Roger found Duplicate Page Content, but all my pages are canonicalized

[solarian]

This post is deleted!

[riyas_]

Hi Lucian Sava

I was just gone through you website and found that your website is using 3 domains for different languages (English, Hungarian & Romanian) and each domain is also hosting another duplicate sub-directory version of other languages

So, These are the 3 domains

English: http://starshiners.co.uk (Having another 2 versions)

Romanian: http://starshiners.ro (Having another 2 versions)

Hungarian: http://starshiners.hu (Having another 2 versions)

I'm giving a short name for each versions bellow

Domain .CO.UK

1. A-EN http://starshiners.co.uk (Original)

2. A-HU http://starshiners.co.uk/hu/

3. A-RO http://starshiners.co.uk/ro/

Domain .RO

4. B-EN http://starshiners.ro/en/

5. B-HU http://starshiners.ro/hu

6. B-RO http://starshiners.ro (Original)

Domain .HU

7. C-EN http://starshiners.hu/en/

8. C-HU http://starshiners.hu (Original)

9. C-RO http://starshiners.hu/ro/

Now just taking English website as an example (CO.UK)

A-EN rel=canonical to A-EN (Itself)

A-HU rel=canonical to C-HU

A-RO rel=canonical  to B-RO

In this case I guess Roger will consider A-HU & A-RO as the duplicate of A-EN (Which is rel=canonical  to itself and all others are rel=canonical  to external)

The situation will be same for other two domains (.RO &.HU)

It will look something like this

Original Page:  http://starshiners.ro/accesorii/cercei-spelbound-turquoise

Duplicate Page:  UR http://starshiners.ro/en/accesories/spelbound-turquoise-earrings

Duplicate Page:  http://starshiners.ro/hu/kiegeszitok/turkiz-spelbound-fulbevalo

How to fix this? (My suggestion)

How about removing the sub-directory versions from all individual domains (I guess you prefer different domains for different versions)

For English make this domain (http://starshiners.co.uk) as default version
For Hungarian this, http://starshiners.hu For Romanian this, http://starshiners.ro

and link each other for different version

Also in order to avoid duplicate issues within versions use proper rel="alternate" hreflang="x"

Above are just my thoughts regarding this issue, do wait for others suggestions too

riyas

[Dr-Pete]

Muhammed's right that rel="alternate" hreflang would be a good bet for your international domains, but our system doesn't look across domains for duplicates, so I don't think that explains the issue you're seeing.

It looks like we may be detecting some of the pages as near duplicates - for example, color variations of your products. Only a few words vary and the pages are mostly images. While these pages are legitimate, they could look "thin" to Google at large scale.

You've also got some variations based on URL parameters (like "size=") that seem to be producing "No products found" pages. Those could look like Soft 404s to Google. They're canonicaled to yet another page, but I have a feeling that the combination and paths may be getting too complex.

Unfortunately, these situations are fairly different and reasonably complex. I don't have an easy answer, and I can't tell for sure if the canonical tags in place are pointing to the right pages. I suspect, though, that rel=canonical may not be the best solution to some of these problems.

[solarian]

Hi, Peter, yes, indeed, my question was about duplicates within one domain. However, the problem Riyas mention is a real one and it was my next question here I will answer him separately.

I would try this:

I have to url with duplicate cantent:

http://starshiners.com/ro/salopete?color=Jeans+&dir=desc&order=date&size=S&type=grid

and

http://starshiners.com/ro/salopete?color=Dantela&dir=desc&order=date&size=S&type=grid

where the color parameters are the only difference in url structure.

Both pages are canonized to http://starshiners.com/ro/salopete

Could be a solution this?:

1. get rid of dynamic parameters and tell the system to form urllike

http://starshiners.com/ro/salopete/Dantela/size-S/type-grid/

instead of http://starshiners.com/ro/salopete?color=Dantela&dir=desc&order=date&size=S&type=grid

and insert in this new url a rel canonic to http://starshiners.com/ro/salopete/Dantela/size-S/type-grid/???

Or do you have maybe other suggestion?

thanks!!!

[solarian]

This post is deleted!

[riyas_]

SEO duplicate issues with your current website (with 12 versions of each URL) are

1. For each URL on the website, there are 3 other exact duplicate pages available on the web (You partially fixed it with rel="canonical")

2. There is 8 other regional duplicate version, each one will create duplicate issues with the first URL I mentioned

3. Your link popularity is diluted  through so many domains

It will be much better if you can make it to any one version. Either a domain version or a sub-directory version

What's exactly in my mind

Since you own a dot.com domain I suggest you to use starshiners.com as your base English version and create sub-directory versions for all other languages.

English Version: starshiners.com (Or you can even 301 entire starshiners.com to starshiners.com/en)

Romanian Version: starshiners.com/ro

Hungarian Version: starshiners.com/hu

Now 301 All Other Domains

301 Redirect starshiners.CO.UK visitors to starshiners.COM

301 Redirect starshiners.RO visitors to starshiners.COM/ro

301 Redirect starshiners.HU visitors to starshiners.COM/hu

At this stage I guess the only SEO duplicate issue on the site will be each URL creating duplicate with 2 other language versions

Here you can easily fix that using the benefit of rel="alternate" hreflang (refer bellow links)

http://support.google.com/webmasters/bin/answer.py?hl=en&answer=2620865
http://support.google.com/webmasters/bin/answer.py?hl=en&answer=189077 https://sites.google.com/site/tomihasa/google-language-codes

What you think?

[Dr-Pete]

It's a bit tricky, because those aren't really duplicates, which may be throwing off our crawlers. I'm also noticing that Google doesn't seem to be indexing the canonical version, which could be a bad sign.

It would probably be better to block some of the parameters from crawling (including with META NOINDEX) and then canonical to more similar pages. I think Google is going to have issues with how dissimilar the pages you're trying to canonicalize are, and it might be better to just prevent the variations from being crawled at all.

Unfortunately, it's a pretty tricky problem on a site that's this dynamic. I don't have a simple answer for you.