Site Speed

A couple of years ago Google introduced Site Speed to it's search algorithm.  This is a measure of how long it takes a page to load. You can track this measure for your site in Google Analytics as well in Google Webmaster tools.

Recently I've been investigating this measure for some of our sites and was quite surprised at how long Google thinks it takes our pages to load. Page load times as measured by Google were several seconds above the one second it takes for our largely static pages to load from the end users perspective. I measured this from several remote locations on different continents and our pages appear to load in about a second. Crucially the time measured by Google is also above their recommended maximum page load time. Why the difference?

Google doesn't just measure the time it takes the main content to load, but also how long it takes all the Javascript to process and potentially load other bits and pieces.  On our pages, a lot of this extra loading doesn't really affect whether the user can get on with reading the content and using the site. We try to make use of progressive enhancement as much as we can.  A further detailed look at the load times showed that the main culprits where actually all the social networking add-ons such as Facebook Like, Google +1, ShareThis and ShareThat.  I am now a little worried that Google may be penalising our sites unfairly. I don't consider social networking buttons as being vital to the proper functioning of our site.

On the other hand I can see why Google have introduced this feature and why they take the approach that they do. All sites nowadays are filled with Javascript and Ajax calls. On the ones who do this properly you won't really notice.  But on lots and lots of sites the page gets reformatted slightly about 2 seconds after it initially got displayed. This is really annoying if you've just read the first sentence and then text text shifts. Or you're about to click a link and then the text shifts so you click another link. Both of these have been happening more often recently, especially when browsing on my six year old laptop. And I must say: It is really annoying.

So I can fully understand why they are measuring this.  I do hope though that their algorithm is clever enough to evaluate what the consequence of the extra loading to the end user is. Do they take into account how quickly the initial content loads? Do they measure how the text displaces after the Javascript has finished? I sure hope so.