Don’t Trust Those Web Browser Statistics

At least, don’t trust the percentage of Internet Explorer, Firefox and Safari site visitors without looking carefully at the source of the data. Obviously, what matters most is your own site statistics rather than outside research numbers. But even then, the numbers may not be extremely accurate.

For many IT managers, Web browser statistics are interesting but irrelevant data—at least at first blush. You have some idea of the people who visit your public website, and a better perception of intranet users (since you may have some control of permissible browsers used on in-house computers). Or perhaps it doesn’t matter overmuch which browsers site visitors use, since your company (bless you) is adament about writing to Web standards and testing the site with Internet Explorer, Firefox, Safari, and every other conceivable browser.

However, browser trends can have major implications for the web designers who create the ever-more-Internet-centric applications the company runs. The data can be helpful data to the site’s webmaster. It behooves us to pay a little attention to the numbers.

Although web development ought to be a simple issue of writing to the published standards, in sad point of fact it is not. A web application probably looks very different in Firefox than in Internet Explorer, and each browser version has its own idiosynchracies. The result is a lot of cusswords, because developers have to put extra effort into ensuring their code works right on every browser. That isn’t trivial. (Firefox, for example, has several handy debugging tools such as Firebug; debugging IE is a guessing game.)

The usual IT answer used to be, “We’ll just support IE, then,” despite foot-stamping from developers (most of whom believe the right answer is to ensure that the code works for everyone) and certainly from site visitors. Thankfully, the general wisdom does seem to have evolved into an intention to support just about every common browser—but the dev team will prioritize based on which browser is most relevant. Thus, the statistics do matter.

Your initial urge may be to look at typical browser usage from IT research firms or from, really, anybody who takes the time to put out a press release on the subject. That information can be helpful, certainly, especially when you’re designing a new site and have little empirical data to work with.

But pay attention to the source of the data reported, because it is accurate only for the sites they watch. Some scale-tipping is obvious based on the target customers; Apple.com surely has more visitors using Safari than IE. Others are less so; W3 Schools publishes global ststatistics (which, incidentally, currently say that IE accounts for half of Web traffic), but the site is for Web professionals; its visitors are more likely to experiment with technology than your average consumer site. Geography can make a difference, too.

I’ve noticed that official-sounding sites tend to report higher use of IE than smaller hosts do—not that they agree, particularly.

My interest in the topic initially was sparked by a “data point” article we published, Microsoft Internet Explorer (IE) Still Dwarfing Mozilla Firefox, Others in Browser Market Share, which said 77% were using IE. Back in July, OneStat.com reported that 85% of Web surfers used IE. TheCounter.com pegs IE at 79 percent. And ADTECH, a provider of ad server technology, reported at the end of January that “Microsoft browsers continue to lead overall, with nearly 40 percent of UK ad impressions generated through Internet Explorer (IE) 7.x. This marks a significant increase over last year’s July figure of 36.3 percent. IE 6.x lost ground in the UK, dropping from 45.5 percent in July to 37.3 percent in January.” From these official pronouncements, you’d think that IE was the only browser that mattered.

However, according to the blog of Bluehost’s CEO, IE has far less dominance:

44.5% of the 1,000,000 unique visitors we had in November [2007] used Firefox!! That is amazing! In Europe Firefox usage is skyrocketing with many countries usage over 50%. Firefox usage in the US is less than in Europe, but is still noteworthy as it approaches the 20-25% mark.

That matches my far-from-scientific observations. W3Counter shows IE with two thirds of site traffic. I’ve looked at traffic for sites catering to a range of visitors, from Little League members to packaging companies to nontechnical nonprofit associations. I’ve corresonded with web designers and developers who report IE at 71 percent or so. A website for a hotel in the Austrian Alps has 76% of users on IE.

My point is: take the numbers with a grain of salt. I don’t think there’s any ill intent in all those sites claiming IE use in the 80+ percent range. I simply don’t think they’re the final word.

The most important browser statistics are (duh) the ones from your own site, because they’ll reflect your user base and you can draw useful conclusions from the information. For example, if you notice that the number of IE 6 users is falling and more people are adopting IE7, that should be reflected in the site design. I don’t mean that you should necesssarily write the apps for IE7 (by all reports, it still has many bugs that makes it tedious to design for) but IE7 is better in its support for web standards and apps can be designed accordingly. When IE 6 use declines to insignificance (which won’t happen anytime soon), you will be able to drop support for it.

For webmasters, browser statistics can provide a glimpse into visitors’ technical competence, if that didn’t follow based on the content you publish. Also, trends in browser usage can help webmasters know how to optimize the site for its primary users (that is, ensure that you’re fastest on those browsers; I’m not saying to ignore others).

Still, there may be more going on with these statistics. I’m not knowledgeable enough about this area of technology to speak authoritatively, but from my reading, several things can mess with the accuracy of the data collected. Says a 2004 article explaining how this stuff all works, “If your home page has 10 graphics on, and an AOL user visits it, most programs will count that as 11 different visitors!”

Plus, I’m bemused by data at a particularly good Wikipedia entry on the topic, which states (among many other interesting things) that statistics can be overestimated by such things as web browsers that refresh the page at a regular time interval, or extra files like CSS hacks and JavaScript hacks which are often sent to Internet Explorer. The implication is that more efficient browsers can get a demerit in site statistics, but I’m not smart enough to know if that assumption is true.

Bottom line, though: don’t automatically trust the numbers. At least without knowing where they come from.