Google Caffeine

Back in August, Google took the unprecedented step of announcing to the world that they were launching a new version of their search engine, and they wanted the world to test it for them.  This new version was named Google Caffeine.  The sandbox environment for testing was available here, and if you just clicked on that link you’ll see that the sandbox is no longer available – instead, you’ll see the following:

After several months of not really hearing much about Caffeine, Google have decided to start rolling it out to everyday users of Google.  Matt Cutts has just confirmed that full rollout will not happen until after the holidays are over, but there has been no comment on the Official Google Blog.

So what exactly is Caffeine, and as a webmaster, do you need to worry about it?  Matt Cutts described it by saying that the “update is primarily under the hood: we’re rewriting the foundation of some of our infrastructure”, and that “some of the search results do change”.  From this, it sounds like any changes will be relatively minor.  However, this all sounds a lot like when Google rolled out the Big Daddy update back in 2005, and although I can’t recall any specific examples, I am aware that some websites took a hit back then.  I guess that whenever Google make any changes to their algorithm, you need to be cautiously aware that things can go wrong.

From playing around with Google Caffeine, I only ever witnessed very minor fluctuations between Caffeine and the main index.  If there was one thing I picked up on, it was at the speed with which user data seemed to being processed.  For example, the day after Google Caffeine was launched, if you did a search for “caffeine” in the sandbox, then a related search was prominently displayed, suggesting “google caffeine” as a related query.  In my mind, the analysis required to do this would be pretty heavy, and wasn’t mirrored in the main index, which might suggest that Caffeine allows Google to process this kind of data much quicker.  Why would they want to do this?  Following the Vince Update, analysing user data has become much more important within their algorithm, and the ability to do this much quicker might allow Google to easily roll Vince out across a wider range of queries in the future.

This is a guest post by Jonathan Stewart, the Head of Search & Social Media at Review Centre. You can follow him on Twitter here.