Microsoft picks Net Applications for web browser stats

For a while, we have been reporting on Microsoft's claims about the market share rise of Internet Explorer 9 among Windows 7 users. On the first of each month, Microsoft posts up a new entry on their official Internet Explorer blog, such as they did on March 1st, with their spin on how IE 9 is doing among Windows 7 users compared to Google's Chrome and Mozilla's Firefox browsers.

Microsoft based its results via data drawn from Net Applications. However, there is another company, StatCounter, that offers up its own web browser market share data results. In a new entry on the IE blog, Microsoft's Roger Capriotti gives the company's reasons why it prefers Net Applications' data gathering techniques over StatCounter.

One reason is that StatCounter gathers data from Chrome even when the web browser is “prerendering” some web pages. Users of Chrome sometimes don't actually visit the pages that Chrome is prerendering. Microsoft points out that starting in February, Net Applications removed Chrome prerendered web sites from its main traffic statistics, with the company saying that "prerendering in February 2012 accounted for 4.3 (percent) of Chrome's daily unique visitors.”

Another reason is that StatCounter just checks absolute page views for its information. Microsoft says that Net Applications also considers "geoweighting” as part of its results. That means it also considers data based on the overall Internet user populations of each country. Net Applications uses information gathered by the CIA to provide country Internet usage.

Net Applications says:

For example, if our global data shows that Brazil represents 2 (percent) of our traffic, and the CIA table shows Brazil to represent 4 (percent) of global Internet traffic, we will count each unique visitor from Brazil twice. This is done to balance out our global data.

By contrast, StatCounter ranks China as the 22nd biggest country for Internet users based on the number of page views StatCounter claims people in that country generate. However, StatCounter's numbers come from the site coverage they have from China when in fact the country has the largest Internet population in the world.

Microsoft decided to compare StatCounter's web browser data with the same information plus adding the real Internet user population of each country. The results are shown in the graphs above where Internet Explorer actually gained nine percentage points when the real Internet population is factored into StatCounter's results. Both Chrome and Firefox lost percentage points.

Finally, while StatCounter uses pure page views for its web browser stats, Net Applications uses unique visitors. Microsoft believes that simply using page views can be misleading when coming up with web browser market place data and can even lead to false results due to Internet bots that can generate lots of page views.

Image via Microsoft

Report a problem with article
Previous Story

TechSpot: HP TouchSmart 520 All-In-One Review

Next Story

IE8 turns 3, still commands 16% of the market

30 Comments

Commenting is disabled on this article.

I've seen some numbers that were not published on that recent blog by Micosoft.

What they are NOT saying is this: excluding China, the statscounter numbers are pretty much bang on. http://i.imgur.com/Zjb1O.png

Looks to me like Internet Explorer is relying on the Chinese market to boost the figures... the Chinese market where Steve Balmer said 90% of windows machines are running pirated software not so long ago... Just saying...

I don't wanna see IE sucks, but it really open web pages slowly and the Javascript error message is everywhere.

It is not the IE blog though. It's the IE marketing blog. Avoid taking anything from WindowsTeamBlog.com seriously. It's full of marketing propaganda.

xpclient said,
It is not the IE blog though. It's the IE marketing blog. Avoid taking anything from WindowsTeamBlog.com seriously. It's full of marketing propaganda.

This advice sounds legit, at least.

xpclient said,
It is not the IE blog though. It's the IE marketing blog.

Marketing firms care about numbers more than you think. Ideally, all decisions are based on sound analysis and corroborated methodologies.

PR firms, on the other hand, spin things right round...

I'm wondering what exactly is happening since Fx has the most users which block such counters... Currenlty the most users who use a Privacy-List are Fx users... Does this have an impact or still they count them somehow?

PC EliTiST said,
I'm wondering what exactly is happening since Fx has the most users which block such counters... Currenlty the most users who use a Privacy-List are Fx users... Does this have an impact or still they count them somehow?

I'm not sure if those blockers (I'm assuming you're talking about software like AdBlocker Plus) block stat counters. Those kinds of software are normally used to either clean up a web page or stop advertisers from tracking them; not keep stat counters from counting their browser. but I don't really know

Addons like Ghostery and RequestPolicy probably do unless you specify to allow it. And both are ranked high in popular security addons listing.

@Matthew, I'm talking about lists like "Easyprivacy" of Ad Block Plus, not the normal "default" list, but still... And there are other add-ons like that one LaP has mentioned, i.e., "Ghostery".

PC EliTiST said,
@Matthew, I'm talking about lists like "Easyprivacy" of Ad Block Plus, not the normal "default" list, but still... And there are other add-ons like that one LaP has mentioned, i.e., "Ghostery".

hmmm, I guess someone could look through all the sites they block, but apart from that I don't know :\

For example, if our global data shows that Brazil represents 2 (percent) of our traffic, and the CIA table shows Brazil to represent 4 (percent) of global Internet traffic, we will count each unique visitor from Brazil twice. This is done to balance out our global data.

Strange i never learned such practice in my university stats classes. I must admit it was something like 15 years ago and as a computer engineer it was introductory classes but back then we learned to calculate a margin of error for a giving sample.

That would be fun. Let's do a pool with 100 people. Then let's not talk about any margin of error and say the result of the pool represent to whole USA because we say 100 x 3 130 000 = 313 000 000.

seems like no one actually read the article. I mean, it could be biased because they count fewer countries, but counting page views over unique users is nonsense.

JohnnyMartins said,
LoL Net Application is completely IE biased. No other metric provider even comes close to that!

... Umm sure? Other statistics often say IE is around 50%..

Pretty strange decision. Net Application statistics has over 25% margin, because being collected just in few countries. That's highly US centrist statistics.

Enron said,
Maybe we should all use SubZenit's Vision Metrics. If he doesn't see it, it doesn't exist!

I never saw I don't see the metrics, but by far - those aren't the real one, I'm pretty god damn sure about that, and their metrics aren't too relevant since they fetch info's only from a few countries.

SubZenit said,
Pathetic attempt Microsoft, you keep saying you gain more and more users, altough I barely see IE users anymore...

How do you know? Do you walk around asking every person you see what browser they use? Just asking.....

Quapps said,

How do you know? Do you walk around asking every person you see what browser they use? Just asking.....

He said, "pretty god damn sure about that". If anything I'd say case closed, QFT and all of that. "Pretty god damn sure" is gospel in some parts.

cork1958 said,
We'll do what ever it takes to make the numbers look better, dang it!!

I was about to say the same thing lol.


Net Applications says:

For example, if our global data shows that Brazil represents 2 (percent) of our traffic, and the CIA table shows Brazil to represent 4 (percent) of global Internet traffic, we will count each unique visitor from Brazil twice. This is done to balance out our global data.


I'm not surprised Microsoft prefers NetApplications, they are basically doubling IE numbers in some countries. Now I know why NetApplications' IE market share is so out of wack with every other stat counter.

Voice of Buddy Christ said,
Close the thread. This is pretty much all that needs to be said about this.

As always just say something against microsoft to be the Messiah.

cork1958 said,
We'll do what ever it takes to make the numbers look better, dang it!!

If you actually read the article and think about it, Microsoft's preference is pretty reasonable.

rfirth said,

If you actually read the article and think about it, Microsoft's preference is pretty reasonable.

Absolutely true. "Pre-rendering" or "pre-fetching" a site without actually visiting the site is a meaningless metric with regards to site visits. It artificially inflates the stats based on the pre-fetch with no regards to the user actually browsing the site.

Just because Neowin is in my favorites doesn't mean I browse it every time I launch the browser and for a browser to cache the site, albeit a helpful thing for it to do, doesn't mean I looked at the site. And for a site to consider my browser went there even though I didn't is a false assumption and as such artificially inflates stats.

cork1958 said,
We'll do what ever it takes to make the numbers look better, dang it!!

Well of course it is crazy talk...

We should be counting all the automated scripts and bots, cause they are people too.

(This thinking is so messed up, it is almost scary. Automated scripts are not users and not people, and have NO relevance to what browsers PEOPLE are using.)

If I have a server doing HTML parsing using Webkit posing as Chrome to grab data from millions of sites, it is NOT page views and it is not 'user' views.

Wow...