Well, it was too good to last. Our PR 5 is now PR 3.

Why? Who knows. My guess is just regression to the mean. And it is my guess that these PR numbers are not only unstable, they are not a lot of use either.

Now, we all know that PR is only one factor that Google takes into account when serving up results against a query which is, of course, sensible.

But there is some preliminary work here in an article about the page strength concept that suggests the correlation between PR and SERP (Search Engine Result Position) is very weak indeed (and page strength is worse still).

I stress that it is only preliminary work because the analysis was done with only one keyword, and you will note from the graph that the integer nature of PR may well be obscuring relationships that might be more apparent had we access to the unrounded PR.

But it is certainly not encouraging for those who are chasing the PR chimera.

One data point, just one, that suggests that low PR may not matter all that much. I did a google search for “keyphrase extraction” and the link to my post on this topic “The start of the art in keyphrase extraction” on this (low PR) blog, came up on page two of the search results.

Why.. it is not obvious at all, at least not to me.

In my humble opinion, PR is a can of worms. I don’t have the time nor the interest to do a proper analysis of this chameleon/chimera.. and I don’t think there is an obvious sensible way to do such an analysis without first figuring out how to sample semantic (query) space..

Postscript:

• another proposed metric is “inbound link quality”, again of unknown value
• any serious statistical analysis would have to take into account
  1. the clumpy,”integer” nature of PR
  2. the fact that estimation of incoming links is non trivial and the results probably not accurate.. for example the syntax (for search.yahoo.com) linkdomain:www.dsanalytics.com -site:dsanalytics.com gives me 13 incoming links, while a google search link:www.dsanalytics.com/dsblog gives me a different number and different sites
  3. Different search engines crawl in different ways .. see for example, the suggestion that Yahoo crawls static links in preference to dynamic links. This is going to mess up the data.
• If you have an Experts-Exchange account (if not, get one) you can follow through a discussion entitled
  Excellent SEO but position 435?????, which should give you an idea of the anecdotal nature of the accepted wisdom in the area .. there are more ostensibly sensible suggestions here
  1. Search Engine Friendly URLs
  2. Search Engine Optimization for Site Owners and Small Businesses

  but we really don’t know if “common sense” and “expertise” would stand up to a rigorous quantitative examination