Tuesday, March 9, 2010

Operating Systems and Market Share Statistics

Did you know that 55% of statistics are made up on the spot? (or maybe it was 68% I don't recall) If you have ever taken a statistics class you know that data is everything, but it's not just about the data itself. Whenever you see numbers you have to ask yourself:
  • Where did these numbers come from?
  • Is it a credible source?
When it comes to market share and operating systems I've heard lots of numbers thrown around over the years and it is difficult to figure out which numbers you should put your faith in on this subject. For instance if your take a peek at the Wikipedia page on the subject it shows the following figures:
  • Windows: 87.67%
  • OSX: 5.9%
  • Linux: 1.1%
  • Other: 2.08%
Those seem like fair figures right? Odds are many will stop there and take them as fact, however take a peak at some numbers from Marketshare.Hitslink.com:
  • Windows: 92.12%
  • OSX: 5.02%
  • Linux: .98%
  • Other: 1.88%
They don't quite match... Windows is ahead by any rate (which we should expect) but still, take a look at numbers from a couple other websites. Statowl.com:
  • Windows: 88.49%
  • OSX: 10.87%
  • Linux: .41%
  • Other: .12%
Wait, what? OSX has over a ten percent market share? That almost twice what our first two sources cite... How about a forth source? W3counter.com:
  • Windows: 84.33%
  • OSX: 8.12%
  • Linux: 1.55%
  • Other: 6%
These numbers from W3Counter seem to be the middle ground of the four sources for OSX, however note the fact that their count for the Windows Operating system is much lower and their count for Linux operating systems in much higher. One thing to note about these numbers is the much larger "other" category. Even further curious I find is that Google's Android operating system, is thrown in with "other" on W3Counter even though it is a derivative of Linux.

What are other things to consider? For starters how do these places collect their statistics? What websites do they pull their data from? The content of a webpage very much determines the type of operating system that a person is likely to view it on. For instance these are the operating system statistics from the last month for my own (primarily Linux-focused) blog :
  • Windows: 44.4%
  • OSX: 8.03%
  • Linux: 44.03%
  • Other: 3.54%
Care to guess which operating system that data set is skewed towards because of the page content?

Beyond just looking at the source of web statistics of operating systems, when it comes to the global market as a whole, you have to consider the countless systems that are offline or are rarely connected to the internet. Unlike OSX where you can count the systems by the amount of Apple hardware sold or Mircosoft's Windows where they can count the number of activations, a single Linux ISO download can account for multiple (sometimes even hundreds) of offline (or online) installations.

Truth be told, will we ever truly know the precise market share of each operating system? No, we will not. From my four sources here (and others you can find around the internet) I'm inclined to believe that currently Windows floats somewhere around 88%, OSX around 8%, Linux somewhere close to 2%, and the rest can get lumped into that wonderful "other" category.

What do you think? Know of another credible source for market share statistics regarding operating systems that I didn't mention? Let me know!

~Jeff Hoogland


  1. Here ar my statistics -
    Site language: German
    Site topic: Law

    2007: Windows 82.40%, Linux 16.75%, MacOS 0.73%
    2008: Windows 85.54%, Linux 12.36%, MacOS 2.10%
    2009: Windows 85.90%, Linux 10.44%, MacOS 3.61%
    2010: Windows 79.70%, Linux 12.65%, MacOS 7.65%

    I already noticed that lawyers and jurists in general like Macs. You can clearly observe this, when visiting talks and lectures.
    But the Linux percentage strikes me as interesting, since members of the legal profession usually are not interested in tech. Maybe these are Linux pc pools at universities, maybe a significant number of law students using Linux for cutting cost. Or maybe the official numbers are simply not related to actual OS usage in any known way.

  2. Here are some statistics from 5 sites I monitor. The timescale is one year. The middle figure is number of visitors. Taken from Google Analytics.
    I think the figures could be skewed in favour of linux since I visit the sites regularly as the linux-using web developer. (Especially for the lower-visited sites, but the other sites are all above 1%)
    Site 1- Worldwide - Non Profit
    Windows 534 86.83%
    Macintosh 41 6.67%
    Linux 37 6.02%

    Site 2 - Worldwide non profit
    Windows 7,619 92.54%
    Macintosh 316 3.84%
    Linux 236 2.87%

    Site 3
    IT related conference site
    Windows 325 60.75%
    Linux 156 29.16%
    Macintosh 47 8.79%

    Site 4 Worldwide non-profit
    Windows 3,257 91.72%
    Macintosh 146 4.11%
    Linux 108 3.04%

    Site 5 - Ireland - Semi-Governmental
    Windows 32,981 92.79%
    Macintosh 2,012 5.66%
    Linux 399 1.12%

  3. Yeah, a fantastic topic to discuss. Being a mathematician AND a Linux enthusiast I always get overly excited when people start citing OS stats. I think you raised some good arguments that make the entire discussion somewhat ridiculous. People throw numbers and stats only when they serve their purpose - even when they claim to be objective.

    Here anyway some stats from sites I maintain. Things to consider: Sites are in Finnish so 99% of the users are coming from Finland. The linux stats include also traffic generated by me from home (at office I run Windows).

    Site 1 (blog)
    Locally interesting topics (including politics, technology, general rant). Operating systems not covered in anyway.
    * windows 74%
    * Mac 14%
    * Linux 10%
    This truly blows my mind. There was a post about Spotify that really got people excited ... but that should not skew the data towards Linux in anyway.

    Site 2
    News and info about gadgets
    * Windows 88%
    * Mac 5%
    * Linux 5%

    Site 3
    Online sales channel (flea market)
    * Windows 94%
    * Mac 3.5%
    * Linux 1.7%

  4. You are right but there is one more important issue about these market share data. Does market share tells us something about Linux? Is market share the appropriate measurement of its success, does it tell us something about the future of Linux? Well, I think market share is completely biased way to measure anything related to Linux. Linux is mostly non-commercial OS. Expecting Linux to beat commercial Windows in market share is like expecting Manchester United to win NFL. It may look similar, it can even have the same purpose but at the end it is completely different game.
    Greetings from the Czech Republic ;-)

  5. It would be interesting to get a google search numbers, those will probably give the best snapshot of current os usage.

    W3 counter is probably the most legit out of the above sources. That being said you raise a valid point about Linux and Android being separated but they probably should be. I imagine that Google will split Android away from the main linux kernel cycle in the future similar to what Apple did with OSX.

    Smart Phones and MIDs will probably cause a drastic rise in the OSX and Android and linux numbers in the future.

    I used to get much more excited about this stuff before but now it's become a ... meh issue. All that matters is that any of these OSs have decent browsers now and access to flash, soon html5 will bring things even closer together. I'm much more concerned about traffic magnitutde than what folks are using these days.

  6. You can not count Windows and OSX by hardware sales because the hardware all works well with free software. My PPC Macs do not run OSX, they run GNU/Linux. There's a whole subculture of people who want to run OSX on cheap PC hardware and, of course, Intel, Mips and ARM work best with GNU/Linux. If you believe sales figures, you would believe that 20% of the world uses Vista. Vista never got past 9% by better counters and may actually have never exceeded 5% because Vista was so broken. Windows 7 is not much better but it is sucking up all of Vista's market share. That leaves roughly 80% of the Pee Cee market contemplating their hard choice of suffering longer under 10 year old XP or tossing out all of their hardware and software to start the Windoze cycle again, or moving on to something that respects their hard won habits and works well with their old hardware. This is a no brainer for most people. GNU/Linux is the right choice for everyone.

  7. Here's my results: 1314 visiters during last 4 days. They are mostly non-geeks, mostly non-academic, just ordinary people, 95% from Finland.

    1. Windows XP 51.1 %
    2. Windows Vista 21.7 %
    3. Windows 7 15.2 %
    4. Mac OS 5.4 %
    5. Linux 4.3 %
    6. Windows 2003 1.1 %
    7. Unknown 1.1 %

  8. 1. Android should be counted separately from Linux — more precisely, GNU/Linux — because while it uses the Linux kernel, it doesn't AFAIK use the GNU utilities.

    2. The statistics from Systematics are more meaningful because they count the different versions of Windows separately. After all, the kernels are different, whereas the GNU/Linux kernels are presumably the same.

  9. Even so, I'd like to know the numbers behind the percentages.
    Many of the Android figures are probably on new specialized hardware, whereas many non counted Linux figures are on existing or older x386 hardware. I myself install the system on hundreds of P4 1.5GHz 512MB machines used for school home work by the disadvantaged.
    Buying XP and Office is hard and expensive so Linux is a better choice as it produces the required end result.
    Counting all the hardware produced since 2000 could give basis for some reasonable guesses, especially as many children from poor families cannot afford an Internet link.

  10. I completely agree with the above comment, the internet is with a doubt growing into the most important medium of communication across the globe and its due to sites like this that ideas are spreading so quickly.

    social network

  11. All stats quoted here are skewed one way or another. But so is the question. In terms of "average home user", Systematics' stats are probably the most accurate (based on what I know of my own circle of friends'n'relations.) The other stats tell you a lot more about the nature of the sites, eg, the ones that score Linux at 5% or above are for computer pros, who are far more likely to make a deliberate choice of Linux. Etc.

  12. The non-USA statistics confirm something I long suspected. In the United States it is clear that Windows is probably 80% of the market with Mac OS X at around 15% and Linux taking the remainder. The old 1% Myth of Linux I think stems from the fact that the websites that do the counting like the Statcounter network are not popular among Linux users and also in English and oriented towards Americans and not Europeans or Asians. This creates a skew because the numbers for Linux based on the above indicate that it is globally a huge market with around global 25% and probably more than that in non-English European, Asian and Latin-American countries. The stats that quote 90% for Windows take a snapshot of a small set of users who regularly buy new PC's every upgrade cycle, are non-technical and almost all live in the United States...

  13. The blind imperialism with which United-Stadians are running their lives and the rest of the world is well known!

    How can linux have any significant market share? It is generally not sold! Linux has therefore no market share by design!

    Now, if one is referring to the linux user base, that could become a fruitful discussion!