Wade Womersley (our CTO) recently published an article about IP’s and their (ir)relevance in the world of SEO in light of IP re-use and changes, below is the full post, the original is at xcitestudios.com.
I recently overheard a discussion in the SEO world to do with IPs and the dangerous* world of link acquisition. The discussion was to do with clients who ask for things like “unique class C” as a requirement. Many moons ago, IP was almost certainly a factor in link acquisition, big bad Google did not like lots of links coming from the same IP address – or at least, it was a common belief. Then, one day, people realised “oh dear, we’re running out of IPv4 addresses” – and sure enough, we really are. In fact, ARIN are due to run out of IP’s to allocate this month.
The thing about IP addresses is, they don’t need to be single use, you can have one IP address serve, not only hundreds of websites, but thousands of services – just take a look at this list of reserved ports. Apache, IIS, nginx etc all support shared hosting, this is where multiple websites can use the same IP address and the domain name determines what files are shown. As an example of this, check out http://rantaklubi.fi/ and http://capslockhelsinki.fi/ – two sites that are on the same IP address (220.127.116.11) but have nothing in common (feel free to check whois as well to see they are unrelated).
As a more extreme case of IP sharing, look no further than Google’s own hosting platforms, specifically the IP address 18.104.22.168. Wander over to http://www.tcpiputils.com/domain-neighbors and put in that IP address, you’ll get a table with some information that looks like this:
This alone really should be enough to show you how little Google care for unique IP addresses, let alone unique class C! But, I wanted to do a bit more digging, specifically in Finland – historically notorious for any kind of link acquisition across a whole range of categories. To that end, I used MediaSkunkWork’s historical crawled data for the last few months to get some numbers out (incidentally you can take a restricted view at the type of raw data the guys at MSW have access to by going to http://search.mediaskunkworks.com/). I got a list of 34,000 SLD’s with a .fi ending, along with their IP address then worked out how many of those SLD’s share the same class B, class C and class D (D being full on IPv4 address). And this is what I found:
Out of 34,000 sites in Finland, at the class B level, 73% of sites share the same class B as another; at the class C level, all class C’s have multiple sites in them; finally, and most importantly, in class D (unique IP addresses), 26% of sites share an IP address. This is a relatively small sample set but it’s across a whole variety of classifications for sites (sports, news, gaming, family life etc).
Now I am not saying IP address does not matter at all, but it only matters as part of a bigger picture, if you have 6 websites all with a link to your site and the class C is the same, but they are owned by unrelated people – you’re going to be fine, even if they are all on the same IP address. What would be a problem is 6 sites in the same class C, in the same subnet, owned by the same person or using identical themes (similar HTML but different images etc).
So the next time you have a client saying “we want unique class C” or even “class D” IP addresses, tell them it isn’t the 90’s anymore, quality matters, ownership matters, IP address is a tiny part that, for the most part, has no impact.
* There is a lot of debate over link acquisition, phrases like “black hat” get thrown around – this article isn’t about the morality of link acquisition (note I didn’t use ethics – it’s not an ethical debate).