iOS 12 adoption and performance – what it means for your business’s app

On September 17th, Apple released iOS 12. And while many innovative new features were announced, the very first feature listed in the release notes was “Performance.” Earlier this year, Apple was heavily criticized for throttling CPU speeds on mobile devices, which drastically affected their performance under low battery conditions. Although they fixed this issue months ago, the stigma is still there. Given the focus on performance in this release, I decided to investigate how quickly iPhone users are updating their devices, and if the OS performance improvements have actually resulted in faster web page load times.

Using RUM data from Akamai mPulse, I was able to analyze the performance and adoption of iOS12 from hundreds of millions of user experiences across thousands of websites. Here’s what I found.

Measuring the lifespan of iOS

Each web browser requesting a web page sends an identifier, known as a User-Agent string, that declares the browser and OS versions in use. These strings make it easy to measure the percentage of page views served to each iOS version.

Prior to the release of iOS 12, version 11 accounted for 92% of the iPhone market share while about 5% still used iOS 10. After the first day of the iOS 12 release, only ~4.5% of requests were served to devices running the new release. Over the next 2 weeks the iOS 12 traffic gradually increased to 28%, while the iOS 10 traffic remained at 5%. One month later, iOS 12 traffic increased to 59% and as of October 24th it reached 62%.

Screen Shot 2018-10-25 at 10.52.19 AM.png

While this seems like a very gradual upgrade rate, it’s not uncommon for OS updates to take longer than application updates. The reasons for this are usually intentional staggering of large scale software deployments combined with the manual effort required on the part of the user to initiate the update.

The graph below shows a similar rate of upgrades from the 2017 release cycle, with 6% of traffic from iOS 11 devices within one day of its release. Within two weeks it increased to 31%. One month later, it increased to 51% and then gradually rose to 80% over the next three months.

Screen Shot 2018-10-25 at 10.53.37 AM.png

The distribution of iOS versions can be used as a signal to understand the percentage of iOS devices capable of using the latest features. It can also help developers decide how long to support older versions of iOS. Given the slow upgrade rates, it makes sense for applications to continue supporting iOS 11 throughout the next year. Based on the percentage of users on each version of iOS at the time of the last two releases, the average lifespan of an iOS release is about two years long. Developers and web teams can use that as a useful guideline for determining their development and support lifecycles.

Performance of iOS 12

So how do the performance improvements of iOS 12 stack up based on real user experiences? Looking at median load times (measured by the onLoad metric of a webpage), we can see a 12% improvement over iOS 11. The iOS 10 devices are likely older iPhones and they are loading pages 45% slower compared to iOS 12.

Screen Shot 2018-10-25 at 10.52.41 AM.png

The graph above is based on medians, but looking at the cumulative distributions of the response times helps provide a better feel for what users are actually experiencing. A cumulative distribution graph like the one below illustrates the range of load times based on the percentage of users that experienced them. For example, 25% of iOS 12 users experienced load times of less than 1.2 seconds and 75% of users experienced load times of less than 4.4 seconds. In the graph below, iOS 12 consistently outperforms iOS 11, but the variance is larger at the higher response times. Unsurprisingly, iOS 10 is much slower.

Screen Shot 2018-10-25 at 10.36.44 AM.png

Identifying iPhone models

The above data shows there is a performance improvement from the new OS. However, how much of this is attributed to newer devices?

Grouping iPhone models based on similar physical characteristics such as screen size and pixel densities can help identify different iPhone models. By correlating that understanding of devices with data about the performance of each iOS version, we can learn more about how they impact each other. The following chart illustrates the relationship between screen sizes, pixel densities and models.

Screen Shot 2018-10-25 at 10.38.13 AM.png


Analyzing the data from September 2018 in this way shows the distribution of popular phone models. It seems that ~13% of iPhone users are using the X series, while 13% have a much older device (5S and earlier). There is also a 2-to-1 ratio of normal to plus sized devices, which shows how popular the larger screens have been with iPhone users. Developers need to make sure they are designing experiences to take advantage of those larger screens.

iPhone Device Distribution.png

After exploring how the new iOS version is being adopted, and how it performs, it is possible to go one step deeper and examine which iPhones models are benefiting the most from the iOS 12 performance optimizations.

The graph below breaks down the model groups identified above, and summarizes the performance ranges. Based on these results we can see that iPhone load times improved across the board for all devices thanks to iOS 12. We can also see the step pattern of improvements with each device.

Load Time Comparisons for iPhones Running iOS 11.png

Load Time Comparisons for iPhones Running iOS 12.png

What can we expect with the new iPhone models?

The latest iPhone versions have faster CPUs, more memory, and high pixel densities among other features. This usually translates into faster user experiences and richer applications. We can see from this data that response times improved significantly.

Despite the continuous performance improvements, it is important to keep in mind that advancements in technology often leaves older devices in the dust. We saw that with the 5% of traffic on iOS 10, as well as the 13% of traffic using the 5S and older devices. While the iPhone 6 and 7 devices are still very fast, how will that change as applications are built for the newer phones? And how will this affect the way that you design applications while the iPhone population is fragmented with a mix of new and old devices?

*** This is a Security Bloggers Network syndicated blog from The Akamai Blog authored by Paul Calvano. Read the original post at:

Cloud Workload Resilience PulseMeter

Step 1 of 8

How do you define cloud resiliency for cloud workloads? (Select 3)(Required)