It's March 2016 - what browsers and operating systems should we be supporting?

A regular client asked me which browsers (version numbers) and devices would be supported by a new responsive web application, which we are currently building out. I was about to fire off my typical email reply to such enquiries, when I realised that I'd been copy/pasting the same answer for the past three years. (How time flies when you're building websites!) So I did a bit of research and a sweep around the office, to get a new consensus on Wiliam's position. And I must say, I like where we're heading.

Setting the trends

Looking at the latest market share statistics* is always quite revealing. It was very obvious that recent announcements by the larger software developers had clear impact on usage trends.

• Microsoft is no longer supporting Windows XP
• Microsoft will also end support for Windows Vista in April 2017
• Microsoft is no longer supporting IE versions 8, 9 and 10 at all, across any platform / OS
• Microsoft is not supporting IE version 11, running on Windows XP or Vista
• Google is no longer supporting Chrome browser, running on Windows XP or Vista

What this indirectly means is that (finally) those users / system admins who have been fighting off pressure to upgrade Windows XP (15 years old) and Windows Vista (9 years old) are now basically having their hand forced. These people can no longer rely on software patches and updates to keep their systems safe. On top of this is the obvious pressure being applied by Microsoft in offering free Windows 10 upgrades to many systems - an offer that is pretty tempting, and one that many mums and dads at home will simply see as a logical upgrade step; a no-brainer given the time-limited free release.

Windows 10 is the first Microsoft operating system not to come with Internet Explorer, and as such, we've seen the total market share for IE (all versions) drop to roughly 13%. Nearly one in four (75%) of people who use Internet Explorer do so on IE11, however these people are also fast adopting the new Microsoft Edge browser which has replaced IE. Thus, when you add Edge, Microsoft's total market share is not down as much in total, but neither is it really moving up, probably due to the fact that Edge is only available on Windows 10 currently.

While IE11 still maintains a fair share of anything from 6-10% of the total market... it's doomsday for the older browsers

So, while we have the new Edge browser to contend with, and IE11 still maintains a fair share of anything from 6-10% of the total market (depending on what source you draw your conclusions from)... it's doomsday for the older versions. Microsoft IE10 is now only used by 0.3% of worldwide internet users. IE9 is down to 0.2%, and the same shows for IE8. It makes no sense to fully support these any longer.

To keep, or show the door

What browsers and devices Wiliam chooses to support is largely based on Internet usage trends. When a popular browser or operating system starts to wane in popularity (generally less than 1% worldwide usage) - or becomes unsupported or vastly obsolete due to technological change - then we will consider dropping it. This doesn't mean that our websites won’t work on these browsers/devices, but it may not be 100% true to the ideal design or function, and we certainly won't go out of our way. This doesn't mean we won’t support a particular platform if asked, but generally we charge additional fees, as our slicers need to undertake additional testing in the specified environment.

I wrote previously about the (ideal) end for the need to test multiple versions of the same browser. and based on the above, it would appear we may have finally reached the tipping point where legacy browsers - that did not support auto-updates - are finally on the demise. This is fantastic news, as it means we (web designers and developers) can now legitimately shed a lot of the extra baggage that we have carried for too long, during the testing and QA process. And so we have decided that Wiliam will no longer - as a rule of thumb - support these browsers.

Reaching a new consensus

Wiliam generally define our web stack requirements into three groups:

• Optimised
• Usable
• Not supported

We have concluded that any browser that is not self-updating, and certainly those that do not support responsive designs, are to be non-supported. We've also finally 'bit the bullet' and dumped support for all legacy versions of Internet Explorer. This is something that I've wanted to do for a long time, and we've just been waiting for the statistics to justify the means. It would come as no surprise that even a browser that is supported by as little as 1% of users worldwide, is still a browser that a large organisation may consider important to support. (Think about what losing 1% of a million visitors might mean to your business... potentially ten thousand customers!)

We also concluded that certain mobile devices needed to be given a 'nudge' in the right direction. One of the biggest pain-points for user acceptance testing, is troubleshooting device specific bugs. While we can often test on Browserstack, it's not always sufficient, and sometimes you need to get hold of an actual device - or perhaps even the client's individual device, which is showing up a unique error because of some obscure accessibility setting or invasive app that they have installed which means the device is no longer displaying websites as it should.

While sometimes you just have to deal with device specific issues, it was interesting that there was broad support to drop default support for Android 4.0.3, which came out in 2011 and now apparently has quite a lot of issues with modern sites. Instead we've suggested Jellybean (Android 4.2) as the cut-off for an optimum mobile experience. Subtle but important.

For the record, here is where we landed. Not to say this is perfect for every new project, but it is a good base to start discussions with your producer or project manager.

Optimised

The website will be optimised for common (high usage statistics) operating systems and browsers. By ‘optimised’ we mean that the website will be fully functional and representative of the designs.

Windows 7 and newer:

• Internet Explorer: 11 and newer
• Edge: latest 2 stable versions
• Chrome: latest 2 stable versions
• Firefox: latest 2 stable versions

­ Mac OS X Mavericks and newer:

• Safari: 7 and newer
• Chrome: latest 2 stable versions

iOS7 and newer:

• Safari: 7 and newer

Android 4.2 (Jellybean) and newer:

• Chrome: latest 2 stable versions

Usable

The website will be usable for some uncommon (low usage statistics) operating systems and browsers. By ‘usable’ Wiliam means that the website will be functional and generally representative of the signed-off desktop designs, however users may experience some layout differences and the site may not be responsive to browser size.

Windows 7 and newer:

• Internet Explorer: 10

Not supported

The website will not be designed, developed or tested for all other older and low usage browsers. Older browsers that do not support responsive websites, are of very low usage, or are no longer supported by the manufacturer including:

• Internet Explorer 9 and below
• Browsers on Windows XP (now unsupported by Microsoft)
• Browsers on Windows Vista

Caveats

• We won’t support phone manufacturers that change the default Android Webkit browser.
• We make no guarantees on compatibility if/when Chrome moves from Webkit to Blink.
• Other semi-common Android browsers (including Firefox, Opera, Dolphin) may or may not work, and will be untested.
• Devices have resolutions with different pixel per inch ratios, so although the site should work, it may appear "bigger" or "smaller" on different phones.

* Market share is always a little bit of guess work. But here are my sources. Take your pick!

http://www.sitepoint.com/browser-trends-march-2016...
http://www.w3schools.com/browsers/browsers_stats.a...
http://gs.statcounter.com/#all-browser_version-ww-...