Regional Browser Roundup

Browsers and their versions routinely ride off into the sunset, destined to become the bleached bones and tumbleweeds in the outdated desert. Long gone are the days when Netscape and Mosaic reigned supreme. The newfangled browsers of today support more robust coding, and tend to stay compliant with the current standard as each new version is released.

Web Standards

When we develop web applications, we always develop based on established web standards. In theory, a standards compliant site or application should "just work" on all browsers. That's the idea, anyway. The truth is all browsers have bugs, anomalies and implementation variations. Some more so than others. And some browsers just don't keep up with newer technologies as well as others.

Supply and Demand

Part of our promise to our clients is that we will make sure our final product works flawlessly in a list of "supported" browsers. This means that if something doesn't work or look right in that list of browsers, we do a workaround so that it does work and look right. But we do have to draw line somewhere with that approach since most clients have limited budgets and don't want to pay for workarounds for browsers with tiny market shares. Hence our supported browser list is determined by market share. This is getting more complicated as Internet usage diversifies across devices: phones, tablets and desktops.

Browser Stats Explained

There are plenty of sources for global browser stats that we could use. But should we trust those in our own little corner of the world? Not to mention there are different metrics used to determine browser popularity. The two most divergent methods are user-based or hit-based. A simplified way to understand this is if you have a small simple site and 5 different users visit one day all with different browsers. Each has a 20% share. But if 4 of them just look at one page, and the 5th user looks at 10 pages, then the user-based stats are still the same at 20%, but the hit-based stats have changed radically in favor of that guy looking at 10 pages. Which is more meaningful?

To illustrate, two reputable and commonly cited sources have wildly different results (February 2014):

NetMarketShare (open in new window):

1. Internet Explorer: 58%
2. Google Chrome: 17%

StatCounter (open in new window):

1. Google Chrome: 47%
2. Internet Explorer: 25%

Both are legitimate ways to interpret raw data, so there is no real "right" and "wrong" here.

This can be even less meaningful to us locally since other parts of the world have very different usage patterns. For instance China, with qazillions of users, has many users on very old Microsoft operating systems that are stuck with older versions of Internet Explorer. This skews "global" stats significantly. India has a very high percentage of mobile users vs desktop users. More skew.

In the case of a local or regional client what we care about is local and regional usage patterns. If it's a redesign, we can look at existing Google Analytics data. If it's not, then we need to set our own thresholds.

By The Numbers

To see what a local average might be, we used our own Google Analytic data from January and February 2014. We then sampled several client sites that have pretty broad demographics, with regional audiences, and have significant traffic to their sites. Google Analytics stats are based on "visits". The results are quite surprising.

Safari, hmmmm. That's somewhat slanted because of the local popularity of iPods, iPads and iPhones. Chrome is popular on mobile and desktop and that causes some distortion. Let's separate out some of the mobile overlap:

Some of the striking differences here are that locally, Safari is 20% on the desktop, yet NetMarketShare and StatCounter have Safari at a little over 5%. Major difference.  This is further corroborated by our raw log data. Internet Explorer always marches to the beat of a different drum, and usage stats are no difference. NetMarketShare has IE8 as the number one IE version at 22%, while StatCounter has IE11 at the number one IE version, but much lower overall at 8%. StatCounter has IE8 as the second most popular IE version, at 7%, while our stats have it in third place at 6% and trending down.

What makes the Internet Explorer stats meaningful for us, we generally "support" the two most popular versions of IE. The ones that cause the most pain are those before IE9. IE8 is technically a dog and requires significant extra effort to do much of anything that is state of the art. So that's good news for our "supported browser list" as the newer, improved IE versions are taking the top 2 slots with IE11 solidly in first, at least in our regional stats. These are much better browsers.

Takeaways

The bottom line is that when you approach a project to develop any online content, you have to plan for the audience in more ways than one. Different audiences and geographical regions will use different devices to view your content. You may not have the luxury (or pain) to develop for all of them, and a good examination of usage data can provide a sensible approach to who you're catering to. Sometimes it just makes sense to cull the herd and reward yourself with a cool sarsparilla at the end of the project.