Last night I posted about what Google has to say on SEO and bounce rates. You can view the post here. A great question was posted by Wilson: “David, I was wondering, why Google want to have two different answer for the bounce rates…? ”
Even broken down into four parts, my response was longer than allowed for comments on this blog, so I decided to make it a post on its own. Here is my answer to Wilson’s question:
Great question. I will get to that, but let us look at another misunderstood part of the Google algorithm.
We have been wondering for years why Google has three different measures for PageRank. The real PageRank calculation used in their algorithm is a complex logarithmic calculation. All other things being equal, a link from a PR4.12964 page is probably worth many links from a PR3.45294 page, for instance (We have no idea to how many decimal spaces the real PageRank is calculated, not whether this has remained steady over the years or whether it fluctuates over time).
Then there is the PageRank in the Google Directory, which supposedly is on a scale of 8. I can’t find any reference to the 8-point scale in the Directory, but the Wikipedia article on PageRank is a good reference on this point. Interestingly, the Google Directory states that…
“The green ratings bars are Google’s assessment of the importance of a web page, as determined by Google’s patented PageRank technology and other factors. These PageRank bars tell you at a glance whether Google considers a page to be a high-quality site worth checking out.”
Note the “and other factors” wording.
Finally we have the famous Toolbar PageRank, a green bar on a scale from one to ten. This is what most webmasters mistakenly refer to as Google’s PageRank calculation. However, it is just an estimation that makes a PageRank of 4.0001 look the same as a PageRank of 4.9999, even though the latter might be worth many times the former. Meanwhile, it makes a PageRank of 4.9999 look much less valuable than a PageRank of 5.0001, even though the two are almost the same. Furthermore, everyone involved in SEO can recount numerous instances where a page “should” have a much higher or much lower PageRank than another page, based on the number and value of incoming links, but the Toolbar PageRank does not reflect that. (For instance, I have noted on many sites that a “links” page with identical link juice to a PR3 content page might nevertheless have a PageRank value of zero.)
What does this tell us about bounce rates?
Just like PageRank, bounce rates is a metric Google shares with its users. PageRank is viewable to everybody; bounce rates are viewable only to the website owner. In both cases, Google is showing a very simple calculation … a number people can use to quickly make comparisons between pages, between websites, between last month and this month, etc.
As I wrote above, “It would be a ridiculously simplistic algorithm that calculates bounces using such simple calculations.” Any serious calculation of bounces applied to a search engine ranking algorithm would have to be such a complex multidimensional equation that it would be useless to you or I as humans viewing it with our eyes (unless you happen to be a mathematical genius – and I mean genius – which I am not by a long shot.
Except to the extent that a search engine chooses to reveal how it treats bounces and other actions in its algorithm, we will never know for certain what plays a role and what does not, nor how big a role each factor plays. This is par for the course with ranking algorithms.
It is also totally possible that Google and the other search engines do not include bounce rates and related user actions yet in their algorithms. Adam Lasnik’s comments quoted in my previous post are good hints, but they are hardly official. Google engineer Knut Magne Risvik speaking in Norwegian at Digi.no and saying that Google can measure how many seconds it takes from when a user clicks on a link to click back to Google, and if it is a short time that visit was a failure, is not quite an official Google statement either. The only search engine that has released anything official is MSN through its BrowseRank paper … and that is not a statement current practice but of future intentions.
As this very young field matures, Google might also change its Google Analytics definition of “bounce rate”. SEO aside, the raging debate over whether a high bounce rate could sometimes be a good thing (depending on the nature of a website) makes a good case for changing the Google Analytics definition, too.
The summary to all this is that I have to answer Wilson with a simple “I don’t know”. But, just like defining “bounce rate” and “PageRank”, such simple answers are really a lot more complex than they look.
Grab The Bookmarketer For Your Site