Search Engine News :: Optimise Web Sites for Organic Search: WebProWorld

Friday, August 13, 2004

WebProWorld :: Proof that Outbound Links improve SERPs?

WebProWorld :: Proof that Outbound Links improve SERPs?: "One of the theories that have been bouncing around in different forums is the possibility that outbound links helps a website (page) improve its ranking position. Everybody seems to have an opinion on this, but I've been hard pressed to find any hard evidence one way or another.

So, I decided to conduct an experiment...

I created three blogs to rank for the search term Nigritude Ultramarine. I designed the original blog to be a psuedo-directory for the term in typical blog fashion. I linked to various websites that I thought gave good information about the contest. My other two blogs were controls. Control A had the same posts and links as the original but I specifically excluded the term Nigritude Ultramarine from the outbound anchor text. Control B had the same posts as the original but no outbound links. This control blog was to determine if the original ranked just due to keyword density.

I had some trouble with Google's duplication filter and had to rearrange the wording on the two control blogs to bypass that. But, other than that, and what I stated above, the blogs are virtually identical

Here is how the blogs ranked this morning:

Nigritude Ultramarine News & Views: 351
Nigritude Ultramarine Control A: 588
Nigritude Ultramarine Control B: 587

Unless there is some flaw with the experiment that I'm not seeing, this seems to indicate that the outbound links in the original blog are giving it a scoring boost for the term Nigritude Ultramarine.

I didn't really set out to prove this point, rather I'm interested in testing a theory. Actually, for awhile there I didn't think there would be a scoring boost. By all means, if someone sees a flaw in the testing, please let me know.

The obvious possible difference is in the inbound links to the different blogs. I tried to control this as much as possible. Yahoo shows all the links as being the same, except for control B which also show a link to itself.

Edit: I checked the rankings through Googlerankings.com"

a brief summary of my interpretation of Hilltop.

Expert pages are qualified by:

1. Meeting a certain "out-degree" threshold. The paper mentions 5. I don't know if this is 5 links or 5%, my guess is that it is a percentile number. (a page with one link in and one link out would have an out-degree rating of 0)

2. Expert pages must be non-affiliated to each other and to the target page.

Affiliated is defined as:

a. The first three octets of the IP address being the same, and,

b. The right most non-generic word of the url being the same. (ex. ibm.com is affilated to ibm.co.uk)

Expert pages are culled from the general population of the web into a seperate inverted index in order to determine authority pages.

Authority pages are determined and ranked by:

1. Having 2 or more non-affiliated expert pages pointing to it.

2. Having the search term query (labeled as "key word") in either the Title, Heading, or Anchor text (labeled as "key phrase) on the page that points to it. (If the search term query is in the title, then all the outbound links are determined to be relevant, if the term is in the heading, then all the links underneath that heading are relevant until the next heading occurs, if the term is in the anchor text, then only the url that link points to is determined as relevant.)

3. Inbound links from expert pages are scored two different ways.

a. Level Score: If the search term query is in the title a level score of 16 is assigned (according to the paper). If the term is in the heading, then a level score of 6 is assigned, and if it is in the anchor text a level score of 1 is assigned.

b. Fullness Score: The fulness score is subtracted from the level score depending on how diluted the key word is within the key phrase.

I assume that this "authority scoring" is then added to the other variables in the general algorithm.