How Do We Test Your Site’s Performance?
If you frequent our blogs, you’ve probably noticed a lot of testing-related talk. Recently, Kendra Skeene published a three part series about how our team has made a huge effort to improve the performance of our platform sites. We did things like compress images and optimize the code, but how do we know that the changes we made actually worked? *Queue more testing related talk*
Using Tools And A Plan
When it comes to testing your site’s performance, it’s important to have the right tools and a plan. Tools could be scripts you run on your site’s server to collect data, or manually running page speed tests using web-based applications. A plan can simply be what we’re measuring, how we’re documenting those measurements and a basis for improvement. In this post, I’ll shed a little light on how we executed our performance testing for your site.
Choosing a Variable
First, we identified what we thought was a good quantitative variable to base our performance on. We decided to go with page load times. Load times are measured by how long it takes for the entire site page’s content to load. Next, we selected the tool we thought best captured the data we needed to measure our pages' load times as well as give us insight into what was affecting those times. That tool is webpagetest.org.
This tool allowed us to discover that one of the biggest factors affecting load times was an uncompressed, or very large, image file. That uncompressed image file accounted for a large amount of kilobytes for the site to load, causing the load time to be slower than desirable. After establishing what we wanted to base our performance on, what tool we needed to use to measure that performance, and what factor was affecting the performance, it was time to record findings and analyze.
After establishing a basis for our testing, we started by taking a sample of platform site pages that had different attributes like large and small amounts of content, different content types like blogs and press releases, and even compared our sites with government sites from other states such as Alabama and California. Using webpagetest.org, we recorded how long it took for each page to load. Initially, we only took load times, but our testing evolved to include the amount of kilobytes that loaded during that time as well. We thought that information would come in handy once we compressed those large images and did another round of testing. This way, we could confirm that we did in fact reduce the amount of kilobytes to load and that the new reduction in kilobytes had an affect on the load time.
Improving the platform’s performance is an ongoing effort, but our testing methods have given us a better idea about how we can make improvements. With each round of testing, we gain more knowledge about how our behaviors like selecting images for the themes, can affect a site’s performance. I encourage your team to take a look at how your site is performing and see if there are areas that could use some attention.