SEO implications of CMS archive sort order

[Sgt_Kickaxe]

CMS = content management system (ie:wordpress, drupal, etc)
Default sort order on most CMS's is Desc, meaning newest content first.
SEO implications = purely the effects on search results, for people the best would obviously be to include an option for both but I'm interested in hearing your thoughts about JUST the SEO implications of one vs the other.

My observations:
- With the default a new entry constantly pushes an old entry to another page. If you have 50 pages in a category at least 49 of them change with each new post.
- Search results leading to a category page become outdated as the content is no longer there, this happens faster on sites with frequent updates.
- Image search results may no longer show the image a visitor was looking for if post images are included on your archive pages.

Pros for using Desc order (newest first)
- Newest content is posted first, no need to click to the end to find it (on page option for both or better pagination, negates this).
- Each archive page is 'refreshed' as the content changes over time.
- Search engines account for, and adjust for, any default that is widely used.

Pros for using Asc order (oldest first)
- The content never changes on any given page.
- Stable link graph, if archive page x leads to article page y the link will not expire from page x.
- images on page x will always be on page x, no frustrated searchers landing on an outdated page.

Assuming content has solid pagination and perhaps an option for a user to change the order, and assuming interlinking is optimized between similar articles, I'm leaning towards using the non default Asc order, oldest first, for SEO purposes. If you can think of a reason for, or against, either option (for SEO and rankings reasons) I'd like to hear your thoughts.

[g1smd]

Yes, this problem also affects forums and blogs, in fact any type of CMS.

It's not just the "archive" pages, but "new products" and "recent updates" pages as well.

[deadsea]

It makes it hard to build an audience if your site never appears to change and new content is hidden on page 53. Oldest first isn't a good choice for blogs that want a following, and it would kill off forum discussions pretty quick.

I'm a fan of two lists on the home page: "newest content" and "best content" where the best content is hand picked or picked by some sort of interest metric.

[g1smd]

Current systems show URL with "page-1" in it with the newest content on it. The content of the "page-1" URL is therefore always changing. After every "n items per page" have been added to "page-1", the content that was at the "page-1" URL is now all on the "page-2" URL - and could be seen as being a duplicate of the "page-1" URL unless "page-1" has been respidered in that time.

What the OP is saying is that the "page-1" URL should show the oldest content, and new content should be added at the "page n+1" URL.

However, it is very important to separate out two concepts here: the actual URLs used for page numbering and the separate issue as to what order the content is listed in.

I do agree that the page with "page 1" in the URL should show the oldest content. This makes the content static at that URL. I do not agree to show "page 1" first. The page with the newest content should be showed first.

In this case, you would show the "page n+1" FIRST and work backwards by page numbers to see the older content.

I wrote a very long post about this a couple of years ago, including having a page called "latest" that shows only the most recent stuff and IS constantly changing in content.

So, with this scheme, today I see the newest items on the "latest" page, then click through to page 53 to see some slightly older entries, then 52, then 51, all the way back to page 1 with the oldest content.

A few months later I return to the site. The newest stuff is shown on the "latest" page, and when I click through to see older information, the next page is page 80, and I can click through to 79, then 78 and so on.

If I had previously linked to an item on "page 20" that content will still be on "page 20".

New content is added on higher numbered pages. The highest numbered page should be shown first.

[rlange]

g1smd wrote:
In this case, you would show the "page n+1" FIRST and work backwards by page numbers to see the older content.

I've thought the same thing for a while. Just now, however, I realized a flaw with that method...

Let's say you display 10 items per page with this method. You're starting out fresh, so your first "post" is the only item on Page 1. No problem so far. You add 9 more posts to fill up Page 1. Post 11 starts off Page 2, which will be displayed as the most recent page. Uh oh. Now you've only got a single item on the most recent page, with a link to Page 1 that contains 10 more posts.

That sort of situation may be a bit confusing for visitors. The "Latest" page you mention might be an acceptable workaround, but I don't know.

--
Ryan

[g1smd]

Yes, that would be a problem but I actually do something slightly different, as hinted at in the previous post.

The "latest" page doesn't contain exactly ten items, nor does it contain a variable number from 1 to 10. It can, in fact, contain up to 20 items, and never contains less than 11 entries (unless it is a new site with less than 11 entries in total). The numbered "page-n" pages always contain exactly 10 items. As soon as item 21 is about to be added to the "latest" page, the oldest 10 items are moved from there to a brand new "page n+1" page. The only disadvantage here is that the 10 "fairly recent" items disappear from search visibility for a short while after leaving the "latest" page and before the new "page n+1" page is indexed.

There's a slightly more sophisticated method that can be used, and this is one which avoids that scenario. In this case, the "latest" page works exactly as before, with up to 20 items. It links to the highest numbered full "page-n" page. That numbered page and all numbered "page-n" pages with lower page numbers always contain exactly 10 items, as before. However, the slight change in implementation here is that the new "page n+1" page is built in the background, growing from 1 to 10 items over time. It is linked ONLY from the highest numbered full "page-n" page. When the content on the "page n+1" page reaches 10 items, the outgoing link on the "latest" page is changed to point at this page as it has now become the highest numbered "full" page. At the same time, the number of items on the "latest" page drops from 20 to 11.

This scenario gives static content on all except two pages. The highest numbered "page n+1" page grows and then becomes static. The "latest" page is always changing and always has between 11 and 20 items. At the point it is about to have 21 items, its outgoing link changes to point to that new "page n+1" page and the number of items on the latest page drops from 20 to 11 and a new "page n+1" page is created with just one item on it.

This means that the only time that a visitor arriving from a search result may fail to immediately find the item they wanted is in the days right after 10 items were moved off the "latest" page and the item they were looking for was one of the 10 items that was moved. In this case the visitor is only ever likely to be one or two clicks away from the new location, depending on how long Google has kept details of the old page version content in its index (the old style "historical" supplemental results). However, Google should already have a copy of the "page n+1" page from recent days indexed too so this might not be such a big issue.

There's yet a further modification that can alleviate that problem. When "page n+1" becomes full, delay dropping the oldest 10 items off the "latest" page for a few days. Also delay changing the outgoing link for a few days in order for Google to fully index the completed "page n+1" page. At the same time, since new items are still being added, build a "page n+2" page. In this scenario "page n+1" always has 10 items (and is not linked from the "latest" page) and "page n+2" is the one that grows from 1 to 10 items.

[Sgt_Kickaxe]

g1smd, best of both worlds, I like it. I wonder if it would be hard to alter the more popular CMS's to accomplish this. I don't see any out of the box solutions or plugins for most systems, yet.

[Sgt_Kickaxe]

You know, this recent pagerank update tells you how importance flows through your CMS order. On Page 1 of categories with PR4 I'm seeing a drop to PR2 on every page that is linked to. ie: if Page 1 of a category links to pages 2,3,4,5,6,7 ... last then pages 2,3,4,5,6,7 and last all have PR2 and pages 8,9,10 etc have PR ranging from n/a to 1.

In terms of performance then it would make sense to have your BEST content on page 1 of an archive or at the very least on one of the pages in the initial pagination options. A PR 2 link is a PR2 link be it internal or from another site.

Assigning an importance score to content pages would help create such a layout but it wouldn't help visitors much so is impractical. The next best option is to figure out whats most important, your oldest content or your newest... and to make sure it's reachable by the initial pagination tree. Making it static to that page, even better.

Thanks for the ideas.

edit: here on WW I'm seeing n/a as pagerank for paginated links, I haven't checked but the theory with that might be to block all but the top page from your internal link graph?

[g1smd]

In terms of performance then it would make sense to have your BEST content on page 1 of an archive or at the very least on one of the pages in the initial pagination options.

Yes, but in this case we're talking about "page 1" only in as far as its click number from the root page, but with that page having a high page number in its URL, and newer pages being assigned an even higher number in their URL.