Google Ignoring Canonical Tag for Hundreds of Sites

[redgatst]

Bazaar Voice provides a pretty easy-to-use product review solution for websites (especially sites on Magento): https://www.magentocommerce.com/magento-connect/bazaarvoice-conversations-1.html

If your product has over a certain number of reviews/questions, the plugin cuts off the number of reviews/questions that appear on the page.

To see the reviews/questions that are cut off, you have to click the plugin's next or back function. The next/back buttons' URLs have a parameter of "bvstate....."

I have noticed Google is indexing this "bvstate..." URL for hundreds of sites, even with the proper rel canonical tag in place.

Here is an example with Microsoft:

http://webcache.googleusercontent.com/search?q=cache:zcxT7MRHHREJ:www.microsoftstore.com/store/msusa/en_US/pdp/Surface-Book/productID.325716000%3Fbvstate%3Dpg:8/ct:r+&cd=2&hl=en&ct=clnk&gl=us

My website is seeing hundreds of these "bvstate" urls being indexed even though we have a proper rel canonical tag in place. It seems that Google is ignoring the canonical tag. In Webmaster Console, the main source of my duplicate titles/metas in the HTML improvements section is the "bvstate" URLs.

I don't necessarily want to block "bvstate" in the robots.txt as it will prohibit Google from seeing the reviews that were cutoff. Same response for prohibiting Google from crawling "bvstate" in Paramters section of Webmaster Console.

Should I just keep my fingers crossed that Google honors the rel canonical tag?

Home Depot is another site that has this same issue:

http://webcache.googleusercontent.com/search?q=cache:k0MBLFcu2PoJ:www.homedepot.com/p/DUROCK-Next-Gen-1-2-in-x-3-ft-x-5-ft-Cement-Board-172965/202263276%23!bvstate%3Dct:r/pg:2/st:p/id:202263276+&cd=1&hl=en&ct=clnk&gl=us

[LoganRay]

Do you have different examples? The Home Depot link doesn't work when trying to view the actual page on the site. With the Microsoft link, the canonical is working, as the version with the parameter is not indexed in Google, but the canonical version is indexed, which is what I would expect for a canonical that is being obeyed.

[redgatst]

Hey Logan -

The Microsoft canonical is not being obeyed. The canonical tag points to the one representative URL for the product whereas the "bvstate" URL is shown as being cached.

http://prntscr.com/aqwnhu

http://prntscr.com/aqwo4c

If you do a search in Google for inurl:"bvstate" , you will see hundreds of sites like H&R Block, Kohls, etc.

[LoganRay]

I just ran this query for bvstate URLs indexed for the H&R Block site. Mozbar shows canonical tags with bvstate in them, and Screaming Frog finds no canonical tags at all. There is a deeper issue that is not simply Google ignoring them.

[redgatst]

Yeah, it's very strange... if you view-source on the BVSTATE url that is cached, the proper canonical tag is in there. Don't know why toolbar apps like Mozbar show otherwise. I think you're right, must be a deeper issue.

[redgatst]

I think I found out what is going on.

I have found that the source code does contain the proper rel canonical tag.

However, the "bazaar voice" plugin generates a code snippet that appears in the page's body where it features a [base_url]. The [base_url] should match up with the canonical tag. For some reason, it isn't. The [base_url] that is generated contains the "bvstate" parameter.

Tools like the Mozbar, and I believe even Googlebot, are extracting out overriding the rel canonical tag with the [base_url] that appears in the code.

Complex!

[ThomasHarvey]

I have had something similar, this is response I received:

You don’t have canonical tags on the URL and that’s expected.

On pages where BVSEO is implemented, canonical tags must be updated or removed when the product contains more than one page (more than eight) of reviews. BVSEO paginates the product page so all reviews are in the search engines’ index. Canonical tags that point away from a pagination URL will cause search engines to ignore the paginated content.

When any of the BVSEO pagination parameters are present (bvstate, bvrrp, bvqap, bvsyp, bvpage), do one of the following:

•Remove the canonical tag. This is the most common, recommended solution.

•Append the "name=value" pair to the canonical URL.