The IE team published an in-depth post over the weekend, raising a few concerns about StatCounter’s methodology (or lack thereof) for reporting browser market share. Their points were interesting to consider, but one of them stood out to me:

You’ll notice some pretty big differences in the weighting of StatCounter versus Net Applications. First and foremost, the most populous country in the world, China, doesn’t make the top 20 for StatCounter, when in fact it represents the world’s largest internet population.

[…]

To further explore this problem, we re-ran the StatCounter numbers and weighted their publicly reported individual country browser share numbers by the CIA internet population data. This calculation would then represent a true country or geo-weighted view of worldwide browser data based on the actual world’s internet population.

It’s true that we should be wary of methodology issues that can creep into data extracted from analytics services that weren’t designed with aggregate statistics in mind. StatCounter’s data is often accepted at face value, without any detailed scrutiny. However, I believe this geo-weighting approach they’ve explored may be as flawed as the raw, unadjusted data itself.


Why does it matter?

What’s the use in splitting hairs about this? When I finished writing this post, I wondered the same thing. Why did I bother writing 1,000+ words about a random IE marketing post?

Microsoft is an incredibly metrics-driven company. You can be sure that these numbers are intended to make a case for ignoring certain aspects of Chrome and Firefox, prioritizing particular Internet Explorer features, and/or confirming that Internet Explorer is turning the tide against Chrome and Firefox.

As long as Internet Explorer is the baseline browser that ships with every Windows-based PC, massaging the numbers to downplay the popularity of what Chrome and Firefox are doing hurts us all. So, that’s why I’m taking the time to voice my doubts about this recent post.

What’s an “Internet user” anyway?

The motivation behind this geo-weighting approach seems to be that the CIA World Factbook reports China’s “Internet users” at a robust 389 million, compared to the US’ 245 million, whereas StatCounter only reports about one percent of the traffic it tracks as coming from IE-laden China. So, maybe it’s reasonable to argue that StatCounter’s numbers for IE usage should be adjusted proportionally.

Not so fast though. If you’re like me, you might be curious what actually counts as an “Internet user” first. Here’s the CIA World Factbook’s definition:

The number of users within a country that access the Internet. Statistics vary from country to country and may include users who access the Internet at least several times a week to those who access it only once within a period of several months.

Reasonable interpretations of that include counting every person that uses an Internet cafe a few times a year and everyone who checks email from an app on their phone as Internet users. Maybe it’s just me, but that seems like an incredibly vague statistic to warrant such a significant adjustment.

It gets worse though. Keep that definition in mind for a minute and let’s talk about another statistic that varies greatly between China and the US.

You can’t ignore mobile

The most current CIA World Factbook stats on mobile phone usage put China at a whopping 859 million subscribers versus 279 million in the US. While Internet access via mobile phone was insignificant in the pre-iPhone US, it has long been the norm in many Eastern countries (some research points to a full two-thirds of Chinese Internet access being via mobile phone).

Given that reliance on mobile access and the nearly 600 million mobile subscribers China has on the US, it’s reasonable to assume that a non-trivial chunk of the “Internet user” gap between the US and China is comprised of mobile users. In fact, the data suggests to me that China’s higher number of total Internet users may actually represent less access via traditional PC than in the US.

Speaking of those mobile users, while we’re swimming in a sea of iOS, Android, and Windows Phone devices in the US, it’s easy to forget that the venerable Nokia feature phone is still the dominant Internet-enabled mobile device in many countries (China included). Behind Nokia, manufacturers like Samsung, LG, Motorola, and Sony Ericsson also sell a combined tens-of-millions of feature phones with Internet capabilities in markets like China each year.

More importantly, only a tiny fraction of mobile phones run any version of Internet Explorer, and no feature phone that I’m aware of runs Internet Explorer. When you consider this aspect of the CIA’s data, it seems plausible that IE’s market share in Eastern countries may actually need to be adjusted downward if anything.

Why not version-weighting too?

The IE team’s blog post specifically mentioning China reminded me of The Internet Explorer 6 Countdown site that Microsoft itself launched last year. One of the datapoints that really stands out there is that China represents a solid majority of Internet Explorer 6’s remaining user base (largely due to pirated copies of Windows XP that can’t easily be upgraded to IE7+):


So, even if you believe that China’s browser share numbers should be adjusted upward to compensate for the potential discrepancy in StatCounter’s data, nearly a full quarter of those users are still stuck on IE6. That’s hardly something to be proud of in 2012, but even more embarrassing in a discussion about IE’s market share as compared to Chrome and Firefox’s rapidly upgrading user base.

Why so mean?

My aim here isn’t to pointlessly bash the Internet Explorer team.

They really are doing some great work on turning IE around. I think IE9 was an inflection point where the Windows team got serious about the web again, and IE10 is poised to bring Internet Explorer forward to a place where it can focus more on moving the entire web platform forward like it did in its heyday 10+ years ago.

Yet, no one benefits from this sort of accounting trick; not even the IE team itself.

  • If they believe that IE9 has similar or higher market share to Chrome, it’s easier to dismiss features like HTML5 Notifications as fringe APIs that can wait till later, when implementing them in IE9 or IE10 would’ve been a win for both developers and users.
  • If they believe that IE9 is more popular than it is and that IE10 is likely to follow in those footsteps, they might be more inclined to continue ignoring the de facto standard touch events API that iOS and Android both support and create a proprietary API instead.
  • If they believe Chrome (and Firefox, to a lesser extent) has been relatively unsuccessful with a rapid release cycle and forced upgrades, it’s easier to remain confident that Internet Explorer’s slow release and upgrade cycle is still as viable as ever.

Rather, I think if Microsoft wants to truly bring Internet Explorer back to dominance, they need to take stock and reevaluate some of their sacred cows:

  • Multiple versions side-by-side: A modern browser may either auto-update on a regular basis or be able to install every one of its currently active versions side-by-side. For example, it’s trivial to test in many versions of Firefox on the same machine and you’ll never need to test in a year-old version of Chrome.
  • Support for XP and Vista: This is related to the previous item, but punting on support for older operating systems is directly responsible for the awful fragmentation that Internet Explorer continues to suffer. Why have Chrome and Firefox figured out how to implement WebGL 3D acceleration on Windows XP, but the company that actually created Windows XP can’t get IE9’s 2D acceleration to work on it? Come on.
  • Embrace the platform: I actually think some of IE9’s proprietary features are compelling. IE9’s work on 2D acceleration spurred the rest of the browsers to take that feature more seriously and site pinning leapfrogs Chrome’s application shortcuts by a long shot. However, that doesn’t make it okay to ignore de facto standards like the HTML5 Notification API and the touch events API, where other browsers have been playing nice with each other.

What do you think?

If you’re part of the 80% of people who visit my site in Firefox or a WebKit browser, what would it take to make Internet Explorer your default browser? Is it possible?

If you’re part of the 8% of people who visit my site in Internet Explorer, why are you using IE? Corporate policy, or are the rest of us missing out on something?