Well, it's that time again – time for the monthly update to Android's Platform Distribution Numbers. Each month, Google publishes the latest figures, letting developers know what versions of Android are currently dominating active devices.

This month, we're seeing a familiar pattern – Gingerbread is continuing its slow descent, hitting 39.8%, down from 44.2% this time last month. Meanwhile the latest and greatest – Jelly Bean – accounts for exactly 25% of the overall distribution, meaning it's finally hit one quarter of all tallied devices. That, for those interested, marks a nearly 9% jump from last month's 16.5% figure.


Ice Cream Sandwich is still strangely climbing, hitting 29.3% up from 28.6% in February. Honeycomb, predictably, is experiencing its own descent, down just 0.1%, while Froyo has been sliced by 3.6% and Éclair and Donut holding relatively stable with tiny drops from the last update.

All of this being said, it's important to note that it appears Google has changed its data collection method, counting only those users who actually visit the Google Play Store within the collection period.

Note: Beginning in April, 2013, these charts are now built using data collected from each device when the user visits the Google Play Store. Previously, the data was collected when the device simply checked-in to Google servers. We believe the new data more accurately reflects those users who are most engaged in the Android and Google Play ecosystem.

This method is more accurate for developers (the audience for whose consumption these numbers are intended), as they – theoretically – only count users actively accessing the Play Store, rather than including devices that automatically check in to Google servers for updates.

Those interested can also find screen density distributions, and the distribution of Open GL versions at the link below.

Source: Google Developer Dashboard

Liam Spradlin
Liam loves Android, design, user experience, and travel. He doesn't love ill-proportioned letter forms, advertisements made entirely of stock photography, and writing biographical snippets.

  • http://www.androidpolice.com/ Artem Russakovskii

    The new collection method is far better for developers, it's good that Google changed it. Remember, this is not a cop-out - the dashboard stats are for developers, not press.

    Before, any checkin to Google's servers would be recorded in the stats, which would count devices without Play Stores or users who don't go install or update apps. Developers use the dashboard to see what people who install apps use, not what the whole ecosystem contains. To them, it's a lot more accurate now.

    • sahilm

      Yes, I agree. I think this move is better, but I'm sure many Pro-Apple or anti-Android blogs will be quick to claim this is Google skewing the numbers.

      • http://twitter.com/ToysSamurai Toys Samurai

        True, but there's no denial that having 40% of its user base running a 2 year old OS isn't a good thing to the ecosystem. Granted, I doubt that even if Google has the ability to upgrade all those Gingerbread phones, those phones could run smoothly -- they probably couldn't run Jelly Bean smooth enough for the general public (die hard Android fans, on the other hand, are more forgiven.)

        • Brandon Southwell

          I don't necessarily see it as a bad thing either if it's not hurting the bigger ecosystem. Anyone using a gingerbread phone right now isn't losing sleep over the newest Android features.

    • Daniel

      Lets be honest, the play store should be restricted to 4.2.2 users only, seriously how can you be led astray from designs made by our lord and savior matias duarte?

      • http://cassidyjames.com Cassidy James


    • PhilNelwyn

      Yeah, the new way of collecting this data is more interesting for devs...
      Now it's devices that visit the Play Store that are counted...
      Prior to this, every device hitting Google servers were counted, even if they weren't consuming Play Store content, which is what devs want to know...


    • Brandon Southwell

      It makes sense. Developers, and maybe even to an extent, the Android development team itself, aren't going to cater to users who are not actually accessing central pieces of Android, like the Play Store.

    • Thomas’

      From the description provided i thought that Google was already using only data from Google Play.

      Funny enough, 3.1 was only used by not-really-used devices.

  • icyrock1

    Donut seems to be on it's last leg, so to speak. It had a good run, but it's time it let go.

    • Jay T

      Definitely. I think it's time the developer docs started recommending that you support back to 2.3, not 2.2.


    What I'm getting out of the changes is that 4.0+ devices probably still don't make up the majority of Android devices checking into Google's servers, but make up a majority of the devices that people use. I have a feeling that someone saw an opportunity to shrink the numbers of older devices for those who love to criticize the stats, as well as boost developer focus on 4.0+ features.

    I would argue that this is less accurate in terms of the devices that are out in the wild, but it is a better approximation of what developers should aim for - which seems to be the intention that both Liam and Artem have stated.

    Unfortunately, even with the old stat counting, mandatory CM-Stats, and a few other metrics that are available, there's no complete way of telling what devices support what and which versions are out in the wild, because you'll still leave out Kindle, Nook, and other users without a Google checkin or other metrics system. Even then, the capabilities of such devices are varied. The platform distribution stats are as close as we'll get to a measure of how Android is in the wild. Perhaps, because this decentralization is coupled with the openness of Android, this lack of a metric is a good thing - Amazon and B&N, etc. can use Android without checking in with Google.

  • IncCo

    Thats a Huge change from last month, +9% JB?? How did that happen all of a sudden ?

    • http://k3rnel.net Juan Rodriguez

      By shrinking other device version numbers, you can raise other version numbers. Remember, this is for devices that have actually entered the Play Store, not just logged into Gmail or Youtube from an Android Device. People who bought (Or still buy!) Android devices running 2.X but never once enter the Play Store aren't counted here.

      • IncCo

        I see.. i thought that these numbers didnt employ that way of counting since the press memo says that they will be starting using that new way of counting in the Beginning of April.

  • James Jun

    There's no reason to support Gingerbread and below at this point for developers. Go 4.0+ developers, its better, easy, and a lot more value for effort return.

  • Cris Aviance Igtiben

    How about a huge chunk of android users in China which as far as I know doesn't have play store?

    • asdf

      They won't factor in anyway.

      AFAIK, they can't ping Google's servers due to the Great Firewall of China.

      • Markoff

        Google Play Store works just fine inside mainland China, the problem with chinese phones is that producers just remove all google services and replace them with chinese alternatives which they consider better suitable for Chinese customers and Chinese also like pirated software available on tons of their available markets. I live in China and have chinese phone, i just had to install Google framework/services (~80MB package) and since then it works fine, although I also use Market unlocker, since chinese play store is not really appealing

    • http://twitter.com/navjotbatra Navjot

      they are not very important for people developing for the play store.

  • http://profiles.google.com/hephastus Kurleigh Martin

    The screen resolution data is where the real story is! tvdpi (N7) and xhdpi (N10) cumulatively are less than 2% of devices.

    • Matthew Fry

      Well... keep in mind that these devices aren't phones- To buy a tablet, you need a convincing reason to have it in addition to your (probably) Android phone.

  • http://twitter.com/Codexx Cody Curry

    Incredible how a minor shift in collection criteria results in a much more realistic and accurate sampling.

    All the people whining about slow adoption of new versions can take it down a notch. The newest versions jumped almost 10%, directly taking numbers away from Gingerbread. 4.0 devices (a version of Android that suffers from far less fragmentation overall) make up over 50% and are growing rapidly. That's a pretty significant step forward, even if it'll take a couple more months for Gingerbread to really die off.

    • http://my.opera.com/rafaelluik Rafael

      What? How do you think it's more realistic and accurate? People who don't access Google Play every week don't use their phones and apps?

      • http://profiles.google.com/bmg1001 Bryan Guerrero

        This chart is for developers to see what versions of android to develop for. This is a good change because only people who regularly download apps are counted. Why would a dev need to know that 2.5% more people use GB or something if they don't even download apps?

      • Freak4Dell

        It's pretty rare that you don't at least get an update to an app or something in a month long period. If you're using your phone as intended, there's a strong chance that you're accessing the Play Store at least once a month, and if you're not, then you really wouldn't care if developers developed for your phone or not. The change helps developers have a better idea of what they should be doing, and that's what matters.

        • http://my.opera.com/rafaelluik Rafael

          As I understand the thing is auto-updates (the default) doesn't count anymore, only if the person access the store intentionally it'll be counted by the new method. Hey I don't need a new app every month!

          • Freak4Dell

            I don't know. That's not how I read it. Maybe Liam or Artem can provide some insight.

  • Boris Iomdin

    "Ice Cream Sandwich is still strangely climbing..." - nothing strange. A lot of Chinese devices (from the "noname" to branded phones) still are manufactured with 4.0.4...

    • Markoff

      that's one reason, other reason would be that just old phones running older versions of Android are just being disposed so whole percentage of pool between newer Android versions is raised even if absolute number didn't grown

    • http://www.facebook.com/profile.php?id=523907787 Maxx Tan

      IF, you have been following the trend of China phones, they are one of the early adopters of the latest OS. They started injecting phones with 4.2.x even before any known manufacturers does. Google for android 4.2 china phones.

  • Dinofan01

    Are there any devices besides Nexus devices that are running 4.2 (not including custom ROMs)? If they're the only ones they make up less than 2% of the market (taking into account custom roms). That's slightly surprising to me. I didn't expect a huge number but I thought the Nexus 7 was popular enough to warrant at least 5% for all Nexus devices. Unless some Nexus owners haven't updated which I don't see as enough to jump a percent.

  • mldi

    Bigger news in my opinion is that 99.7% support GL2.0

  • http://www.facebook.com/Samurai89 Daniel Harris

    What about people with custom roms? Does it count those too?

    • akshay7394

      It does, because it looks at what OS you're running when you connect to the Play store to download something.

  • http://p.mobile9.com/abilashanmwg/ Justice Seeker

    GB, ICS & JB are high because people buy android phones and use as it is. They aren't aware about ROM updating.

  • http://twitter.com/Metshatrocker Metshatrocker

    why are people upset with the results. Either yall love complaining or out of touch, Developers need to update their old GB style apps. end of story

  • Bernardo

    accounts for exactly 25% of the overall distribution, meaning it's finally hit one quarter of all tallied devices. That, for those interested, marks a nearly 9% jump from last month's 16.5% figure."

    16.5% to 25% isn't a increase of 9%. It's a increase of around 57%!! If it were 9% would be around 18%. What you wanted to say is 9 percentage points, not 9%.

  • Bill Joyce

    On ICS numbers climbing. Maybe there are a few more HTC Thunderbolts out there than the tech community realizes, since regular people don't upgrade often. Just a thought. There have also been a lot of deals lately on refurbished tablets running ICS, but I suspect tablets are mostly showing up as Jelly Bean numbers.