Mobile App Stability
Outlook 2023

High standards for mobile app stability

Crash-free session rate: The percentage of mobile app sessions that did not end with a fatal error.

As the undisputed bane of mobile app stability, crashes are extremely disruptive to the user experience, terminating it altogether. The data reveals a median crash-free session rate of 99.94%, suggesting a high level of stability across all apps.

At the 75th percentile, the top-performing mobile apps score an impressive 99.99% crash-free rate, confirming that ‘five 9s’ stability is a valid target for mobile teams. On the other end of the spectrum, apps at the 25th percentile lag behind with a crash-free session rate of 99.76%.

This data makes it plain that mobile teams targeting anything lower than a 99.95% crash-free session rate should revisit their stability targets or risk being outperformed by the competition.

Android scores a surprise win

Surprisingly, our data set sees Android apps scoring a win over iOS in terms of stability with crash-free session rates of 99.95% and 99.87% respectively. This is true even among the top 25% performing apps on each platform at 99.99% and 99.96%.

Despite the high degree of fragmentation in the Android ecosystem, high levels of stability have proven to be quite achievable. Mobile teams should not accept lower stability from their Android apps and must give priority to their app stability to achieve good results.

Platform Breakdown

*Higher is better

A marked difference between industries

Breaking down our data by industry reveals a relatively high degree of stability overall, with some clear winners and losers.

Leading the pack, educational apps scored an impressive median of 100% crash-free sessions, further confirming the viability of the ‘five 9s’ stability target. Automotive apps came at a very close second with a 99.99% crash-free session rate.

Food and beverage services scored the lowest median crash-free session rate at 99.79%, followed closely by media and entertainment at 99.84% and advertising at 99.89%. The remaining industries registered a respectable median crash-free score between 99.92% and 99.94%.

Looking through a different lens, we can see that the leading apps across all industries maintain a commendable score, with four industries achieving an almost flawless score of 99.99% or above and only two scoring below 99.95% at the 75th percentile.

Industry Breakdown

*Higher is better

5-star ratings require 5-star stability

Arguably the most important indicator of a mobile app’s user experience, app store star ratings are a make-or-break factor for mobile app success.

App store ratings and reviews are the only user-facing indicator of app experience to potential users and the primary standard against which mobile apps are judged by app stores. Improving the quality and quantity of your app’s ratings and reviews allows it to rank higher in user searches, driving traffic to your app store page and boosting downloads and conversion rates.

Stability is not the only factor affecting app store rating, but our data shows a clear correlation between an app’s crash-free rate and its star rating, with apps under 3 stars scoring a significantly lower rate of 99.89%.

This relationship is especially evident when we consider the crash-free score of the least-performing apps, with apps rated between 3 and 4.5 stars scoring around 99.7% and ones above 4.5 stars scoring 99.84% at the 25th percentile.

This indicates a ’critical stability threshold’ of around 99.7% crash-free sessions to reach a 3-star rating, and around 99.85% to exceed 4.5 stars.

App Rating Breakdown

*Higher is better

Application Not Responding errors (ANR)

Definition: Instances where an app is unresponsive to user input for more than five seconds, prompting the operating system to notify the app user.

ANR errors are a serious and highly disruptive stability issue that severely impacts the mobile app experience. The median number of ANR errors that occur across all apps stands at 0.79 ANR errors for every 10K sessions demonstrating their low overall incidence rate.

However, breaking this down by rating reveals that apps rated below three stars scored a significantly higher median score of 4.46, with the rest registering scores between 0.24 and 1.02. This suggests a threshold of around 4 ANRs per 10K sessions above which a mobile app’s star rating might start to suffer.

ANRs per 10K Sessions

*Lower is better

Out Of Memory errors (OOM)

Definition: App terminations by the operating system due to excessive memory usage.

While not technically crashes, OOM errors have an equally fatal impact on the user experience. Not far from ANRs, the median number of OOM errors across all apps stands at 0.83 OOM errors for every 10K sessions, also registering a low overall incidence rate.

Breaking this down reveals widely varying OOM scores between 0 and 3.07, with no clear correlation between OOM scores and app store ratings. This could be explained by the low incidence rate of OOM errors overall which doesn’t affect a significant number of sessions.

However, considering their similarity to ANRs, it might be safe to assume that OOMs have a similar incidence threshold of around 4 OOMs per 10K sessions above which app store ratings are impacted. Mobile teams must ensure they are able to capture and debug OOM and ANR errors occurring in their apps, keeping an eye out for any spikes and maintaining them below this threshold.

OOMs per 10K Sessions

*Lower is better

App hangs

Definition: Instances where an app is unresponsive to user input for more than two seconds, but less than five seconds.

While not as disruptive as the stability issues previously mentioned, app hangs introduce significant friction to the mobile user experience and can be a leading indicator for the occurrence of ANR and OOM errors.

The median number of app hangs per 10 thousand sessions steadily increases with app ratings, from 76 per 10K sessions for apps below 3 stars to 132 per 10K sessions for apps above 4.5+, with app ratings seemingly unaffected by the incidence of app hangs.

However, when we look at the data for the 75th percentile, we can see that apps with a rating above 3 stars all hover at around 200 app hangs per 10K sessions, while apps rated below three stars scored 334 app hangs per 10K sessions, almost 70% higher.

This suggests that there is a threshold of about 200 app hangs per 10K sessions above which an app’s star rating could begin to suffer. Again, mobile teams should have the appropriate tools and processes to monitor and debug app hang occurrences to avoid negative app store feedback.

App Hangs per 10K Sessions

*Lower is better

Forced restarts

Definition: Instances where an app is manually forced to terminate, then restarted within 5 seconds.

Forced restarts don’t always indicate the presence of a stability issue, but they are almost always a sign of user frustration. ‘Turn it off and on again’ is the first reflex for mobile app users who face a misbehaving app, making forced restarts an important metric to monitor.

The median number of forced restarts across all apps stands at 134 per 10K sessions, though the breakdown by app store ratings does not reveal any clear trends or correlations. While difficult to provide an explanation or recommendation based on this data, it establishes a benchmark for forced restarts that mobile teams could find helpful.

While not a north-star metric, tracking the incidence of forced restarts allows mobile teams to watch for spikes that indicate underlying issues and analyze app sessions that experienced forced restarts to understand what caused them.

Forced Restarts per 10K Sessions

*Lower is better