Net Applications: Windows 8 up slightly in April to 3.84 percent

The good news? Windows 8 keeps adding to its market share among PCs worldwide. The bad news? According to the just updated numbers from research firm Net Applications for all of April 2013, the growth of Windows 8 is still moving at a very small pace.

Net Applications's website shows that worldwide, Windows 8 was installed on 3.84 percent of all PCs in April, compared to 3.31 percent in March. Net Applications breaks down Windows 8 in three parts; Windows 8 at 3.82 percent, Windows 8 Touch at 0.02 percent and Windows RT Touch at 0.00 percent (no, that's not a typo).

It's not a shock to learn that Windows 7 is still the most popular PC OS in the world. The numbers from Net Applications for April show it was installed on 44.72 percent of PCs, down by just 0.05 percent in March where it had 44.77 percent of the PC market share.

Windows XP, now less than a year away from its support cut off date from Microsoft, was still used by 38.31 percent of PC's in April, down compared to 38.73 percent in March. Windows Vista is also still sliding; it had a market share of 4.79 percent in April, down from 4.99 percent in March.

Source: Net Applications | Image via Net Applications

Report a problem with article
Previous Story

Windows Server Hyper-V can now run on SUSE cloud-based infrastructures

Next Story

IE10 claims 6.02 percent of web browser share in April

31 Comments

Commenting is disabled on this article.

So let me guess, Windows 8 and Vista are run on more machines than flavors of OSX and they are the considered epic failures of an OS

MrHumpty said,
So let me guess, Windows 8 and Vista are run on more machines than flavors of OSX and they are the considered epic failures of an OS

Kind of interesting how the marketshare of OS X has become the new yardstick for the success of Microsoft's OS...

CSharp. said,

Kind of interesting how the marketshare of OS X has become the new yardstick for the success of Microsoft's OS...
Considering the model OS is usually OSX I love those stats.

I'm actually pretty amazed that Win8 has gained the numbers it has so far. There isn't really a compelling need to move into a new OS and there isn't a compelling reason to replace hardware these days. Back when Win7 was out... that wasn't remotely the case.

Edited by MrHumpty, May 1 2013, 3:23pm :

MrHumpty said,
So let me guess, Windows 8 and Vista are run on more machines than flavors of OSX and they are the considered epic failures of an OS

If you repeat a statement often enough, people will believe it.

It's fascinating that the majority still call Windows Vista a failure, and yet almost no one ever mentions *why* they feel this way.

What's worse, these same people praise Windows 7 like it's a completely new operating system, when it isn't.

I'm glad that I'm using Windows 8 Enterprise x86, which makes better use of my oldish laptop than Windows 7 Ultimate SP1 x86. Definaltey boots faster and once you get used to the Start screen and the Power user menu it does an amazing job to bring some freshness into computing.

So "Windows 8 Touch" (whatever that even means) somehow lost over 80% of its market share between March (when it was 0.12%) and April? How does that even make sense?

I agree. This doesn't make any sense at all. OEM's (including Microsoft Surface tablets) sold over 3 million tablets last quarter alone. That number should be up a good percentage if anything.

The only thing I can think of is that people with touch-screen devices may be using apps for information instead of the web browser, which may make these numbers look falsely low.

The same really goes for all of Windows 8 devices. I have a non-touch laptop (now use a Surface Pro) in which I used more of the apps than the web browser. If this is the case, Windows 8 numbers are actually better than NetApplications can capture.

contextfree said,
So "Windows 8 Touch" (whatever that even means) somehow lost over 80% of its market share between March (when it was 0.12%) and April? How does that even make sense?

They might have changed the way they're recording the data. Either that, or It just shows that there is a significant amount of noise in their data collection. I have no idea why Neowin diligently reports on Net Applications data month after month, when they have some of the worst statistical methods out there for this kind of work.

They have an exceedingly small network of 40000 sites which generates only a small number of pageviews per month. They then take this number and cut it down to unique visitors only, leaving only 160 million pageviews per month. They further manipulate the data (to account for their small network) by weighting it by country based on Internet usage statistics that are 4 years outdated. All this massaging leaves a lot of error (which they do not report).

Other sites like statcouner, which uses two orders of magnitude more data, paint an increasingly different picture. For April, they show that Windows 8 was on 5.06% of desktops (Windows,Mac,Linux). This is a substantial discrepancy that appears to be based in the fundamentally small sample size for Net Applications' data.

ModernMech said,

They might have changed the way they're recording the data. Either that, or It just shows that there is a significant amount of noise in their data collection. I have no idea why Neowin diligently reports on Net Applications data month after month, when they have some of the worst statistical methods out there for this kind of work.

They have an exceedingly small network of 40000 sites which generates only a small number of pageviews per month. They then take this number and cut it down to unique visitors only, leaving only 160 million pageviews per month. They further manipulate the data (to account for their small network) by weighting it by country based on Internet usage statistics that are 4 years outdated. All this massaging leaves a lot of error (which they do not report).

Other sites like statcouner, which uses two orders of magnitude more data, paint an increasingly different picture. For April, they show that Windows 8 was on 5.06% of desktops (Windows,Mac,Linux). This is a substantial discrepancy that appears to be based in the fundamentally small sample size for Net Applications' data.

I find it funny how you complain about NetStats properly weighting their data by doing unique views only, geo weighting, and filtering out Chrome's pre-rendering.

MS addressed the two services last year: http://blogs.windows.com/ie/b/...owser-usage-share-data.aspx Now I know that report is produced by Microsoft. But their logic is sound. If you can't make your sample random, you need to do *something* to weight it properly. Regardless of network size, you do need to weight it properly.

MS's main concern seems to be the geoweighting issue. Statcounter addresses this explicitly:

We do not impose artificial weightings on our stats - this is a conscious and deliberate decision. Weighting stats means that the stats are only as good as the weighting methodology used. If the weighting data is inaccurate or out of date, then it renders the data completely incorrect. Further, applying a weighting factor to inaccurate data does not turn it into meaningful information - no matter what weighting factors are applied, the geographical spread of the initial stats is very important. For these reasons, we choose not to weight our data.

We would also have a significant difficulty in choosing and applying any system of weights to our data. For example, how would we handle changes in the weights? Would we rerun and restate all our prior stats? Would we apply a smoothing factor? How often would we update the weights?

They go on to talk about the problems with using the CIA data that NetApplications uses:

The data is collected at different times and may be out of date. For example, in March 2012, the CIA data per country is based on figures from 2001, 2008 and 2009 for different countries.

The data is vague and inconsistent - "This entry gives 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." In other words, users who access the internet several times per day are given equal weight as those who only visit an internet café once per month - this seems to be inherently problematic.

No distinction is made between mobile and non-mobile users. It would not, therefore, be correct to weight browser stats using the CIA data. In China, for example, some estimates put mobile internet usage as high as 66%. Weighting browser stats without adjusting for potentially significant Chinese mobile internet usage would incorrectly inflate the stats for desktop browsers. This could significantly overstate the stats for desktop browsers which are popular in China. Similar problems would be encountered in other countries with high mobile internet usage.

Notice that last point, which the IE Marketing Manager neglected to mention.

This CIA data is at least 4 years old and very inconsistent. Net Applications offers no guidance on how they weight past data, which makes comparisons to prior data very murky (for example, comparing Vista growth in 2006 to Windows 8 growth in 2013, which so many like to do with Net Applications data).

With Statcounter the data is straightforward: from N people, x% of those used Browser/OS 1 while y% used browser/OS 2. With a sample size two orders of magnitude larger, their data has a much better hope at being representative of people who are actually using the internet.

Seriously, do you think starting out with data that isn't clearly representative and weighting it based on 4-12 year old data collected in an ambiguous and inconsistent way somehow makes the data representative?

By the way, Statcounter does filter out pre-rendering, and unique views have their own problems: i.e. how do you define a unique visitor? NetApplications doesn't say. What if that visitor uses IE once and Chrome 40 times on a given site. Are they given as much weight? Statcounter's data indicates the Chrome browser was used 40 times and IE used once. NetApplications records that both Chrome and IE were used once, or maybe that IE was used and Chrome not at all depending on how they define a unique user. How exactly do you come to the conclusion that this is the "proper" way to do things.

ModernMech said,
MS's main concern seems to be the geoweighting issue. Statcounter addresses this explicitly:

They go on to talk about the problems with using the CIA data that NetApplications uses:

Notice that last point, which the IE Marketing Manager neglected to mention.

This CIA data is at least 4 years old and very inconsistent. Net Applications offers no guidance on how they weight past data, which makes comparisons to prior data very murky (for example, comparing Vista growth in 2006 to Windows 8 growth in 2013, which so many like to do with Net Applications data).

With Statcounter the data is straightforward: from N people, x% of those used Browser/OS 1 while y% used browser/OS 2. With a sample size two orders of magnitude larger, their data has a much better hope at being representative of people who are actually using the internet.

Seriously, do you think starting out with data that isn't clearly representative and weighting it based on 4-12 year old data collected in an ambiguous and inconsistent way somehow makes the data representative?

By the way, Statcounter does filter out pre-rendering, and unique views have their own problems: i.e. how do you define a unique visitor? NetApplications doesn't say. What if that visitor uses IE once and Chrome 40 times on a given site. Are they given as much weight? Statcounter's data indicates the Chrome browser was used 40 times and IE used once. NetApplications records that both Chrome and IE were used once, or maybe that IE was used and Chrome not at all depending on how they define a unique user. How exactly do you come to the conclusion that this is the "proper" way to do things.

The size of your sample is not what makes it representative... it's randomness is.

They are both a factor. If your sample size is a single person from a population of a million, no matter how random you choose that one person he will not be representative of the population statistics. For the type of data in question, the member network *necessarily* biases the sample. The only way to counter act this is to 1) weight based on the demographics of each network member or 2) have a large enough network to counter-act the biasing effect of any given member site.

For instance, if a stat tracker has a network of only english speaking sites related to business, no matter what kind of geolocation weighting you do, your sample will still be severely biased towards people who speak english and are interested in business.

Net Applications has a network of only 40,000 sites to try and capture a representative sample of the entire world's interent users. I just don't buy that this could ever yield useful statistics, no matter how they try to compensate by massaging the data. Garbage in, garbage out.

This is a story that can just write itself on the first of every month with the latest numbers plugged in. I wonder if one can script that...

Just look at that kick ass percentage for that total POS Vista though!! WOOT!!

Windows 8 is about the equivalent of Vista due to that stupid start screen, amongst a few other things.

Frankly,
I'm glad XP is still kicking ass. It was and still is an awesome OS, even if it is ancient! That was one OS MS got right, along with Windows 7!!

cork1958 said,
Just look at that kick ass percentage for that total POS Vista though!! WOOT!!

Windows 8 is about the equivalent of Vista due to that stupid start screen, amongst a few other things.

Frankly,
I'm glad XP is still kicking ass. It was and still is an awesome OS, even if it is ancient! That was one OS MS got right, along with Windows 7!!

It's funny how you praise Windows 7 yet criticize Vista. If you must spread your hatred for it, then at least have the courtesy to explain why.

cork1958 said,
Just look at that kick ass percentage for that total POS Vista though!! WOOT!!

Windows 8 is about the equivalent of Vista due to that stupid start screen, amongst a few other things.

When XP came out it was just as **** as every other system. Service pack 2 made is usable...
vista was running on my notebook desktop replacement for 4 years without problems. maybe people should have bought proper hardware not trying to run 10 year old machines with a new system
Frankly,
I'm glad XP is still kicking ass. It was and still is an awesome OS, even if it is ancient! That was one OS MS got right, along with Windows 7!!

cork1958 said,

Frankly,
I'm glad XP is still kicking ass. It was and still is an awesome OS, even if it is ancient! That was one OS MS got right, along with Windows 7!!

You are crazy. Windows XP has literally no security, you need to install an A/V software, and even A/V softwares can only go so far.

Windows XP isn't even fast, have you booted up a 2-year old XP installation? Takes forever. Windows 7 is so much faster than XP on today's hardware.

I still don't get how this is impressive. Windows 8 is the only Windows OS sold in stores. If you want a new PC, your are stuck with 8. This is not even news.

News flash: the 2008 Mercedes E350 loses a few percentage. The 2012 Mercedes E350 gained 3.5% this month.

JHBrown said,
I still don't get how this is impressive. Windows 8 is the only Windows OS sold in stores. If you want a new PC, your are stuck with 8. This is not even news.

News flash: the 2008 Mercedes E350 loses a few percentage. The 2012 Mercedes E350 gained 3.5% this month.

Thats the same with any OS release I don't see how Windows 8 is any different?

Windows 8 is not competing with just Windows. It's competing with OSX, iOS, Android, Chrome OS, and Linux. Fact is, there is more choice in OS now than there ever was in history, and yet Windows 8 is STILL the fastest growing OS out there today.

As much as you refuse to admit it, I can go out to the store *today* and buy any of the above machines instead of a Windows 8 device. Many people are doing this, as evidenced by declining PC sales. I can even go online and buy a Windows 7 machine if I wanted; all the major OEMs still offer them. But again, despite this, Windows 8 is growing faster than any of them.

Statcounter has Windows 8 marketshare at 5.06% of desktops for April (4.75% when you add in iOS and Android). That's quite an achievement for an OS you and others claim is terrible every day in every thread on Neowin.

Windows 8 steadily acquires 0.5% of market share each month. With that growth rate it'll be installed on 15% of all PCs in two years. Impressive!

lexp said,
Windows 8 steadily acquires 0.5% of market share each month. With that growth rate it'll be installed on 15% of all PCs in two years. Impressive!

If it holds at 0.5%, it could very well change once we see better priced touch devices. Right now any really good device is also pricy.

lexp said,
Windows 8 steadily acquires 0.5% of market share each month. With that growth rate it'll be installed on 15% of all PCs in two years. Impressive!

Well - Win XP still has 33% - Win7 must be a really impressive system except for half of the user

primexx said,
already has more than the 2 latest versions of OSX
You do realize your talking about an OS that is used by billions and comparing it with a very small market in OSX?

JHBrown said,
You do realize your talking about an OS that is used by billions and comparing it with a very small market in OSX?
Does it matter? Eveeryone is blaming that Windows 8 is a complet failure and nobody wants it, however, it looks like people like it more than OS X.

Looks like that is gonna happen sooner than later.

Acer to launch affordable Windows 8 touchscreen laptop ($399 Touchscreen laptop)

http://www.itproportal.com/201...ndows-8-touchscreen-laptop/

Tablets are coming down in price too.

Acer Iconia W510-1674 Tablet (Windows 8 Intel Atom)

http://www.microsoftstore.com/...lqd0-5yhC9vJwoWL6pJCF43z4uw

Once these low priced devices start to become available in stores that's when Windows 8's market share will rapidly increase.

Studio384 said,
Windows 8 is a complet failure and nobody wants it, however, it looks like people like it more than OS X.

It also "looks like" people like the decade-old Windows XP much much more than modern OS X. Funny how that works, eh?

primexx said,
already has more than the 2 latest versions of OSX

Problem is that Windows competes with itself not IOS......