An Outside-In View of Subdomains – Answers

The first blog post on An Outside-In View of Subdomains - Questions is essential reading before continuing here.  That post encouraged you to answer three questions that may well separate our readers into two groups with very different views of domains.

My answers to those 3 key questions are:

• Is domain authority real?        No
• Are links between subdomains internal?        Yes
• Can subdomains be separate 'websites'?        Yes

In a sense whether you agree or have a different viewpoint is somewhat academic.  What really is important is how Google views all this.

Google's Answers On Domains And Subdomains

Since it is Google's views that will determine how websites and web pages will rank in SERP's for keyword queries, that's what counts.  Unfortunately not too much is revealed by Google on how domains and subdomains may influence rankings.

Trying to back-engineer how Google's algorithms work is an impossible task.  The algorithms are very complex and are in constant evolution.  Making tests to try to measure the effect of different factors may take some time since changes are not indexed immediately.

Can You See The Website?

In the absence of clear signals from Google, all we can do is work with the simplest hypothesis that fits the facts.  Although the SEOmoz view suggested that the domain itself is an important factor in determining rankings, there is nothing to prove that and indeed there are some findings that would put that assertion in question.

If your eyes are beginning to glaze over and you are wondering what all the fuss is about, then be very careful because this is the 'I see the pretty girl' moment in this discussion.  If you believe that a website or domain is a clearly defined group of web pages that have an affinity that is more than just the hyperlinks that connect them, then you are seeing the pretty girl.

Rand Fishkin eloquently explained all that in the video.  His picture of domains have boundaries around them to clearly identify what is in the domain and what is not.

Those boundaries are the 'pretty girl'.  However some of us believe that those boundaries are not really there.   Convincing you that you are not really seeing that pretty girl will not be easy.  However let us examine two aspects of all this to shake your faith.  Remember this is not just a philosophical discussion.  The view you have will cause you to do different things and they will have different degrees of effectiveness in getting search engine visibility..

The simplest view is that Google indexes individual web pages and registers all the hyperlinks that emanate from these individual web pages.  In other words the Internet looks somewhat like the image below.  There are no tightly defined websites such that individual web pages are either in or outside a clearly defined 'space' that is called a website.

That is not to say that individual web pages are all completely independent.  Instead any web page has a URL that is like a label.  Some of these labels have similarities.  For example a group of web pages may all include in their 'labels' the characters mysite.com.  You might consider this group of web pages as a loose bundle of web pages.  Their closeness if you try to visualize the 'placing' of these web pages is determined solely by the hyperlinks that interconnect these web pages.

If you accept that a website is this loose bundle of web pages, then you will not make mistakes.  If you assume that domains and subdomains are tightly defined groups of web pages, then you may run into trouble.  Let's explore why.

Camel-humping

The first aspect to try to shake your faith is a rather subtle one.  Camel-humping describes the situation that you may sometimes write your URL for a web page as http://mysite.com/index.html and sometimes as http://MySite.com/index.html  Beware because Google will index these as two distinct URLs.  With luck your hosting service arranges that either spelling will ensure your visitor will see the expected web page.  But not always.  Sometimes a 404 error will turn up.  There's a good discussion about all this in the Google Webmaster Tools Forum entitled lower case, UPPER CASE, or Mixed Case urls...is there a standard to follow?  Of course Google will only index URLs it finds in its crawling but if someone types 'the wrong one' and there is no resulting 404 error then the URL will be indexed.

If you believe that a domain or website is a clearly defined group, how do you treat this possible infinity of URLs that may exist.  Does your website include all the variants that have already been crawled?  Does your website include all the theoretical ones that might come up in the future?  These are tough questions that are better avoided.

www or not-www

If you found camel-humping a somewhat subtle topic, the next one is very practical and has important implications.  Indeed Rand Fishkin skated quickly over this one in his video.  He did mention that www.mysite.com is a subdomain.  That does of course mean that the two URLs http://www.mysite.com/index.html and http://mysite.com/index.html may well both be indexed.  You might say that one is in a subdomain and one is not.

There is a great deal of discussion and help available on how to avoid the problems created by this double indexing.  For our present discussion, the only challenging question to take away from all that is what constitutes the website here. Is it both the www and the non-www versions?  If the website does not include both, how do we define the 'space' where the excluded URLs are?  Even if you retain the notion of a website, that picture does not really help in practical problems and may not be the way Google sees it.

Our three questions

At this point, perhaps some readers are beginning to abandon the notion of domain authority as something that applies to the domain and may not apply to particular subdomains.  All that matters is what is determined by the hyperlink interconnections among web pages.

If this is so, then subdomains should be regarded as equally useful and with equal authority as subdirectories or subfolders.  The slight disrespect for subdomains should be eliminated and they should be considered and used where they can create advantages, even though they may require a little more effort to set up.

What about our two remaining questions and in particular, how would Google answer them?

• Are links between subdomains internal?
• Can subdomains be separate 'websites'?

The answers may change with time, which points to why simple notions on domains and subdomains may run you into trouble.  However currently the following is true:

• Links between subdomains are identified by Google as internal links.
• Subdomains have the appearance of independent 'websites' since they can appear in the SERP ranking lists as additional entries even though the main domain itself may already be represented by two entries. 

It is almost as if the Google algorithms assume a slightly greater separation between web pages that appear on different subdomains.  This extra appearance for a subdomain in the SERP does increase the possibility that a prospect may click through to some other part of the domain space.

Geotargeting a subdomain

The Google Help on this does explain that geotargeting is possible for both subdomains and sub directories.  It does not mention that the geotargeted subdomain may appear in SERP listings in addition to entries for the main domain.  In the video there is a somewhat surprising comment that a given web page on the subdomain may be presented as a web page from the targeted country or region.

This has been confirmed in the following real-life case.  The main domain website was a .com domain and was not geo-targeted since the aim was to sell around the world.    The website was hosted outside Australia but did refer to products that were associated with Australia.  A subdomain was set up geotargeted on Australia but still hosted outside Australia.  It is indeed the case that the subdomain web pages are now identified as web pages from Australia and thus would appear in searches where the searcher specifies only web pages from Australia.

Geotargeting a subdirectory

What we are seeing for the subdomain web page can be contrasted with what may well happen if a subdirectory had been geotargeted on Australia.  Since it is part of the main domain it would not have been shown as an additional item in the SERPs.

Also it could be that it will not be defined as a web page from Australia, although we have not tested that.  If that is true, then it would not appear in searches where the searcher specifies that only web pages from Australia are considered. If anyone has proved that this is not true, then it would be interesting to have their comments on that.

Summary

It thus appears that the conventional wisdom of geotargeting subdirectories rather than subfolders may be questionable advice.  This is because Google treats web pages from subdomains as being somewhat separate from the domain web pages.  If you accept the view that domain authority is not a real factor, then there should be no reluctance in working with subdomain web pages rather than subdirectory web pages.  Provided both have  equal numbers of inlinks from the rest of the website, both should have equal ranking in searches.

Some may still feel that domain authority is a reality and with those people, we must agree to differ.  The important thing is to test alternatives and measure results.  If you convince yourself that domain authority is working for you then by all means keep it in your picture of how the Web works.  Any proof you can offer here would be most interesting. Some of us will stick with a simpler view until proved wrong.