October 24, 2013 Leave a Comment
Last week saw a 17% increase in page speed compared to the control period before using Cloudflare. I had a feeling I wasn’t seeing the full speed boost. Because lesjones.com has so many pages and images, it seemed likely that Cloudflare hadn’t seen them yet, so it hadn’t cached them.
To solve that problem I ran a linkchecker that accessed every linked page and embedded image on the site. That seems to have helped. Compared to the control period, performance has improved 34%. For my site that’s an average load time that’s about 2.3 seconds faster. Not bad at all considering it took almost no effort.
One thing that concerns me is that the page load sample for last week was extremely small – just seven pages compared to 24 for the control period. For various reasons Google Analytics can’t* and doesn’t** collect page timings for all pages.
Both of those page load samples are very small as a percentage of traffic – just 0.45 for last week. At work the number is about 16%. I’m really not sure why the sample here is so small, but I’m going to continue the test for another week to be sure the results aren’t a fluke.
* From Google Analytics Help: “Site speed tracking occurs only for visits from those browsers that support the HTML5
Navigation Timing interface or have the Google Toolbar installed. Typically this includes: Chrome, Firefox 7 and above, Internet Explorer 9 and above, Android 4.0 browser and above, as well as earlier versions of Internet Explorer with the Google Toolbar installed.”
** That same help section says that “By default, a fixed 1% sampling of your site visitors make up the data pool from which the page timing metrics are derived.” However, that doesn’t match the numbers on this site or the one I manage at work.