Part of the entertainment of watching browser market share statistics is the interpretation of data, the consideration of validity as well as the reaction of browser makers.
Obviously, statistics carry marketing value in a negative or positive view and there is a clear motivation to comment on numbers that are primarily published by NetApplications and StatCounter (even if there are plenty more sources such as Clicky, StatOwl, or W3 Browser Statistics).
While Mozilla and Google have been largely quiet about market shares (but certainly care, as mentioned by Mozilla CTO in this blog post), Microsoft has used market share numbers only from Net Applications to highlight the decrease of importance of IE6 and the runaway success of IE9, especially on the dedicated blog Exploring IE. Last week, Microsoft reacted to data published by StatCounter that Chrome has exceeded IE market share for the first time. Microsoft's opinion is that StatCounter's data are slanted and incorrect and best. The reason? StatCounter evenly counts market share across its user base and does not consider geoweighting. To make matters worse, StatCounter also considers Google's prerendering feature, which obviously puts a positive spin on Google's data.
Without going too deep into analytics, Microsoft complains that the pages that are pre-rendered in Google's Omnibox count for market share at StatCounter. In February, pre-rendering counted for 4.3 percent of market share, Microsoft says. Also, Microsoft believes that, depending on the available data set, browser market share data has to be weighed against the entire Internet population. For example, China (where Microsoft dominates the browser market share arena) has 21.39 percent of the Internet population, according to CIA Internet user statistics, but it has only 0.96 percent in StatCounter's data.
According to Internet World Stats, the CIA estimate may be a bit conservative, as the user share in China could actually approach 25 percent. So, does Microsoft has a point? Sure. But we should not forget that NetApplications has also its problems as the scientific process of data collection isn't entirely transparent with missing margins of error, for example. My personal gripe with NetApplications has been that the freely available data change on a frequent basis and the company has denied consistent access to the same data sets over time. For example, Net Applications has strongly limited access to market share data to the fragmented versions of IE. Were they removed because of monetization reasons? Because of integration into IE overall? We have no idea as the company declined to comment. I cannot help but have some doubt about the data distribution between browser versions and within a browser itself. I am not accusing NetApplications of shady data publications, but there are clearly inconsistencies that make it difficult from the outside to assess the value of the data.
Like NetApplications, StatCounter has its issues as well. It is common sense to assume that no market share estimate can be entirely correct and there will always be opportunities to criticize the data published and interpret the data in the way you prefer. Those on the outside have to remember that browser market share statistics are merely trend indicators. In any case, absolute numbers should always be taken with a grain of salt.