Just read Travis Loncar's YouMoz post and I have a question about Pagination

[danatanseo]

I understand exactly what you are saying Jared. However, here's the problem, the canonical tag is exactly the same....for every single subsequent page in a series across the entire site.

No matter what is searched. The canonical tag remains:

Wouldn't that mean that all search results pages, regardless of search term, are viewed as the same page?

I have heard this discussed before come to think of it. In this case, wouldn't it be proper to block all dynamic search results pages from being crawled or indexed by Google via the htaccess file or robots.txt file?

[JaredMumford]

Hi Dana - Let me see if I understand this correctly:

In question 1 you asked if this would be a duplicate content issue. The canonical tag retains the exact same URL regardless of the search parameter (and resulting search results). Therefore, regardless of the search being made, Google and other crawlers will not index page with a search parameter since the canonical references to the original url (http://www.ccisolutions.com/StoreFront/category/search-return). This means that when Google accidentally lands here http://www.ccisolutions.com/StoreFront/category/search-return?q=countryman it sees the canonical tag and understands that it should not index this page as it is only a variation of the core page.

This would of course be a problem if you actually wanted Google to index every query page. Alternate methods could be to disclude the query parameter in WMT or Robots.  But the canonical is built in for you so that you dont have to.

In situations like this I also like to add site search to analytics and block the query parameter so no query pages show up as landing pages.

[RobertFisher]

Dana

Great and informative question,

Jared

Great Answer

[MagicDude4Eva]

I raised a similar question in the following Q&A - http://www.seomoz.org/q/duplicate-title-tags-with-pagination-and-canonical

My concern or question (we have rel=prev/next) would be more towards what the canoncial should be. There seems to be different opinions:

1. Use the current paginated page as the canonical - in our case GWMT reports duplicate titles (I suppose appending a page-number should sort this out)

2. Use the base search URL as the canonical - perhaps not a bad choice if your site's content changes and Google indexes page 50, but over time you only have results for 40 pages (resulting in an empty result page)

I currently only can conclude that having the prev/next implemented is a good thing as it will hint Google in pagination (in addition to setup the URL parameters in GWMT). I do plan to change the canoncial to the base search URL (and not having multiple paginated URLs) and see how this will affect indexing and SERPs.

[danatanseo]

Hi Gerd,

Yes, this is a separate issue we are also struggling with on the site. I believe Travis' YouMoz post from yesterday made a pretty good case for using multiple paginated URLs, and he even illustrated how to accomplish this with sorting parameters like "color" and "price"

You raise a very good point about duplicate titles and descriptions potentially being a problem in this scenario.

Does anyone have any ideas about how to handle that? Could the backend be programmed to dynamically create unique titles and descriiptions based on some rules for naming conventions? (assuming you have access to that level of the code of course)

Really interested to know some points of view on this!

Dana

[danatanseo]

Yes, Jared, this is a great answer. I understand completed. It looks like we are ok then with Google Custom Search as it is. Thanks so much for your thoughtful answer. Now, if we can only get our paginated category pages sorted out, we'll be on the right track!

[JaredMumford]

For the most part, I would choose to use rel=prev/next for pagination, including both pagination with dynamic urls and static URLs. There are some cases (as with this original thread question) where you should use canonical, but as a whole you should use rel=prev/next.

The best way to explain it is:

Rel Prev/Next:

Your site: Hi Google, I have all of these pages that very similar so I'm just letting you know that I only have duplicate content here for usability reasons and am in no way inferring that you should index all of these pages and rank them #1!

Google: Ok great, thanks for letting us know. We'll index the pages we feel are appropriate, but you wont get penalized for duplicate content. We may only index and serve one page, "page 1", or we may index multiple pages. Thanks for letting us know.

Canonical:

Your site: Hi Google, I have all these paginated pages that look like duplicate content, please do not include any of them in your index, and don't penalize me for duplicate content. For the record, the page you should index is Page 1 and no other pages.Any links that point to the paginated pages should be counted towards Page 1*.

Google: Great, no matter what we will not index any pagination and only Page 1.

With rel=next you are simply letting Google know, but not dictating how Google should act on the situation. If fact with ecomm sites, youll find that a lot of timees when you use rel=next, Google will actually index the 'view all' page if you have "view all" as an option around your pagination links

*many articles suggest that link juice is passed to the canonical URL - I'm have not seen any direct evidence of this but is worth a different discussion.

[JaredMumford]

This would all depend on what the site was built on, and the flexibility. There's no questions that this can be done. "Back in the day" we had a few sites that had tens of thousands of page due to sorting, and we had everything generated including:

Title, meta d, meta k, breadcrmb, H1 and short description.

Those were the days!!!

[danatanseo]

Considering that the larger of the two sites I work on is on a platform from 1996, I might actually be living "back in the day!" lol - Thanks again Jared!

[JaredMumford]