A Reminder About Android Adoption Rates

And how long it’ll take Oreo to be eclipsed (sorry)

Version 8 of Android, which we now know is code-named Oreo, was officially released today through the Android Open Source Project, a year almost to the day after the release of Android Nougat. So it’s worth taking a minute to remember that a new Android release isn’t like the release of new versions of iOS (which we’ll see next month). In fact, it’s likely that no more than a third of Android’s installed base will ever use Oreo, and that it’ll take about two years for the majority of the base to be using Oreo or versions that come after it. (I wrote a similar analysis here last year on the day Nougat was released).

Every Version of Android Gets Lower Penetration in the First Year Than the Last One

The first thing to note here is that, for the last few versions of Android, each version has achieved lower penetration in the first year than the version before it, as shown in the chart below:

Jelly Bean achieved just under 40% penetration of the Android base at the end of its first year, KitKat achieved 30%, Lollipop 25% and 24% respectively, and we’re still waiting to see where Nougat will be after its first full year, but chances are it’ll top out at around 17–18%, the lowest by far. That backsliding has occurred despite Google’s repeated efforts to speed up the process of rolling out new versions, with its long developer betas, getting bits to OEMs earlier, and so on, over the last several years. Project Treble is the latest attempt at this, and we’ll have to see if that changes things.

One thing we should note here: these are percentages, and not absolute numbers: the overall Android base continues to grow, so it’s possible that Google is making progress in terms of raw numbers of devices, but it doesn’t provide total Android installed base data regularly enough for us to measure this.

The Most Popular Version Today is Two Years Old

The most widely-used version of Android today isn’t even the version released a year ago, but the Marshmallow version released in October 2015, nearly two years ago. And that just barely became the most widely-used version in June of this year — before that, it was Lollipop, released in November 2014:

At any given point in time, most Android users are using something other than the last two versions of the software. Compare that to iOS, where new versions become the most used within the first few weeks, and dominate the base within the first three months.

Features in New Versions Take Two Years to Reach the Base

For all these reasons, features launched as part of new versions of Android have in the recent past have taken nearly two years to reach even half the installed base. Marshmallow and Nougat combined haven’t yet reached 50% of the base as Oreo launches, so we don’t know how long that will take, but they’re at just under 46% today with Marshmallow at 22 months, so it’s likely that it’ll be around 2 years.

That’s arguably why Google has increasingly abstracted formerly OS-level features from the core operating system and put them at the app or services layer, which can be updated by users independent of OEM and carrier rollouts. But that’s still not a universal pattern, and Google’s blog post about its release today touts several features — autofill, new emojis, speed improvements – which are still OS-dependent and therefore won’t reach the median user until the second half of 2019.

When Will Oreo be Eclipsed?

Interestingly, despite the slow growth of new versions, they’re still often eclipsed by a subsequent version pretty early in their lifecycles, as new versions start to gain traction six to nine months into their launches:

KitKat and Lollipop both stopped growing as a percentage of the base at around 15–16 months in, as Lollipop and Marshmallow respectively began to gain traction. But Marshmallow is still growing at this point, and it’s likely that it will begin its decline in the next couple of months as Nougat starts to take off (a year after its launch). Oreo, meanwhile, won’t likely make a serious dent in the user base until early next year. And it will likely be the most-used version of Android sometime around late 2019, and eclipsed by Android Peppermint in 2020.

Timing is Everything

What’s interesting to me is that, even though Android’s annual release date has moved around quite a bit for the last several years, it still releases at a time of year that makes it almost impossible for a new version to make it onto new flagship devices from major vendors in a timely fashion. There are basically two windows in the year when these devices hit: spring, often announced at or around Mobile World Congress, and fall, in the case of Samsung’s Note line in particular. By releasing a new version in August, Google almost guarantees that the flagships released that fall won’t carry it. This year, we’ve already seen the Essential Phone launch with Nougat, and Samsung’s new Note will almost certainly launch with it too.

Meanwhile, perhaps the Pixel 2 models Google announces later in the fall will be the first phones to carry Oreo. At some point, you have to start to wonder whether the timing of these new Android releases is deliberately intended to give the Pixel an advantage and every other flagship a disadvantage.