitor can immediately see two informations:
1. In which category he is on the website
2. Which products are displayed.
Now Matt Cutts happily tells us: We can resolve this duplicate content issue by picking one of those product pages and making it our "canonical page" with the new "canonical tag".
So here is my problem: I do not think those pages have a duplicate content issue. I do believe that those four pages are entirely different and for some users the essential information lies in the 1 percent of information that differs in those four examples and that this information for me as a shop owner can decide if I make the sale or not.
What if a user searches for: "Gelatine free swedish sweets"?
Will my product page even show up now in the search results if I use No 1 as my canonical URL?
And even if it does - won't my potential customer be confused if he specifically searched for "gelatine free swedish fish" and this important information does not show up in a prominent place because I directed him to my "canonical URL". What about the user who wanted "swedish salty sweets". What about the user who wants "swedish gummy candy".
What if I made the URL "salty sweets" my canonical. Won't a visitor be confused if he does a search for "gelatine free sweets" and then ends up with "salty sweets"? That wasn't the information he was looking for.
When I understand the implications of the canonical tag correctly I might loose 75% of my sales on swedish fish if I use this tag.
I checked my own online shop. Tried searches like
product_type + part of product name
brand + part of product name
For both searches Google showed me another URL. However the correct product at its correct location in my website - for the specific search. One in the "brand category" one in the "product type category" How it should be. Like a visitor would expect it.
I would be nuts to implement this tag. It's fixing something that is not broken. Or did Matt Cutts only use a bad example?
| 12:18 am on Mar 1, 2009 (gmt 0)|
|In which category he is on the website |
This is always going to be problematical.
If the user browsed from a category page fine, showing category, or path thru category tree, is good a idea, be it a breadcrumb or heading.
If however the user gets there by other means, say direct from search engine, and products can be under multiple categories, showing a page suggesting that product belongs to a specific category is wrong, particularly if no category is suggested by the search terms.
Showing all categories the product belongs to is probably the best way to go.
Knowledge of how the user got to page ( search terms or referer ) could be used to modify page using script, if this is felt to be needed.
Be it a good idea or otherwise, this tag allows "where from" information to be encoded in the url, with the canonical tag telling the search engines to ignore it.
| 3:08 pm on Mar 6, 2009 (gmt 0)|
My initial tests threw up some interesting behaviours with this new tag:
I duplicated a record in my database and created a new page with an seo url where the only difference was the use of an "_" between 2 words instead of an "-". We'll call them test_page.html (the new page I am trying to canonical redirect to) and test-page.html (the existing page). Both pages were otherwise identical in content and structure.
My initial test was to see if the canonical tag would work if the canonical url (test_page.htm) was an orphan page not linked to anywhere else on the site. After setting the canonical tag on test-page.html I waited 10 days for any results.
As hinted at by Google by their "other factors" comment, they rightly ignored the new (unlinked to) page and continued to index the old page (test-page.html).
My second test was to link to both pages in an identical place (my html sitemap). Would the canonical tag page now take precedent as intended?
3 days after adding a link to the new page (test_page.html) the canonical tag was recognised BUT the initial page was dropped from the index and the new page WAS NOT added in it's place! Err.. oh.. where has all my traffic gone?!
A further 5 days after this the canonical url has now been indexed in place of the old url in exactly the same ranking position the old page had been. (phew)
My test posed Google with the problem of a brand new url as the canonical tag target. This was as brutal a use of this tag as I could think of and probably not what it's use would be in most circumstances.
However I didn't think the loss of rank (and therefore 4 days of traffic) was ideal and it got me thinking that there must be two independent processes at work with this tag, we can cause this kind of ranking issue.
Process one drops the old url from the index as it is non canonical, and the second process picks up the new one (probably on googlebots next visit).
This makes the canonical tag a much less efficient and more dangerous than other common techniques to repair canonical issues (such as traditional 301s).
| 11:32 pm on Mar 6, 2009 (gmt 0)|
tootricky: It may be interesting to search the server log file for test_page and test-page to see how pro-active the bot is in actioning the tag.
| 1:05 am on Mar 7, 2009 (gmt 0)|
This tag should have been here in 2005, but better late than never I suppose.
| 5:41 pm on Apr 2, 2009 (gmt 0)|
Is it best to include the trailing slash in the URL string in canonical tags or no? For example:
<link rel="canonical" href="http://www.example.com">
<link rel="canonical" href="http://www.example.com/">
<link rel="canonical" href="http://www.example.com/directory">
<link rel="canonical" href="http://www.example.com/directory/">
| 6:01 pm on Apr 2, 2009 (gmt 0)|
Many say that the best practice is to include the trailing slash for example.com/ and example.com/directory/ but not for example.com/page
But it is quite common practice to include the slash after /page/ as well, especially when a dymnamic site does not have a native directory structure and rewrites to search-friendly urls. Google clearly can deal with that practice.
Most of all, be consistent in your canonical tagging and your internal linking.
[edited by: tedster at 6:24 pm (utc) on April 2, 2009]
| 6:21 pm on Apr 2, 2009 (gmt 0)|
Thanks for the tips Tedster. I'll stick with / as internal linking follows that structure.
| 6:25 pm on Apr 3, 2009 (gmt 0)|
Did I miss something? Can someone tell me if on my home page I should use:
I use the one without the slashie currently and google is now showing an error in wt's looking for:
Could this be the problem?
| 9:23 pm on Apr 4, 2009 (gmt 0)|
@texasville, @tedster, @teenwolf re: trailing slash
For a plain address, add the slash (http://www.example.com/)
For a directory, add the slash (http://www.example.com/path/)
When in doubt, add the slash. Why?
Watch what most browsers do to the address when you enter it.
Click on http://www.example.com results in http://www.example.com/ in the browser address bar.
A plain address adds a slash after it, it's actually the standard, although every browser does the right thing there anyway.
As for the paths, if you DON'T put a slash for a directory, many web sites actually submit a 301/302 (IIS, and Apache, I believe), then redirect to the slashed directory.
The only time you shouldn't use a slash after a directory is if you have a site using mod_rewrite or ruby on rails, where every URL is passed through a handler which does normalization and page generating. In which case, the trailing slash is dependent on the app.
However, when in doubt, use the slash. If you really don't want to, use a HTTP header sniffer and see what the web server returns for the non-slashed version, if it gives a 200 then you are OK not using it.
| 11:02 pm on Apr 4, 2009 (gmt 0)|
*** If however the user gets there by other means, say direct from search engine, and products can be under multiple categories, showing a page suggesting that product belongs to a specific category is wrong, particularly if no category is suggested by the search terms. ***
*** Showing all categories the product belongs to is probably the best way to go. ***
Indeed, you should have search links on this product page pointing to "search for more salty sweets" and "search for more gelatine-free sweets" too.
*** http://www.example.com/%20-%2020k ***
That suggests you omitted the trailing " quote mark on the URL itself.
| This 137 message thread spans 5 pages: < < 137 ( 1 2 3 4  ) |