The importance of user context and performance

February 27 2015 By Posted in Expertise

Webmasters beware: do not let your users’ browser ruin their online experience… and your online revenues. Based on two Czech examples, this article takes a look at how browsers can affect the response times and reliability of a website and discusses whether this is worth worrying about.

 

Back in the old days (sometime before 2010), Internet Explorer was a dominant player. As a website owner you could be confident that your website was ready to go live if “it works with IE”. These days, there is no longer any single reference browser. All over Europe, browser market shares are equally split between Microsoft, Google and Mozilla. To complicate matters, the last three versions of any given browser are widely used by your audience. As a website owner how can you manage this complexity? And even more to the point – is it worth it?

 

Google Rates

Usage rates of Google Chrome versions over time in %. Full year 2013 – ip-label global Czech audience.

 

 

Google, Mozilla and Microsoft claim to provide better and faster web experience with every new browser version. Without any clear evidence, I personally trust them and go for automated updates of my browser. Unfortunately, having the latest version can actually be a bad idea on some websites. That’s what we learned at ip-label when helping European webmasters who were striving to deliver fast and responsive websites…
The example of one major Czech Media website monitored over the past 6 months yields interesting results. Performance by the latest version of Chrome (39) and Firefox (34) are comparable but Internet Explorer 11 is almost 50 % slower.

 

response_times

News Site (response times in ms.) Latest browser version (left) versus Previous browser version (right).

 

… but looking at last month’s versions shows a different picture: This time Chrome 38 is still a clear leader but it’s now Firefox 33 which takes 50% more time to display the page.

 

If this is true for one website, can we infer anything for other websites? Well … No! Looking at another reference Czech websites from a global European brand, we get completely different results: this time Internet Explorer works much better … and for both versions …

 

response_times_browsers

CZ Site global Brand (response times ms.) Latest browser (left) versus previous browser version (right)

 

 

So, now you are most probably convinced : the browser used to access a website has a direct impact on the customer experience. And the speed difference from one browser to another can be very significant – up to 100% in some cases.

 

Taking it from here, how do you manage this and optimise performance for everyone ?

Well, the first thing is obviously to start measuring response time for majors users contexts with one of so called “real user monitoring” tools on the market. And since there are new browser releases every quarter or so, there is no choice but to do it permanently. Then it’s a simple part of the QA process to get alerted automatically when response times for these key configurations exceed your company expectations or SLAs.

There are more dramatic cases where your website will completely fail due to browser incompatibility. That’s the case we recently had with a Belgium insurance company. Due to web coding issues, the purchase button was not visible on the order page for some versions of Internet Explorer thus preventing 20% of website users from effectively subscribing to their insurance policy. Fortunately, the same real user monitoring tools can trigger alerts when some browser regressions appear, allowing you to active action plans to resolve the problems without delay.
So, with browser versions changing monthly if you do not yet use real user monitoring on your website, it might be the right time to give it a try.

 

Sources: ip-label Real User Monitoring Data – Between July and December 2014 – Selected websites – Czech audience only

Article published in partnership with http://computerworld.cz/

Leave a Reply

Your email address will not be published