PageRank has not gone away, still not dead

Google 6.10.2011 1 Comment

Last night at about 3:45am BST Google made a change to the toolbar PageRank service which caused 3rd party checkers to stop working.

It was a trivial change to the API URL, but some it ignited yet another round of “hooray, pagerank is useless and has gone away” tweets and a little bit of panic amongst those who still have some (maybe too much) faith in it.

Let’s get some things straight though.

PageRank does not:

  • Tell you anything about how a site will rank for a given search term
  • Provide anything like an up to date assessment of the quality of a URL
  • Prove that one site is “better” than another
But, it isn’t completely useless because PageRank does still:
  • Have a correlation to ranking overall, when you look at enough data
  • Give a reasonable impression of whether a site has been stable for some period of time
  • Give a rough idea of the number of indexed inbound links a site has pointing to it
This means that it’s worth a glance when considering whether to obtain a link on a particular site, but not much else. It may be worth giving some consideration to when you’re buying established sites, since it does correlate to rankings in a broad sense.
From an analysis of around 330,000 URLs, the distribution of different PageRanks to 1st page position looks like this:
The first thing that’s obvious is that sites with no PageRank at all have no problems taking the top spot. This may come as a surprise to some, but it doesn’t account for the variations in longevity of a particular ranking. It’s likely that sites with no PR won’t stick around too long, so while they may comprise a significant proportion of the top place it’s unlikely that they’re going to be the same set of sites holding the same position for long periods.
Because the graph above shows such a high proportion of the sites to have no PR, it’s also helpful to look at the upper section alone, with the “no PR” sites taken out:
Here, we can clearly see a correlation between higher PR sites holding higher positions in the results. As you go down the first page, the likelihood of a site being a lower PR increases, whereas at the top there is a much higher change of it being at least a PR3. The tipping point seems to be around a PR2 which remains more or less constant all the way down the results, with PR3 and upwards getting a clear benefit.
Before anyone lays into me with the correlation vs causation argument, yes this is correlation but if I were buying links on a site or looking to buy an aged site, I’d be stupid to not pay any attention to the PR.

One Response to "PageRank has not gone away, still not dead"

  1. It’s good to finally see someone use some numbers to show the relevance and use of pr instead of the usual speculation.

Leave a Reply