Google Switch Https indexing and ranking problems

[mimo]

Hi,

Since Google switched to the https version, our company website (http://we.example.com) indexing switched to the https version (https://www.example.com).

After that the ranking on Google dropped for almost every keyword.
On each page from years there is the correct meta rel canonical. No spam, and WMT is ok.
Could please help about it?

Thanks

[edited by: aakk9999 at 12:19 pm (utc) on Nov 8, 2013]
[edit reason] Examplified, no URLs please as per ToS [/edit]

[aakk9999]

Google switch to https could not cause your site to be indexed under https. To me it seems there could be a technical error somewhere that allowed https URLs to leak out.

I am wondering whether you could provide more details without being specific with regards to your domain name and niche. In particular:

- How are you linking internally within the site, are you using http or https?
- From what you have described, this is not just http/https problem as it appears that the subdomain changed too (we. into www. ) Have you done any changes in this area?
- You say "since Google switched to https version" - can you give a better timeline on when your issues started?
- Have you done anything to try to rectify the issue?

[mimo]

Thanks for reply
- The internal links are all in http, and in https. If you are https are in https (they are all relatives)
- No changes have been made and the subdomain is in that way from 8 years: the main url has always been http://we.example.com
- Google started this indexing switch around the 15 October

[edited by: aakk9999 at 2:04 pm (utc) on Nov 8, 2013]
[edit reason] De-linked URL [/edit]

[Robert Charlton]

I've been expecting to see lots of reports of http /https problems since Google has gone https.

At the least, try a canonical link tag on each page, with the appropriate http protocol included.

If possible, also go to absolute urls, including the http protocols, in your internal nav.

What really needs to be done, though, is proper canonicalization on your server. We've had several recent discussions on canonicalization. Perhaps someone can jump in with a thread reference.

[bumpski]

with the appropriate http protocol included

I agree this is critical. Not long ago Google started indexing one of my sites as both https and http. My webhost, who hosts millions of sites, by default, allows all pages to be crawled with or without https.
This webhost also provides a "secret" subdomain path to all sites, such as:
"domain-abbreviation.giantwebhost.com".
Google is crawling hundreds of thousands of sites using three paths, http, https, and the "secret" domain path. This secret domain path is very easy to overlook when researching duplicate content problems, and when implementing canonical meta-tag code.

I believe I triggered Google to crawl using https myself. After disabling all the browser warnings about the certificates, I was surfing the site using https and I'm sure I visited Google somewhere in that process. So Google seeing the https referrer started crawling. At the time the canonical code on the site did not take into account https, but after adding http to the canonical tag google de-indexed the https pages after several weeks.

Actually I'm tempted to get a secure certificate to allow visitors to surf in whatever manner they desire, without the certificate warnings. That's why the canonical tag solution, versus using 301 redirects.

[Robert Charlton]

aakk9999 wrote...

Google switch to https could not cause your site to be indexed under https. To me it seems there could be a technical error somewhere that allowed https URLs to leak out.

I wrote...

I've been expecting to see lots of reports of http /https problems since Google has gone https.

For clarification, aakk9999 is correct. The protocol "leakage" shouldn't occur between pages on which the protocol has been declared... ie, between Google https pages and your http pages.

The leakage generally occurs in an internal nav situation using relative urls, where no protocol is specified in the destination page, but where https may be specified in the linking page.

I'm not sure I understand bumpski's post at all, but that's what prompted me to revisit my post.

[JD_Toims]

The protocol "leakage" shouldn't occur between pages on which the protocol has been declared.

I wouldn't be 100% surprised if Google decided to test https responses much they way they do by requesting "unknown" URLs or Query_String parameters, meaning there may not have been a "leak" at all.

[bumpski]

Since Google switched to the https version, our company website (http://we.example.com) indexing switched to the https version (https://www.example.com).

For clarification I assume this means that:
http://www.example.com/example.htm AND
[example.com...]
are both showing up in the Google index, causing a duplicate content problem.

In this case the rel=canonical tag should contain the desired path to these identical pages; typically
<link rel="canonical" href="http://www.example.com/example.htm">
If Google sees this meta-tag it will only index the page using

http://

, the duplicate page accessed through [example.com...] will be removed from Google's index.

[lucy24]

Thought. gwt has a setting for "preferred domain", meaning with or without www. Wouldn't it be useful to have a "preferred protocol" setting as well? Where do they hide their suggestion box?

[mimo]

Since 2009 I've:
[example.com...] canonical to http://we.example.com
[we.example.com...] canonical to http://we.example.com

Since 2005 I've:
http://www.example.com Redirect 301 to http://we.example.com

Preferred domain: you can't choose subdomain like (we.example.com) as preferred domain in WMT when you've 301 from www--->subdomain

Thanks you for your support, but unfortunately, I think, the only thing left for me to do is to change my primary domain:
- from we.example.com to www.example.com

[edited by: aakk9999 at 3:55 pm (utc) on Nov 9, 2013]
[edit reason] Unlinked URLs [/edit]

[aakk9999]

If you really want that we. subdomain is your canonical, there is no issue with this, but you need to make sure that:

- you are internally linking to the correct URL (including protocol http)
- you have 301 redirect set up from other variants of the same page to your canonical page

The first step I would do is to check what has Google indexed, by using site: operator. The command:

site:example.com -site:we.example.com
will give you the list of all URLs that are indexed that do not belong to we. subdomain.

site:example.com inurl:https
will give you a list of all URLs that are indexed as https

site:we.example.com inurl:https
will give you a list of all URLs on we. subdomain that are indexed as https

And so on, you can use various combinations of operators above to slice and dice and get the picture of what has Google indexed.