My good friend Bill over at SEO by the Sea blogs about a lot of patents filed by the major search engines, I mean A LOT of patents. His blog is one of the most interesting blogs in the SEO / Online Marketing sphere, and should definetly be in your RSS Reader, when you´re considering yourself a true SEO expert. I tweet most of his posts, and the patents he talks about in his posts, but the patent he talked about two days ago, is worth a whole post & a closer look, cause I consider it to be one of the most interesting and important patents from Google in recent years. Well, at least in my humble opinion
The patent is called “Ranking documents based on user behavior and/or feature data“, and was filed in 2004 (!) by 3 Googlers. It just got granted two days ago, and shares some insight, about how a search engine might look at “a search engine might look at a wide range of factors to determine how might weight each link on a page may pass along“.
A link just ain´t just a link anymore, there´s a whole lot of stuff search engines are looking at, when determining how much juice (or whatever you wanna call it) the link is actually passing to the page it links too, and how much it will actually help that peticular page to rank.
There´s a lot of stuff, that has been floating around the web for a long time now, but there´s never been any proof. The patents talks about some things, which one might consider a proof, that these theories might indeed be true. And you gotta think about one fact -> the patent has been filed in 2004, so what just sounded pretty smart back then (where stuff like this most definetly wasn´t used), might be the fundament for some ever heavier stuff, that´s in use right now – 6 years later (!) – or will be coming into play in the next couple of month / years. So you should definetly think about that kind of stuff now, to plan ahead, and anticipate Google´s next move. Cause we all know, that paid links are a BIG pain in the ass for Google right now, and they´re definetly working on some stuff, to get a better grip on that one thing, that can still fuck up their best efforts to deliver the best searchresults for the user.
Bill lists a whole bunch of examples, that might be used to determine the strength of any one peticular link -whether features of the link itself, and / or features of the source and / or target document the link points to. It basically seems to come down to a “reasonable surfer” – the more likely the common user will click on the link, there more juice it´s going to pass to the target document. This can be determined, by either using user data collected via a web browser or the Google toolbar, or by certain attributes of the link, like it´s position on a page, and the font size & colour etc. etc. just “the probability that a reasonable surfer will access the document after following a number of forward links“.
A real easy example would be a whole bunch of links in the footer of page – I think it´s pretty easy to give those links a lot less power to pass on juice to their respective target pages, since it´s quite clear, that these links don´t get clicked very often, and are (probably) only there to manipulate the SERPs. Like I said before, this tip, you´ve probably heard before -> “Don´t buy links in the footer of a page” has been floating around the SEO forums & blogs for quite a while now, although there´s been no indication in any patent – it just makes sense!
“Examples of features associated with a link might include [...] commerciality of the anchor text associated with the link”
Whoa – now this is interesting! This is actually something I keep telling people for quite a while now – it just makes sense, to take a look at the commerciality of the anchor text, since it can be a strong indicator, whether this peticular link has been set voluntarily / organically or was set to manipulate the ranking of the target document in the SERPs for that anchor text. If you have a website talking about linux, and 98% of all links from that website point to linux related websites (“a topical cluster with which the anchor text of the link is associated“), BUT there´re also some links with non linux related anchor texts, that are commercial (“payday loans“, “play poker” etc.) and are also in a specific area without much other content around it (so basically just a bunch of links in the footer or sidebar, like you still see them all around the internet) – there´s a good chance, these links were just put there to manipulate the SERPs. I think it´s even so clear, that you can devalue these links algorithmically without having to worry all too much about any collateral damage. Ok, this is a pretty easy example, but there´s a whole lot of stuff listed in the patent, that can make other somewhat more difficult cases look pretty clear just as well.
So read Bill´s post, scan through the patent, and try to think about that kind of stuff the next time you´re
buying getting a link organically