Haavard

Same number of users, yet wildly different market share for Opera and Opera Mini?

Saturday, January 2, 2010 10:52:00 PM

stats, browser stats, net applications, hitslink, statcounter, chrome

Net Applications reports an increase in market share for both Opera Mini and Opera (desktop) in December 2009. What's odd is that Opera for desktop is reported as having a market share of 2.40%, while Opera Mini has a mere 0.53%.

You may recall that the desktop version was reported as having 45 million users early in December 2009, while Opera Mini had nearly 42 million users in November 2009. Presumably, that number grew somewhat in December, and I wouldn't be surprised if Opera Mini has closed the gap to the desktop version even more.

With these numbers in mind it's quite odd that the desktop version has a reported global market share which is several times higher than that of Opera Mini.

This is far from the first case of market share oddity. Around the time we reported 45 million users for the desktop version late last year, Google announced that Chrome had reached 40 million users. Despite Opera's higher number of users, Net Applications reported Chrome has having a market share which was nearly twice as high as Opera's.

StatCounter reported a market share for Chrome hich was more than two times higher than Opera in November and December 2009.

There must be something really odd about the way Net Applications, StatCounter and other statistics providers are handling browser statistics.

It's too bad that people are using such flawed statistics as a basis for browser related decisions.

A pretty graph as we enter the new yearJon steps down as CEO of Opera

Comments

endless lovepersianweblog # Sunday, January 3, 2010 12:05:58 AM

Daniel HendrycksDanielHendrycks # Sunday, January 3, 2010 1:03:05 AM

Originally posted by haavard:

StatCounter reported a market share for Chromw hich

*Typo*

Rafael Luikrafaelluik # Sunday, January 3, 2010 6:32:15 AM

I have a doubt.
Does the search / bookmarks partners pay less to you because of these "flawed statistics"?

J. Mauricewiz # Sunday, January 3, 2010 6:38:18 AM

It's because these stats are generated from webserver logs and pageviews, so they only represent the ratio of browsers that pages were served to, not "market share" as they claim.

beeblebro # Sunday, January 3, 2010 8:24:22 AM

I guess you get quite accurate Opera Mini stats from your server park, but how do you measure "active" Opera Desktop users?

Sami Serolaserola # Sunday, January 3, 2010 8:44:16 AM

"Market share" just makes statistics sound hot

However, at least StatCounter has about-page where the method is explained. Statcounter's 3 million websites is not very much. It may have statistical credibility but I still wonder on what kind of websites StatCounter is used and how does it affect on results?

HitsLink looks very odd because they seem to compare mobile Opera Mini browser with desktop browsers. Does this mean there are desktop and mobile Safari browsers bundled together? If Opera Mini has 0.53% share compared to desktop browsers, then I would say it's rather huge because the share of mobile users is still very small compared to desktop users.

What comes to mobile browsers, stats based on visits does not give stats for browsers faking their user agent. Therefore these statistics should always be seen only as giving a "very general estimate".

Tomas Thorsteinssontth # Sunday, January 3, 2010 11:20:10 AM

nrof_users != nrof_pageviews so I don't see what's not normal when approximately the same number of users leads to 20% of 'market share' (when this is defined in terms of page views as it seems to be). I use my O-mini a lot less than my desktop browser for one. Not that I don't agree that the statistics are generally flawed, but you're comparing apples and oranges here.

Purdi # Sunday, January 3, 2010 2:57:35 PM

Originally posted by rafaelluik:

I have a doubt.
Does the search / bookmarks partners pay less to you because of these "flawed statistics"?

What on earth are you talking about? If people decide not to support Opera because of bogus stats, THAT is a bad thing for Opera.

Purdi # Sunday, January 3, 2010 2:58:29 PM

Originally posted by tth:

nrof_users != nrof_pageviews so I don't see what's not normal when approximately the same number of users leads to 20% of 'market share' (when this is defined in terms of page views as it seems to be). I use my O-mini a lot less than my desktop browser for one. Not that I don't agree that the statistics are generally flawed, but you're comparing apples and oranges here.

How about educating yourself instead of spouting nonsense? Net Applications clearly state that they are counting unique visitors and not page views:

"The site unique visitor and referral information is summarized on a monthly, weekly, daily and hourly basis."

BTW, Opera Mini is the ONLY way a lot of people can use the web.

Charles SchlossChas4 # Sunday, January 3, 2010 4:49:33 PM

Daniel HendrycksDanielHendrycks # Sunday, January 3, 2010 6:35:46 PM

Originally posted by beeblebro:

but how do you measure "active" Opera Desktop users?

Users that check for updates. (Automatically (every threee hours) or manually)

Cutting Spoonhellspork # Tuesday, January 5, 2010 6:43:32 PM

Phone users tend to view much fewer pages than desktop users. Especially in the pages covered by that stat-counter. Simply put a real computer loads pages faster, and may be navigated more easily. The Opera/Chrome nonsense is old news, Chrome's behavior inflates its count and Opera's behavior deflates its count.

Charles SchlossChas4 # Tuesday, January 5, 2010 8:07:23 PM

But for a lot of People Opera Mini/Mobile is the only way they can get on the web in their area

Cutting Spoonhellspork # Wednesday, January 6, 2010 10:05:43 PM

Yes Charles, and most of the pages they visit in those countries are not on NetApplications' or StatCounter's lists. Not to mention that Opera Mini and Turbo traffic are routed through server parks. It is difficult to quantify how much this harms the perceived share, but we may assume that it is a statistically significant penalty.