How does Google evaluate an inbound link??

[compar]

I just posted what follows on another forum and thought that it would also be of interest here.

I know any discussion of what Google is doing post Florida is still highly speculative, so lets discuss what we think we knew pre Florida.

Can we agree that PageRank was then completely query independent. Simply a number based on the sharing of that same number calculated for all the pages in the link exchange?

Can we also agree that there was an additional evaluation taking place and at a minimum this involved Google reading the anchor text, or alt tag with a graphic link, and deciding on the page's relevance based on the this query dependent information?

If we can agree on this then the question is how did Google use these two evaluations in deciding where to place a page in a SERP?

We have all seem pages with higher PRs placed lower in the SERPs than pages with lower PRs. So we can only conclude that PR is not Google's primary consideration for placing a page in the SERP. This is only reasonable when we consider that Google's avowed purpose is to deliver the most relevant web pages in response to each search query.

PR is query independent therefore it would follow that it should not be used as a relevancy measurement.

Now the problem or confusion with PR arises out of the fact that Google decided to give us a numeric value for this one measurement factor. If PR attempts to measure relevance at all it is simply by implication. The implication being that in a perfect world a webmaster would predominantly link to pages of interest to his visitors, hence relevant to the topic at hand.

But this is the only measure we have. This is the only peek that Google gives into the results of any of their algos. So people place an unreasonable and unfounded amount of importance on the numeric value of PageRank. In fact many people are of the opinion that page rank reflect the total value of an or all inbound like.

So the question remains how did Google measure relevance and how did it factor into the placement of a page into the SERP? In another Forum one of the members proposed this formula for how Google determines SERP placement: (In fact the other forum referred to was IMF.)

PR + Page Copy = Rank for non competitive words

Anchor Link Text = Ranking of competitive words

Would anyone care to comment on this formula? If this is correct then PR has value only for the more unusual or unique search terms. And relevance no matter how determined is the determining factor in the SERP placement for high volume search terms.

Now I think this discussion is worthwhile, because while we don't know exactly what changes Google made in Nov. it is the major consensus that these changes involved more sophisticated methods of measuring link relevance. Nobody has suggested the calculation of the PR numeric value has been changed in any way.

So why are people still possessed and engaged in the headlong pursuit of PR value? Obviously to rank well for any popular search term we should be concentrating on relevance factors. This was assumed to be predominantly anchor text prior to Nov. 16. Today it may be a number of other factors. But to succeed in getting good SERP placement from our inbound links, our efforts should be concentrated on increasing our link's relevancy factors not on increasing the numeric value of our PR.

[I, Brian]

My understanding is that the old-fashioned Google SERPs worked on a system of general ranking elements being scored, and then this score being modified by PR.

In short:

"ranking elements x PR = ranking"

For the connection between PR and rankings: 1,000 high PR pages with a link to your site as being much more valuable than 1,000 low PR pages with a link to your site (link weight and anchor text being otherwise equal).

The point here is that if other ranking elements remain the same, the highest PR site would win out.

Google bombing I thought was seen as a more extreme manipulation of this formula - anchor text being a particular weakness that allowed elevation of the initial "ranking elements" factors to an extreme level with repetitive linking. Thus PR itself as a ranking factor becomes muted in the face of the otherwise very high "ranking elements".

I'm afraid that I can't pull any papers out on my initial perception of PR at the moment as I've a lot to do here - had to destroy and rebuild my partitions today, reformat, reinstall...and now there's a server error buggering up my business site. I'll try and find something over the next couple of days to try and support my perception - presuming that I'm not corrected with a cite in the meantime. After all, my early days of learning SEO were done at SitePoint, reading Chris Beasley's articles and links (John has as similar high regard for him as he does for Jill Whalen. ).

Florida, of course, complicates everything. Although some of the system seems to have remained the same (or put back to similar), a certain swathe of general results seeks to require an additional "selection" criteria to even be considered for ranking for certain terms. This is where the theory of Hilltop's involvement says that experts pages and authority sites are used in some measure to "pre-select" paegs for ranking - whereas the TopicSensitive PageRank theory says that the actual context of the links is now the deciding factor.

Personally, I see issues of theme and context as issues that SEO needs to take into account now for long-term SEO projects. Even if PR alone was a ranking factor - that links were merely links without context - I don't see a long-term basis for following that reasoning anymore.

2c.

[I, Brian]

PS - compar - if you give me a week I'll have a public experiment set up, using some of my sites, to address not only this question, but hopefully any degree of association (if there is one). .

[compar]

Brian,

You can have all the time you like. Let us know when you have something you think is informative or valuable.

Interesting stuff though. Wouldn't you say?

[I, Brian]

No problem.

I'm actually setting up a number of public experiments.

[Johan007]

I posted than formula up after reading JohnScott articles about the search term Jobs and is the reson I joined this forum!:
http://www.internet-marketing-resear...ound_links.php

Here he has another article:
http://www.internet-marketing-resear...erank-wins.php

where he concludes:

In fact, anchor text and inbound links is 95% of effective search engine optimization.

Its not rocket science, so I will stick to my formula cos its simple and easy to read (too easy for many people to accept!).

I have refined it...


Page PR + Page Copy = Ranking for non competitive words

Number and Quality of Anchor Link Text = Ranking of competitive words

The above is for Google, for Inktomi its not Anchor Link Text but the BOLD/Title/Heading text from the inbound linking page. I have proof somewhere in this forum and if you want it I can dig it up - however you should be able to do it yourself.

[John Scott]

I think on page factors don't matter too much. I mean, if you anchor text your way to the top, you'll stay there. If you try to keyword-density your way to the top, I think you'll end up with ugly, keyword-rich copy.

[compar]

Page PR + Page Copy = Ranking for non competitive words

Would you consider changing that formula now in light of John's "Blueberry Pancakes Fraud Strategies and Opinions" results?

Maybe it should be 'Anchor Text + Page Copy = Ranking for non competitive words'

There is no evidence in John's experiment that PR had anything to do with it. The PR of the linking page was 3.

Now the PR of John's page with the content was 7. But I doubt that was involved.

[Johan007]

yes you could do as I know that would have worked.

I choose to exclude it for practical reasons. We are talking about a very non competitive words and its really not worth the expense of using up valuable external links for something that could easily have been done using Copy Writing.

[I, Brian]

With really competitive phrases you're going to be in a field filled with Google Bombs anyway - so the effects of PR by itself may be quite drowned out. But I'll have that experiment up for you.

[tosheroon]

But, since the last 2 go arounds, is Google stemming or interpreting the inbound link text, if I've got a 100 inbounds for a wedding pictures site would I better to have them all read "wedding photgraphy" or split them up "wedding photography" "wedding photographer" wedding photographers"?
or better yet a hundred links each different yet semantically related -"marriage photographs" "wedding pictures" "photographers for weddings" and so on.
Google bombing suggests that identical inbounds are most effective but no-one has tried any other types of googlebomb yet we may learn something as the competiton to find out who is a "miserable failure" increases.