QUOTE(maleman @ Jan 28 2006, 12:54 PM)
Mr. Martinez. Can you expand on the above?
What exactly does Google look at? How does G decide what's relevant aside from the link popularity?
Linkage has nothing to do with relevance.
When you type in the query "better mouse trap", do you really believe the best page out there is the one with 10 times as many links as the other pages? Yahoo! would win that race every time regardless of the query.
The number of links pointing to a site has nothing to do with whether that site's content is relevant to the search query
Google hasn't publicly revealed exactly
what they look at. But anyone who says or implies that Google primarily
looks at or relies upon linkage is just plain wrong.http://www.google.co...bmasters/4.html
1. How Google ranks pages.
Google's order of results is automatically determined by more than 100 factors, including our PageRank algorithm. Please check out our Technology Overview page for more details. Due to the nature of our business and our interest in protecting the integrity of our search results, we limit the information we make available to the public about our ranking system.
Emphasis is mine. Everyone talks about PageRank, but it's hard to define what the other stuff is.
Rand Fishkin of SEOMoz took a good stab at it (the best I have ever seen, in fact) in his Google Ranking Factors
article. However, when he first published it, he only reflected his personal choices. Several of us (perhaps many people) expressed publicly and privately some reservations about Rand's personal weightings. So he invited about a dozen other people (includng me and Jill) to suggest our own weightings, and Rand then aggregated the results and provided a very interesting group assessment of what appears to be important to Google.
The most important factor, according to the aggregate assessment, is the title tag.
Now, 13 people most certainly can be wrong. But the fact that so many others besides me put that factor ahead of inbound linkage underscores just how poorly the SEO community's senior members have communicated the message.
People keep talking about links. Links are important. They are vital. But if you don't look at all the other possible
factors that are influencing your position in the rankings (and you have no control over many factors), you short-change yourself.
G is (from what I understand) fully automated. How can a fully automated program read the contents of a page and decide whether or not it's relevant and moreover, how does it decide that "page a" is more relevant than "page b"?
Well, a lot of people question just how fully automated Google really is. While I don't wish to involve myself in that particular debate, I can say there are some well-known cases where Google outright banned some sites. They simply don't appear in the search results no matter what you type in.
Whenever Google manually removes a site from the index, one other site moves up to take its place -- and often a whole chain reaction of upward movement occurs.
But relevance is determined by looking at both the use and meaning of words. In an ideal world, Google and other search engines would be analyzing our sentence structure and equating synonyms and separating hononyms and doing all sorts of esoteric linguistic semantic stuff.
In reality, they start out with word lists that link to all the documents which "contain" or "include" or are "associated with" those particular words. Our queries are then paired up with the word lists and from there Google looks at things like proximity of the query terms to each other, emphasis that is placed on them, and how many of the query terms appear in the documents' word sets, etc.
It has been established beyond question that inbound link anchor text is somehow associated with a target document by Google. Hence, many people look to links to assert relevance for specific search expressions, but they don't do much on their own pages.
I've got numerous pages that don't have many inbound links, much less inbound links with precise anchor text, to outrank other sites with hundreds and thousands of such links in highly competitive expressions.
Some people have argued that I can produce my own highly valuable links because I have tens of thousands of pages of content. But I don't have time to do that. Instead, I place links on my site maps and a couple of other key pages and then just leave them alone.
I presently have a page I launched last year which has remained in the top twenty for a very intensely competitive business search expression that produces in excess of 130,000,000 results (it was 100,000,000 last year). To date, that page has fewer than 10 inbound links, all of them from my own network.
That page has on several occasions moved up into the top ten, and I have seen it listed as high as 7th place.
According to Google, the site in 1st place has over 1100 references to its URL (this is as close as you can get to a link search in Google). Today (or, I should say, at this time -- Google appears to still be in the throes of an update) Google is reporting 325,000,000 results for this 4-word expression. My page is showing at 11th (recently I have seen it drop down as far as 17th place).
The 2nd listed site has almost 20,000 references to its URL. So, number 1 has 1100 and number 2 has 20,000. And I'm competing with them with fewer than 10.
This expression, which I have never disclosed, is closely related to the search engine marketing industry. There used to be about 4-6 major, well-known SEM-related businesses who consistently ranked in the top ten. I went after this expression last year after I got sick and tired of one individual's grandstanding about how important links and Alexa rankings are.
His site no longer appears in the top twenty, probably not because of anything I did. But this is an actively searched query and I do get visitors from it (I've even gotten some inquiries, so my call-to-action must not be too bad).
That is the power of understanding how search engines work. Imagine what I could do if I were ever to do a link-building campaign.
Edited by Michael Martinez, 30 January 2006 - 04:29 PM.