• http://www.tiv.net/ Gregory

    Thank you Vanessa, all very clear and not confusing at all.

    The question is: why do we want Google to index search result pages at all?
    I guess, the content on those pages is a repetition of what’s on individual product pages. So, why not to make all of them “noindex, follow”, and skip all the markup you mentioned?

    … after I wrote that, I started thinking that maybe the best would be both: markup, AND still noindex. So, Google will know how to travel and group, and “pass the juice”. Does that make any sense?

    Thank you!

  • http://ninebyblue.com/ Vanessa Fox

    Definitely for search results, the best bet is to noindex,follow them. This is more for category pages that have product listings. This page, for instance, wouldn’t be considered search results:
    http://www.rei.com/category/4500304

    It would be the page that REI would want to rank for a [snowboards] search.

    Whereas this page would be noindexed:
    http://www.rei.com/search?query=snowboards

  • jrock

    Hi Vanessa,

    Thank you. Your post couldn’t come at a better time. And what a great name by the way. Vanessa Fox. Nice.

    My wife and I are launching a new e-commerce site and many products are very similar and come from the same line from the same brand. There will some duplicate content in some product descriptions. For instance, we have 6 variations of probitoics from one brand.

    Any tips on home some newcomers to e-commerce like us coul I feel like I should jsut forward your blog post to our developers in India and hope they get everything you saying.

  • jrock

    Sorry my son just jumped on keyboard before I could edit my post.

    I was wondering what some good first steps would be to apply pagination. It really seems like we need it.

    Thanks again – great post

    J

  • http://eywu.com Eric Wu 吴 ¯_(ツ)_/¯

    I’m wondering what would be the negative ramifications if you didn’t dynamically insert the query parameters based on the fetched URL?

    While it’s simple for some developers to do this, I could see that it would be more difficult for others. Especially considering that you would have to sanitize the query parameters since the values are being rendered into the HTML, and query parameters could easily be injected by a potentially malicious user.

    Given the example above, assuming you don’t want the “&referrer=twitter” parameters to be indexed, if you dropped the “&referrer=twitter” effectively breaking the chain, would Googlebot fetch the canonical? And then in fetching the canonical it would see the corresponding next/prev that are now correctly in the chain?

    Seems like it might break some of the crawl efficiency, but the trade off in dev hours and maintenance seems like it might be a worthwhile trade-off.

  • http://www.returnondigital.com/about-us/dave-ashworth.php dave_ashworth

    It has been interesting to see how Google have used the rel/prev tags as an issue arises with the use of WordPress as a CMS – we have seen a client site transfer from static site with rankings across many pages for different keywords lose all but home page rankings due to the way in which WP implements the tags – if you set up a series of pages it automatically adds the tags in to aid users with sceren readers/disabilities etc to help navigation.

    The non techies amongst us would be hard pressed to work out why, I did write a blog post about it:

    http://www.returnondigital.com/blog/wordpress-default-canonical-issue-and-how-to-avoid-it

    Would be interesting to know if there was any dialogue between WP and Google to avoid this issue in future versions of WP

  • https://plus.google.com/105720802884434674393?rel=author Kenichi Suzuki; 鈴木謙一

    Thank you for a very helpful post, Vanessa. I have a question.

    You says,
    > The pagination attributes can only link together URLs with matching parameters.

    How about a case like this:
    http://www.site.com/products
    http://www.site.com/products?page=2
    http://www.site.com/products?page=3

    Only the first page has no parameter but pages after it have parameters. Is this use fine?

  • http://ninebyblue.com/ Vanessa Fox

    Eric,

    I talked to Google about this and they said they paginate before they process canonicals. I asked specifically about your scenario, thinking it would make sense to canonicalize, then paginate from a processing perspective. But they told me their backend system doesn’t work that way. So in the case you describe, the &referrer=twitter page wouldn’t be consolidated into the paginated entity.

  • Jerome

    Kenichi > I have exactly the same question !

    What do we do for the first page, that can be accessed either with ?page=1 or with no parameter ?

    On page 2, should we put rel=prev on http://www.site.com/products?page=1, or on http://www.site.com/products ?

    Should page 1 be canonicalized to http://www.site.com/products?page=1, or http://www.site.com/products ?

    Buzbez

  • http://eric eric

    A little late, but hopefully someone will answer my question. My website already has rel=”prev” and rel=”next” added automatically to paginated pages though Yoasts SEO plugin. However, since these pages are not set to noindex they are clogging up the search results with duplicate title tags and creating duplicate title warnings inside google webmaster tools.

    I was under the impression when Google first introduced rel=”prev” and rel=”next” a few months ago, they would follow a similar principle to using a canonical tag and therefore paginated results shouldn’t appear within Google’s search results.

    Am I wrong?

  • http://ninebyblue.com/ Vanessa Fox

    Eric,

    Do you have examples of these search results? You’ll likely see the duplicate titles message in webmaster tools (since they are technically duplicate titles) but you should only see one page from the set (nearly always page 1 unless another page is significantly more relevant) in search results.

    The exception should be if you’re doing a site: or inurl: search. All the pages in the set will likely show up with those, since they are all in the index.

    Kenichi, Jerome -

    In this case, /products is fine because that page will point to page=2 and page2 will point to /products.

  • Pet Assure

    Great and thorough article! One question. You write:
    “A page can’t contain multiple rel=”next” attributes.”

    I am assuming that only means multiple rel=”next” to two separate links, but 2 links to the same page (When on page 2, the number 3 AND Next button will both have rel=”next”) is ok, correct?

  • http://twitter.com/djfriedmann daniel friedmann

    Hi, I’ve been using pagination to bundle UGC in a Q&A forum where users have used the same title for their questions – in order to avoid keyword cannibalization.
    To illustrate, 10 users have posted a question called “How do I do improve SEO”
    which are linked as
    site.com/HowdoIimproveSEO-1.html
    site.com/HowdoIimproveSEO-2.htmlsite.com/HowdoIimproveSEO-3.html….Of course, each questions is somewhat different and so are the answers.Do you think this is an appropriate use for pagination (prev and next)?
    Or would you suggest a different tactic (moderation / curation, noindex on older questions, ?)