Browser Wars: Metrics Firms Battle Over Internet Explorer, Chrome
A pair of Web metrics firms that track browser share have traded sharp blows, calling into question how their rival measures usage, and which browser -- Microsoft's Internet Explorer (IE) or Google's Chrome -- is the planet's most popular.
The dispute over usage numbers isn't new: In March, Roger Capriotti, director of IE marketing, made his company's strongest case up to then for the data published by Net Applications, and dismissed the numbers from Irish metrics company StatCounter because of what he labeled omissions and flaws in the latter's methodology.
Capriotti posted his criticism of StatCounter just a day after that firm announced Chrome had passed IE for the first time in a single day's tracking.
U.S.-based Net Applications has long claimed that IE remains the world's most-used browser, while StatCounter said that Google edged IE for all of May to become No. 1.
Net Applications had IE with a 54.1 percent share in May, Firefox in second with 19.7 percent and Chrome slightly behind Mozilla's browser at 19.6 percent.
Meanwhile, StatCounter said Chrome was in the top spot with 32.4 percent, IE in second with 32.1 percent and Firefox third at 25.6 percent.
The clash over methodologies ignored Firefox and concentrated on IE and Chrome.
Two weeks ago, Vince Vizaccarro, who heads Net Applications' marketing, emailed Computerworld with a rebuttal to StatCounter's claims that Chrome had replaced IE as the most-used browser.
"IE is still the leader, and it's not even close," said Vizaccarro, citing Net Applications' numbers for May. "[Chrome is] very close to passing Firefox for the number 2 position. But they are not threatening the number 1 position held by IE, and it doesn't look like they will for the foreseeable future."
Vizaccarro wasn't kidding when he said Chrome and Firefox were "very close" last month. Net Applications' initial data, released early on June 1, had Chrome passing Firefox in May. Later that day, however, the company revised its figures and said Firefox had held onto the No. 2 spot.
As part of the June 1 revisions, Net Applications also changed IE's share, boosting it by half a percentage point from the preliminary 53.6 percent to the final 54.1 percent for May.
Vizaccarro went into detail about why his company believes its numbers are different from its rival's. His explanations were largely repetitions of the arguments that Capriotti made in March, that Net Applications tracks unique users rather than page views -- so that each user is counted only once per day per website, rather than give more-active surfers greater influence -- and that it weights the data by country.
Microsoft has long cited Net Applications' data to tout the success of IE, particularly IE9 running on Windows.
Unlike Net Applications, StatCounter tallies page views, not users, and does not weight those results.
Both factors are responsible for producing the wildly-different numbers, said Vizaccarro, who rejected page views as susceptible to automated bots "designed to influence market share." None of the browser vendors, however, have ever been accused of trying to game the system with such bots.
Country weighting gives a more accurate estimate of browser share, said Vizaccarro. Net Applications weights its data to account for the lack of Western insight into browsing habits in nations like China, where IE is the overwhelming favorite.
"If our traffic were concentrated in one or more regions, our global data would be inappropriately affected by those regions," said Vizaccarro. "Country-level weighting removes any bias by region when you calculate a worldwide number like StatCounter has done. StatCounter's global data shows a regional bias toward the countries where Chrome is used more."
StatCounter didn't take the criticism lying down. In a long blog post Monday -- titled "An Open Letter to Roger Capriotti, Microsoft," the Irish company rebutted what Microsoft and Net Applications said were its failings.
StatCounter said page views, not unique visitors, were a more accurate measurement of browser use
"Page views is the only valid metric to look at when talking about browser usage as it directly measures how much activity (or usage) is happening on each browser," said StatCounter. "It is just plain wrong to claim that 'browser usage' is measured using unique visitors."
The company also parried the issue of weighting, saying that the CIA's data on the number of online users in each country, which Net Applications uses to modify its data, was outdated -- the most recent from the intelligence agency is 2009's -- and so doesn't take into account rapid increases in activity in Brazil and India, where Chrome is the leader.
But the biggest factor in its favor, said StatCounter, was its larger sample size: 3 million websites compared to Net Applications' 40,000, and 15 billion page views each month to Net Applications' 160 million unique visitors
StatCounter also called Microsoft's affection for Net Applications' data hypocritical, since while much of the boost IE gets in the latter's data comes from China, a considerable amount of that originates from IE6 and from machines running pirated copies of Windows.
Microsoft has been trying to kill the 11-year-old IE6 since mid-2009, and has taken a strong anti-piracy stance in China, where it acknowledges counterfeit-use rate is high.
"It seems odd to rely on China to boost your global browser usage figures ... [but] usage is usage, huh?" said StatCounter.
To show the disparity between its data and Net Applications', StatCounter highlighted a five-month stretch between December 2011 and April 2012. During that period, Net Applications tracked a rebound by IE to the tune of 2.2 percentage points, from 51.9 percent to 54.1 percent. In the same five months, Chrome fell close to three-tenths of a point, from 19.1 percent to 18.8 percent.
StatCounter, however, saw a different picture in the same timeframe: IE fell by 4.6 percentage points, from 38.7 percent to 34.1 percent, while Chrome increased by about 4 points, growing from 27.3 percent to 31.2 percent.
"We're sorry to have to break it to you [Roger Capriotti] that IE actually trended down and not up over the period December 2011 to April 2012," said StatCounter. "Don't shoot the messenger!"
Microsoft said its opinions of Net Applications' and StatCounter's data had not changed.
"We stand by our assessment that Net Applications provides a more accurate analysis of the browser landscape," a Microsoft spokeswoman said in an email late Wednesday. "[But] it also unfortunate that StatCounter decided to respond to our critique of their data by staging a personal attack on one of our employees. We believe in a healthy dialogue and personal attacks in a professional environment are unwarranted."
Microsoft is a Net Applications customer, and like many other firms, pays the analytic company for access to browser usage data that Net Applications does not offer the public or the media.
Gregg Keizer covers Microsoft, security issues, Apple, Web browsers and general technology breaking news for Computerworld. Follow Gregg on Twitter at @gkeizer, or subscribe to Gregg's RSS feed . His e-mail address is firstname.lastname@example.org.