Worldwide Desktop & Tablet Browser Statistics, February to March 2015
The following table shows browser usage movements during the past month.
Worldwide Desktop & Tablet Browser Statistics, March 2014 to March 2015
The following table shows browser usage movements during the past twelve months:
|Browser||March 2014||March 2015||change||relative|
(The tables show market share estimates for desktop browsers. The ‘change’ column is the absolute increase or decrease in market share. The ‘relative’ column indicates the proportional change, i.e. 8.8% of IE9 users switched browsers last month. There are several caveats so I recommend you read How Browser Market Share is Calculated.)
There’s no significant news. All browsers gained a little at the expense of Internet Explorer. Perhaps that’s not surprising given Microsoft has publicly announced IE’s demise and Spartan will be the future. IE is within a point of dropping below Firefox — a prediction I made earlier this year.
Opera made another small step forward but I suspect Vivali could cannibalize some of the user base once a stable release is available.
Worldwide Mobile Browser Statistics, February to March 2015
Mobile usage regained some lost ground and now stands at 33.44% of all web activity. It’s the first rise in 2015 but won’t be the last.
The top mobile browsing applications:
There were few surprises other than a 1% jump for UC Browser which is popular in China. Chrome was the only other winner.
StatCounter vs NetMarketShare
Last month, rkellogg commented:
I’m continually surprised at the extreme differences in browser statistics from StatCounter and NetMarketShare.
|Browser usage March 2015||StatCounter||NetMarketShare|
Which is correct?
The answer: neither. It’s impossible to accurately gather browser statistics. The results are based on an unstable tower of assumptions. Changing a single variable or the data collection method can skew the numbers.
That’s a cop out! Which is more accurate?
The answer: it depends what you’re looking for. The companies use different data collection and analysis techniques…
|records data from 3 million websites||analyzes data from 40 thousand websites|
|analyzes 17.5 billion page views per month||analyzes 160 million unique visits per month|
|make no adjustments or weightings other than removing bot activity and pre-rendering||applies country weightings according to CIA Internet Traffic data. For example, if Brazil accounts for 4% of global traffic but only 2% of NetMarketShare data, the results are doubled|
|is independent and has no commercial partners||is a commercial company with partners paying for data|
|reports are based on page views||reports are based on daily unique visitors|
The key difference is StatCounter records visits while NetMarketShare records users. Consider the following scenarios:
- You visit 1,000 pages using Chrome then a single page with Internet Explorer.
StatCounter records 1,000 “votes” for Chrome and one for IE. NetMarketShare scores one vote for both.
- You use Firefox all day visiting 500 pages. Your Gran visits one site using Safari.
StatCounter records 500 votes for Firefox and one for Safari. NetMarketShare scores one vote for both.
- You use Chrome, IE, Firefox, Safari and Opera during the day.
Your usage is weighted on StatCounter, e.g. you used Opera twice as much as Chrome. NetMarketShare will rank each browser equally.
A vast number of people are light web users who access via their default browser (probably IE). A small minority are power users who visit thousands of pages daily with their preferred application (probably Chrome). NetMarketShare records people; both types of user are treated equally. StatCounter records usage patterns; a heavy web user will sway results more than others. Therefore:
- if you pick a random page visit on a random website, according to StatCounter there’s a 49.07% probability it will be from Chrome.
- if you pick a random Internet user from a random country, according to NetMarketShare there’s a 24.99% probability they’ll be a regular Chrome user.
Both are “right” even though the figures appear incompatible.
So why do you use StatCounter for Browser Trends articles?
StatCounter has a large sample set, simpler methodology, does not manipulate the figures and reports browsing activity. This is an important factor for web developers.
Reducing to a simple example: presume your application had two users — one who accessed continually with BrowserX and another who visited once per month with BrowserY. Your development and testing tasks should prioritize BrowserX first. The situation becomes more complex as traffic increases but page visits will remain a better indicator of browser usage than unique users or sessions.
Does any of this matter?
The reports are useful for spotting trends such as Chrome’s meteoric rise, IE’s downfall and the prevalence of smartphones and tablets. The information can help you plan future development tasks. However, your own website statistics are more indicative of real usage than average worldwide figures.
Craig is a freelance UK web consultant who built his first page for IE2.0 in 1995. Since that time he's been advocating standards, accessibility, and best-practice HTML5 techniques. He's created enterprise specifications, websites and online applications for companies and organisations including the UK Parliament, the European Parliament, the Department of Energy & Climate Change, Microsoft, and more. He's written more than 1,000 articles for SitePoint and you can find him @craigbuckler.