Effects of switching between 'www' and 'non-www' for established site?

I did it and boy am I paying for it!

[ichthyous]

I was wondering whether anyone knows definitively whether Google sees the www and non-www versions of pages as completely different content? I recently replaced my old site with a new one and decided to switch to non-www urls after almost 5 years of using www on my domain. I placed some code in htaccess to 301 redirect all www version to the non-www versions, updated all the internal links on the old pages to non-www and replaced my google sitemap with a new one reflecting the changes. While the new site's pages have been indexed fairly quickly/thoroughly they aren't ranking well at all in comparison to the old pages. At the same time my old pages have vanished almost entirely from the index. They are not in supplementals, just gone. The effect of this is to have traffic plummet 90% in one week. My question is, does Google see the new non-www urls as new content or does it recognize the pages as old content without the 'www'?

[g1smd]

Some of the www pages will likely reappear as Supplemental Results in a few weeks time. I have seen that happen before.

I guess that it will take Google at least one PageRank update to reassign everything where it needs to be, but it might take a while to fully recover.

Google may see your site as a brand new site. You didn't mention how many pages it has. If it has many thousands of pages, then I guess that it would be a big blip on their radar.

Some of the new URLs are likely being treated as being duplicates of the content at the old URLs, and I guess that effect may take a week or two to clear.

You need to really make sure that there are no internal links that point to the "wrong" version now. In particular, pay attention, to any links to folders where the trailing / is missing from the URL.

Those may force a double 301 redirect via the wrong domain: domain.com/folder ---> www.domain.com/folder/ ---> domain.com/folder/ and that would be a big problem. Resolution of a missing / happens internally in the server, and over-rides what is in your redirect code.

[flanok]

IMHO I think the way you have done the changeover may make the process take a little longer.

Before, you had all you internal links to your homepage and sitemap pointing to www's
You then 301 redirected your home page from www to non www.
And then changed your internal links also.

If you had left your internal links in place (to www), the PR would have been transered to the non www. (The 301 would have taken care of all internal and backlinks, that were to www)

But because you have also redirected your links to the non www, Google may see these as new internal links and it may take time for google to recaluculate the value of these links to regain your PR.

My Honest Opinion

[ichthyous]

I guess that it will take Google at least one PageRank update to reassign everything where it needs to be, but it might take a while to fully recover.

So then Google does see the switch from www to non-www as new content? I had the 301 redirect in place before I updated the internal links. I saw some pages change to the non-www version quickly while others stayed www. Now all but very few pages are missing from the index. The few that are there are listed as non-www. My page rank hasn't budged throughout the change.

Google may see your site as a brand new site. You didn't mention how many pages it has. If it has many thousands of pages, then I guess that it would be a big blip on their radar.

Combined it has around 1500 pages now. It used to be closer to 2,250...I moved some content offsite to another site and used a 301 redirect to the new site. A few hundred other pages I removed entirely. The pages that are being redirected offsite and the ones dropped are now in supps.

Some of the new URLs are likely being treated as being duplicates of the content at the old URLs, and I guess that effect may take a week or two to clear.

Not too sure about that since they would be in supps if they were seen as dupe content, no? Of course, they may show up in the supps in a week or two, but usually when I make changes I see supps showing up very quickly

You need to really make sure that there are no internal links that point to the "wrong" version now. In particular, pay attention, to any links to folders where the trailing / is missing from the URL.

Ok, I didn't have internal links to folders in the past...only to individual html pages and to the domain name (without trailing slash)...is it an issue that the links to my domain name don't have a slash at the end?

[g1smd]

It might be, but probably isn't for the root, but it can easily be so if it is a folder. You can easily have a link that points to domain.com/folder get redirected to www.domain.com/folder/ first, before being redirected onwards to domain.com/folder/.

Likewise a link pointing to www.domain.com/folder can get get redirected to domain.com/folder/ first, before being redirected onwards to www.domain.com/folder/.

The middle URL is on the default hostname (which may not be the one that you actually wanted to promote!). The default hostname is what is used when resolving the problem of a missing / on the end of a folder URL.

It takes precedence over all other redirects, such as your non-www to www (or vice versa) redirect that is supposed to be solving your Duplicate Content issues, and can make things worse.

Fortunately, the effect is usually easy to see. When running Xenu LinkSleuth over your site and then generating a sitemap, you get twice as many listed pages as you expected, and half of them are shown with a title of 301 Moved Permanently. Or you could just use a HTTP Header Checker and check the response you get for various test URLs.

[edited by: g1smd at 5:33 pm (utc) on Oct. 15, 2006]

[ichthyous]

Not to negate your observations because you are both obviously more experienced at this BUT.... I just ran a pagerank lookup for the few lowest level (old content) pages that had any rank at all...they still rank exactly the same as before. How is it possible for them to stay at a PR 4 when they aren't even present in the site:mysite.com results? Shouldn't my PR have flattened out to 0 or 'not ranked' while Google recalculates the switch from www to non-www? One other thing...when I first put up my new site my traffic skyrocketed and I noticed that Google had started to group the old content pages with the new ones of the same theme...i.e. the old page at example.com/new_york_photos.html was subset under the new page called example.com/photos/new-york/. I thought that was logical and was relieved to see that google didn't treat the old pages as dupes of the new pages or vice versa. After a week or so that changed and the old html pages vanished. Now just the new pages remain, and are ranked much lower than before. Is that an indication that Google started to compare them and see similar content so perhaps it removed one and 'demoted' the other? Usually that happens from the beginning, no? Also, there are no signs of those old pages in supps at all. This all strikes me as very un-google-like behavior

[edited by: tedster at 5:39 pm (utc) on Oct. 15, 2006]
[edit reason] use example.com [/edit]

[g1smd]

I'm seeing several examples of un-google-like behaviour in several areas at the moment: site searches, supplemental results, moved pages, and certain redirects.

Something is brewing.

[ichthyous]

Wonderful...I don't think I can take much more of Google's manic-depressive behavior! I spent a whole summer tearing my site apart to try and get better results...the effect (so far) has been lower traffic. Of course, I think it's still to early to tell. On a positive note...at least changes seem to be updated in Google's index more quickly than they were during the summer.

[ichthyous]

One other question: if two pages are seen as duplicate content do they both go into supps, or just one of them? Does the one that doesn't get 'supped' get 'demoted' in some way because the other one is in supps?

[g1smd]

Usually one goes Supplemental and the other stays in the normal index. However, sometimes both go supplemental.

Once you get the redirect from non-www to www actioned, the www version usually gets back into the normal index within a few weeks, and the non-www verison stays Supplemental for a year (and the cache date is stuck), and then it drops out of the index.

However, it might not always work this way in the future if spammers are already taking advantage of these "dual listings" in any way.