Haavard

Net Applications admits to skewed statistics

Wednesday, December 3, 2008 12:56:55 PM

browser stats, opera, net applications, stats, hitslink

In an article at "The Industry Standard", Net Applications apparently admit that their numbers are skewed. More specifically, they admit that they are skewed towards certain regions of the world (US-centric, anyone?). In such an amazing moment of honesty, I wonder why they didn't also admit to actively editing their own statistics.

Update: Their claim to be skewed towards Europe doesn't really match XiTi Monitor's statistics for Europe. This would indicate that Net Applications is heavily skewed towards the US.

Google closes the Web in korea?Net Applications again: Opera vs. Chrome - actual numbers vs. claimed market share

Comments

Anonymous # Wednesday, December 3, 2008 2:25:06 PM

Anonymous writes: i'd recommnd reading the article all over again if you are negatively surprised that NA tracks only certain websites (these with their codes included) you really should sit and read some more on how does webtraffic tracking works. reasons stated by NA are valid - more browsing from home, less from office - more home-installed visits, less corporate-born visits = more alternative browsers (safari, ff, surprisingly not opera..), less corporate browsers (ie, ie, ie..). what is 'evil' about that? maybe this that opera didnt benefit from those events? and no, they are not us-centric, lots of pages in europe also have their codes, your included if you happen to know how to measure this data better, do it, youll be VERY wealthy, because A LOT of companies seek accurate and reliable data about market shares, and still yet nobody has managed to create one. as for now, I think that you are oversensitive about Net Applications as a whole, while you might be right in some cases, in this one it is very wide miss.

Haavardhaavard # Wednesday, December 3, 2008 3:59:18 PM

I am not surprised that they only track their own customers. That has been obvious all along.

What does surprise me is that they actually go a small way towards admitting that their statistics are not really representative of anything. But they don't go all the way: The numbers do not even represent their own customers, since they actively manipulate the data.

And yet Net Applications is widely quoted as an authority on global market share.

I don't know how to better measure global browser statistics. Currently, it is not possible to get statistics for a representative sample of the global internet population. Not even within a more limited region can you get a representative sample.

So rather than pretending that everything is fine and dandy in statisitcs land, they should be more honest about what their numbers represent, and both journalists, bloggers and other people should take the time to consider what the numbers represent.

I'm sure his other comments about home vs. office usage are perfectly valid, but they do not negate my overall criticism of both Net Applications and the people who don't check their sources, and report the numbers from Net Applications without even bothering to consider how representative they are of the global Web.

Khaled KhalilKhaled-Khalil # Thursday, December 4, 2008 12:30:02 PM

ok, now an African citizen like me have a good reason to no more trust them, they don't care even about me, if their main goods is confidence, they don't have a lot to serve me

Anonymous # Monday, December 8, 2008 10:10:31 AM

Anonymous writes: If NetApplications was over-representing Opera by a significant margin, I imagine that you would be cheering them and announcing every new NetApplications report on your blog with great enthusiasm. It sounds more like bitter grapes to me than a real dispute with the reports.

Anonymous # Monday, December 8, 2008 10:41:21 AM

Anonymous writes: So wait, the last idiot is saying that criticism is invalid because.... yes, because what exactly? This idiot has not disputed any facts, he merely spewed out garbage about "bitter grapes". If you can't argue with the facts, spew out nonsense! Look over there! Pay no attention to the facts!

Khaled KhalilKhaled-Khalil # Tuesday, December 9, 2008 6:24:10 PM

hey, please there is no idiots here, we can differ in opinion with respect in each other.
@Anonymous of 12:10:31, the matter is not about cheerful news or bad news, it is about real stats or fake ones (human edited/focused toward specific users or community/estimated instead of counted and calculated/any other way that make stats not dependable), it is about fidelity not loyalty.
btw, i don't work for Opera, don't expect me to be biased toward it.

Charles SchlossChas4 # Friday, December 12, 2008 1:45:40 AM

And then there is Opera having to mask because of back web coding

Haavardhaavard # Friday, December 12, 2008 12:42:22 PM

I've made an update showing XiTi Monitor's statistics for Europe. They are completely different from Net Applications, indicating that Net Applications is overwhelmingly US-centric.

Anonymous # Friday, December 12, 2008 8:12:05 PM

Anonymous writes: haavard, another way to write that is that XiTi Monitor is overwhelmingly Euro-centric. :-) I'd put forward that XiTi is even worse than Euro-centric, their sampling is overwhelmingly biased towards French users and the further away from France you go in their data, the less accurate it's likely to be. So, XiTi Monitor is a decent measure if you look at their France only data. They're a poor measure if you look at their European-wide data. And they don't have any relevance at all outside of Europe. Now, compare that with NA which has strong US sampling. Decent European and South American sampling and poor Asian sampling. Of course all these are biased but you have to look for the data sources that are least biased in the ways that matter. If you want the most accurate view of the world, NA is it and XiTi isn't. If you want the most accurate view of France, XiTi is it, and NA probably isn't. None of them are perfect, but they're all useful.

Anonymous # Saturday, December 13, 2008 10:07:09 AM

Anonymous writes: The last clueless anonymous doesn't get it. XiTi's stats ARE for Europe. They are not claimed to be global. That XiTi is crappy like all statistics services claiming to know browser market share is probably true though, LOL. But he is equally clueless when he claims that NA gives a more accurate view of the world, considering that they claim that Chrome has more users than Opera, when Opera actually has THREE TIMES as many users as Chrome. Did the clueless anonymous guy read the other blog post about this? These statistics aren't even useful. NA actively edit their stats, and are clearly completely wrong considering their claim that Chrome has a higher market share than Opera when Google reports 10 million users for Chrome and Opera reports 30 million users.

Anonymous # Sunday, December 14, 2008 12:01:01 PM

Anonymous writes: You do not understand that there is a difference between users and usage? Opera could have three times as many users as Chrome and if the Chrome users visited more sites for more hours, they could have more usage than Opera. You are also trusting the companies, Opera and Google to tell the truth about the number of users they have? More power users are on Chrome than Opera. See sites like TechCrunch have 3x more visits from Chrome users than Opera users. See this: http://www.techcrunch.com/2008/12/11/chrome-shines-a-little-brighter-drops-the-beta-tag-with-new-release/ If power users are moving to Chrome away from Opera, that could hurt Opera a lot. Power users have more usage than regular users. Power users have more influence with NetApplications because that measure is usage and not users.

Anonymous # Sunday, December 14, 2008 12:24:34 PM

Anonymous writes: Opera has to tell the truth because it would be illegal to mislead their shareholders. Your user/usage excuse is pathetic to say the least. Opera's caching makes sure that it's downloads resources rarely compared to other browsers, and Opera's useragent string masking ensure an artificially low "usage share". And Chrome users don't browse more than three times as much as Opera users. The TechChrunch numbers are useless. One site where Opera is claimed to make up a small amount of the usage, which is just silly when you think about the caching again. Power users are using Opera and Firefox since Chrome is stripped of features, and you can't add anything to it. And Net Applications is still a bunch of liars who edit their own stats and have been busted doing so several times.

Khaled KhalilKhaled-Khalil # Sunday, December 14, 2008 6:22:41 PM