It’s over a year since Internet Explorer 7 (IE7) was released so I thought it was time to revisit the issue of browser market share, especially after the dramatic take-up of IE7 in the first few months.

However, first things first. Before getting to the detail of versions it’s worth looking at the overall market share for the browsers themselves. Over on Net Application’s Market Share site they’ve got this graph for November 2006 – 2007. You’ll see that whilst there’s some variation over the period, in general there’s not that much movement – Internet Explorer under 80%, Firefox hovering around the 15% mark, Safari reaching 5% and the rest less than 2%. The statistics collected by TheCounter for November 2007 show a roughly equivalent situation.

There’s a gradual shift from Internet Explorer (IE) towards Firefox and Safari that appears to be continuing from their earliest figures in December 2005, but only a few percent a year. Interestingly if you switch over to the operating system trends then you’ll see that there is an increase in the use of Macs of about 1.5%. That covers a fair proportion of the change, especially as Microsoft stopped developing IE for the Mac.

When it comes to the specific versions of the browsers the most significant development is the slow-down in the take-up of IE7, especially when compared to the use of the latest version of Firefox. The increases in the first few months (9% from November to December 2006 and 7% from December 2006 to January 2007) is of the same scale as the total increase from January 2007 to November 2007 – 11%.

Microsoft made the upgrade part of its automated Windows Update system, but it hasn’t been classed as critical. Presumably we’re now left with the users who either don’t use this facility or who ignore the optional upgrades – the majority are using Windows XP which supports IE7, so it’s not a matter of incompatiblity.

There is also the factor of the corporate IT policy, where the reasons for upgrading need stronger justification and have greater repercussions (e.g. support, training, etc.). For example, the US Department of Transportation is reported to have placed a moratorium on Microsoft upgrades.

So after sifting through all these figures, what’s the real significance?

For those of us developing web sites it’s the decline of IE6 into obscurity that we’re looking for. It’s only with version 7 that Microsoft have fallen in line with the standards that other browsers have supported for a number of years, and there seems to have been an interesting number of bugs to deal with as well. So having to deal with IE6, and until recently the different problems with the earlier versions, has added to the workload of developing and testing web sites but without increasing the features available.

There’s also the issue of whether what influences real users in their choice of browser. At a recent Bristol Skillswap there were a number of lively/provocative discussions about this subject. Looking through these statistics though, I can’t see that the majority of users make any conscious decision. Even in the refuge of the alternative community, the Mac, we find that most people stick with the first browser that is installed with their system – in this case Safari (even though it hasn’t been able to access online WYSIWYG editors & until the last month or so it hasn’t been possible to restyle buttons in forms).

That could mean that the balance could shift when/if mobiles and other devices start to be used to access the web in meaningful numbers, as Opera is far more dominant in that market.

Essentially the idea is that you have (at least) 3 grades of web browser:

• an identified list of browsers (+ operating system) that will take full advantage of the capabilities of modern browsers and will be tested thoroughly (grade A)
• an identified list of browsers that will be able to access core functionality & content, but will only have a representative sample tested (grade C)
• the rest that will either take conform to current web standards like the grade A browsers or be able to function at the minimum level of the grade C, however there is no testing and no support (grade X)

Which browser gets put into which grade is going to depend on their market share for your audience (or the audience you desire). So in Yahoo’s case it’s approximately 96% in A, 3% in C & 1% in X.
It also takes up the idea of progressive enhancement of web pages rather than the more established idea of graceful degradation – i.e. that the first concern is about presenting the core content & functionality and then building on that, rather than putting in the bells & whistles and then trying to ensure older browsers didn’t trip up.

I must admit that each time I wrote ‘gracefully degrade’ in a proposal or documentation it felt like I was writing the next verse in a Tom Waits’ song (“We’re decomposing as we go”). So not only does the change in perspective make sense, but it will also make me feel better if I have to think about it on a Monday morning.

Microsoft were reporting a higher market share a month ago (25% across all websites in the US compared to TheCounter.com’s 20%) as they notched up the 100 millionth IE7 installation. Given the vagaries of web statistics then it’s pretty much in the small ballpark.

An interesting aspect of the change in figures is that the increase is established (more or less) within the first few days of the month, and no significant variation thereafter. This is likely to change now that Windows Vista has been launched to the consumer market (on 31 January) as users upgrade or buy new machines (currently not shown in the operating system breakdown). So the stepped effect should be reduced in the future.

TheCounter.com offers a web site analysis tool for its clients and produces global aggregated result for free. The browser summary for November shows IE7 with 7% of the users (3rd in rank behind IE6 and Firefox). So far in December it’s moved ahead of Firefox (that is all versions of Firefox) with approximately 12% of market share. Over the same period IE6 was reduced from 76% to 70% – matching the rise in IE7. With approximately 80 million users in November and 17 million so far in December it’s fair to say that it’s a reasonable picture of current ‘normal’ web usage (the corporate market are likely to lag behind this take-up). The percentages and change in rank might not be absolutely accurate, but it is rare for a new browser version to be so prevalent so quickly.

The potential user base can be seen in the statistics for the operating systems being used. In December Windows XP is shown to be used by 83% of the users – all of these people will be offered the opportunity to automatically update to IE7. I wonder how long it will be before IE7 moves ahead of IE6?

If you haven’t check through your website with this version then we’d recommend that you do it as soon as possible. Microsoft have now come into line with how most other browsers understand the HTML and CSS that are used to implement the look & feel of a page. That’s good news from the point of view of web standards, but it may be that your site has been created with the assumption that IE is always a special case and incorrectly thinks it should make the same adjustments for IE7.

[In an earlier posting you'll find details of how you can have both IE6 & IE7 running on your computer using a couple of freebies from Microsoft.]

One way around this is to have virtual copies of Windows running on your computer, each with their own single, different version of IE. Microsoft have such a product and have now made Virtual PC 2004 free to download (Virtual PC 2007 version has gone into beta testing, so that’ll be the version that you’d have to pay for in the future). To run it you’ll need a reasonable chunk of memory (after all you’re running another copy of Windows inside your normal Windows), either Windows XP Professional or 2000 Professional (Home versions of either won’t work), and a license for another copy of Windows (the one that’ll be installed in the Virtual PC).

You can also get a free, time-limited version of XP with IE6 (Service Pack 2) set up from Microsoft as well. So you can upgrade your normal desktop browser to IE7 & then use this version to check that your pages still look fine in IE6. They’re intending to put another time-limited version up before this one expires, so you’ll need to update your copy every now & then. Obviously Microsoft don’t want to be giving away copies of XP at the moment.

If you’re looking to go further back in time then you’ll have to use earlier versions of Windows as you can’t install anything earlier than IE6 on XP. So you’ll need to use:

• Windows 98 (the original version not Second Edition) for IE4
• Windows 2000 for IE5
• Windows 2000 for IE5.5 (you upgrade from version 5 to 5.5 and you can find the appropriate files at the evolt.org browser archive – a good place to get other past versions of browsers)
• Windows XP or Vista for IE7

If you want to go the whole hog (as we do) then you’ll also want to have a duplicate of the same version of IE that you use normally but with the default settings (you might have changed them for your day-to-day work). For IE6 you could also have versions with and without Service Pack 2 applied. This was the update that put in the extra security features such as the pop-up blocker so it shouldn’t make a difference to how your website looks, but you might find differences in how it works.

The same technique can be used on other systems – I use Parallels Desktop for Mac as Microsoft’s Virtual PC for Mac doesn’t work on the new Intel systems (but that is a good option if you’re on a PowerPC Mac).