Google still cares about C class ips

[goodroi]

In case you were wondering, Google will give significantly more link power love if your inbound links are on different c class ip addresses. I just finished a project and it was a dramatic ranking boost when we made this change.

If you want to get really crazy, then you might want to get your own dedicated IP to reduce the tiny chance of gaining a link from a real site that coincidentally is using the same hosting company. Getting really crazy its possible to have a dedicated IP and still have the other site in the same c block.

Anyway, no surprise that Google still cares about links from different ip addresses than your own. Funny how much of the old information is still relevant.

[brotherhood of LAN]

Interesting. Still makes sense, though not so much class C as subnet (though calculations are probably lots more trivial if just looking at the /24)

Don't mention class C's in hosting circles, they go nuts since the 'net has been classless [en.wikipedia.org] for a long time, and prefer to use the CIDR notation. I discovered this when sourcing hosting from a bunch of places.

It seems like it'd be more logical for Google to look at the subnet rather than a default /24, but not sure if they do.

[Robert Charlton]

...it was a dramatic ranking boost when we made this change.

goodroi, thanks for confirming this. I've been a strong believer in the dangers of interlinking sites with common hosting and/or common link sources... since 2003... and I've occasionally taken some heat for being as concerned about it as I have been. Google's Historical Data patent of 2005, was, I believe, the first official word that the line was drawn at shared Class C. The patent warned against obviously coordinated linking practices, and I think still applies in a range of areas.

I'm not sure that it's as simple as shared class-C IPs, btw.... I've seen sites with consecutive class-C IPs get hit... and, before the nofollow attribute, changing hosting company has occasionally been necessary (not always possible if hosting is internal to an enterprise... eg, if linking among a company's brands).

When things are this bad ^^^ , though, it's usually a sign that there are link inadequacies or coordinated patterns elsewhere. It is possible for independent sites hosted on the same class Cs to internlink for identification purposes and not affect each other adversely... but, to err on the side of caution, I've tended to use nofollows.

BoL... that Classless Routing article can produce headaches. ;)

[aristotle]

Well what about the old idea that Google can use Whois records to determine when someone is inter-linking their own sites, regardless of where they are hosted?

[Robert Charlton]

I'm thinking that Whois records might not scale efficiently nor be that accurate if there's private registration. Google is a registrar, but I don't know how much extra information that enables them to see, and particularly whether it's enough to determine ownership of third part sites.

Also, I'd say that Google is looking for signs of "coordination" in linking patterns... not merely common ownership... but common hosting is a strong enough signal to raise a flag.


PS: Edited for clarity.