The people at SEOBlogger wanted to get opinions on the subject of subdomains and since I think this is quite an interesting topic, I have not written much about it and thanks to the Deutsche Bahn whose train broke down, giving me more time than expected, here my evaluation of this matter.

Google works with hostnames. For a normal URL the hostname is the part between the http:// and the first “/”, in this case it would be “www.sistrix.com”. The subdomain which holds our SEO-Tools (tools.sistrix.com) is another hostname and also has its own IP-address. To reflect the hierarchy correctly, hostnames would need to be read from right to left – single elements are separated by the period: com > sistrix > www or, in this case, tools. If we leave out the first part, meaning the domain, we get exactly this sequence in which Google passes on strength as well as possible problems. Those with a strong domain will also rank considerably with subdomains thereof, as can be seen with Amazon's Astores and the many subdomains belonging to Ebay, for example. As far as possible penalties go, we have the same picture: inheritance along hierarchy lines. This should not come as a particular surprise but rather something most have already expected. It gets interesting when we consider that penalties are not passed on from inferior hostnames to superior ones – if you “caught something” on spam.example.org, you do not have to worry about the main-domain as well as any other subdomain – as long as there was no manual penalty. One of the larger auction houses is regularly burning through single subdomains without it having any repercussions to the main domain – though we could also observe this conduct at considerably smaller projects.

Everyone who comes up with the idea now to just considerably increase the number of subdomains and sub-subdomains should be aware of the “subdomain-problematic” that I have already touched on and consider it carefully.