A snail was walking to work one day when she got run over by a turtle. A police officer arrived to take a statement, and asked what had happened. The snail replied “I’m not sure, it all happened so fast.”
How many times have you gone to a webpage and abandoned it because it took too long to load? I’ve done that and I expect you have too. Well what if that’s your website and you’re losing visitors because your site takes too long to load?
We’re a bit obsessed with site performance around here. We spend a lot of time making sure our clients’ sites load quickly. But how do you evaluate your site’s speed? The most obvious way is to load the website in the browser and gauge how quickly it appears. But why am I going to tell you that’s not the best way? While it’s important to test your site for yourself here’s why you need to supplement with other tests:
- When you test in the browser, the speed test is affected by factors unique to you that might not affect other users. For example, you may visit your site regularly so your website is cached on your computer and performs faster than it might for other visitors. Or conversely, you may be logged into your website and it actually performs slower because it doesn’t use the website cache when administrators are logged in. Confused about caching?
- Your qualitative test doesn’t tell you what’s slowing down your site or how to fix it
- Your test isn’t reliably repeatable because of factors outside your control (like other traffic on your network slowing down your connection)
We use various third-party tools to help with speed testing. These give results that we can share with clients and colleagues. And they give a report showing what’s working and what needs improvement. And they also help us see and measure things that happen too quickly for anyone to detect on their own. Then we can work on those problem areas to accelerate the website and we don’t have to be like the snail that got run over.
Here are our favourite speed testing tools:
- GT Metrix is my go to
- DebugBear
- Pingdom
- PageSpeed Insights
Can we rely solely on these tools? Automated tests are very useful however they can be inaccurate in a few different ways. They need to be paired with human testing. A couple of ways they can give a misleading reading:
- Something interferes with the measurement. In October, we were doing extensive speed testing and we determined that an error message on one page was distorting the results, making the performance appear much poorer than it actually was.
- Automated tests don’t benefit from caching because every test is independent. The typical user however, will visit more than one page and since your browser caches page resources, every page past the first benefits from a speed boost that automated tests don’t see
- Ultimately, automated tests don’t truly emulate the human experience. A page may quickly load the top part of the page but delay the lower portion. The automated test may give a poor speed rating to the page even though functionally, to a human user, it appears very fast
Ultimately, you should run a combination of automated and manual tests to best gauge your site’s speed. Your site feel slow? We’re happy to discuss ways to make it faster.
Photo credit: Wolfgang Hasselmann on Unsplash