PercentMobile on, “Mobile OS web-browsing share” by @marcoarment

reblogged from: marco

Some people are criticizing John Gruber’s piece on iPad and Android browser share because Apple-product owners are more likely to visit his site (a bias he clearly acknowledged). I was curious to see more widespread numbers, so I got permission to post Tumblr’s OS percentages from Google Analytics for the tumblelog network.

This includes most human visits to all Tumblr-hosted blogs, not the tumblr.com site itself, to best represent “average” people online who happen to come across Tumblr-hosted sites, not just Tumblr members. Granted, this still isn’t perfect, but it’s probably the biggest and least biased sample that we’ll be able to find in the indie-Mac-pundit world.

        
Left: Including “normal” computers. Right: Only mobile devices.
Sample from May 9-15, 2010, as measured by Google Analytics.

The most surprising part of this, to me, is how well the Macintosh is faring against Windows. But in the mobile space, Android is actually doing quite well, given its tiny installed base relative to iPhone OS. My premise in this post may have been completely wrong.

The iPad is putting up an especially impressive performance given that it’s only available in the U.S. so far, has only been on sale for 6 weeks, costs at least $500, isn’t subsidized, isn’t always in your pocket, and isn’t being given away in two-for-$99 sales by the largest cellular provider in the country.

PercentMobile:

@marco (and Tumblr), thanks for the breath of fresh air.

Most of the numbers we see online are from mobile analytic services that primarily report on people visiting apps or ads on iPhone, Android, or Blackberry devices. Traffic from the rest of the mobile ecosystem is usually underrepresented and as such marginalized.

The data from the tumblelog network can certainly provide a more realistic sample.

Unfortunately, improving the sample is only part of the equation. The other part is correctly collecting and processing that tumblelog data. Google Analytics is not mobile-specific enough to get the job done.

The report you share shows that Smartphone traffic to tumblelog sites represents approximately 93% of tumblelog mobile device traffic. Trends across PercentMobile’s network of sites tells us that the Smartphone crowd should more likely weigh in at around 47%-68% of total mobile traffic. As such, we suspect considerable Feature Phone (and lesser device traffic) is present but not counted. What does that mean to Tumblr? Hard to place a number on it since we’re at arm’s length to your data but you may have up to 2X the mobile traffic you think you have.