Study: Browser performance on European websites
Based on ip-label observations of a panel of four European websites, this post takes a look at how browsers influence website response times and, in the worst cases, degrade user experience.
Nearly one in two visitors uses Chrome
Our most recent data from September 2017 show that European customers favour Google Chrome, with an overwhelming 45% market share. The competition is battling it out for second place, with Microsoft Internet Explorer, Apple Safari and Mozilla Firefox each garnering similar numbers of supporters. ip-label’s European data also indicate that apart from these big 4, no other browser has gained a significant foothold, whether Opera, Edge or Vivaldi.
8 to 16 different browser versions
Like in a preceding study conducted in 2015, we wanted to know how many versions of browsers must be included in order to be “representative” of a website audience. Results taken from four major European websites (located in Germany, Spain, Sweden and France) provide useful insights.
For our panel, the dominant browser is used on average by one third of all users. Testing on 3 browsers instead of 1 would cover half of the audience. It is expensive to be more specific: today you need 16 different versions of browsers to represent 90% of a website audience. That’s a huge increase from two years ago when only half as many (9) browser versions were required. Why is this the case? One main reason is that older browser versions tend to remain in operation longer than before.
These results suggest that a significant fraction of the public must be using an outdated version of a browser (by “outdated” we mean a version that is not the latest stable release). Drilling down into ip-label data, we find that 35% of web visits in Europe at present are performed with outdated browsers.
A variation in performance of 30 % or more between browsers is commonplace
After looking at the demographics, we then turned to performance. For this, we focused our research on our four-website panel. For each website, we sorted the browser versions from fastest to slowest. Then, we computed the % of response time increase for a given version against the fastest version.
Looking at the results, we see that on average response times are 50% higher on browser #4 with respect to the dominant browser. This is only a rough average: the Spanish site is the most stable, with response times only 20% slower. The Swedish website was the most greatly affected; response times doubled with respect to the reference.
These panel results are interesting but limited. They confirm that the user’s browser has a direct impact on customer experience. To put this into perspective within the broader picture, we crosschecked it with our overall European data: in September 2017, 46% of page downloads measured by ip-label were concluded in 3s or more; this is widely considered as poor user experience. It means that for any website, almost half of the audience is not having a satisfactory customer experience!
So what can you do about it?
This being so, what can you do to manage website performance? Well, obviously the first thing is to start measuring response times for your own website with a market-standard “real user monitoring” tool. This will let you instantly identify major areas for improvement and help you focus troubleshooting efforts.
And since new browsers are released every month or so, it is a good idea to include this tool as part of your best practices. With such tools, you can be alerted proactively when response times violate your SLAs or when significant variations in performance occur.