Reduced App size by 42MB — Was it worth it?
--
Theory
App size used to be a problem since the beginning of the mobile apps era, and there is no compelling resolution on the horizon. The reason is simple — device storage is one of the most expensive features of the phone. Not that it’s expensive to create it, but because it’s convenient for manufacturers to make two variants of the phone and charge extra money for the extended storage. Because of these pricing options, people are still sticking to small device storage sizes, and therefore are more likely to halt the installation process when they realize the size is too large. Moreover, when the device storage is full, the user cannot install the app until he clears some space.
Another issue with the app size is the increased proportion of cellular data usage. The fear of running out of data can lead to decreased conversion rate on the Google Play or App Store page as well.
Last year there was exciting research by Sam Tolomei presented on Google Play’s blog. It revealed the negative impact of APK size based on a very reliable source — Google Plays’s internal data. Quoting from the article — “For every 6 MB increase to an APK’s size, we see a decrease in the install conversion rate of 1%”. It was also interesting to see the impact on conversion rate by country. As you can see on the figure below, increasing APK sizes was more influential in emerging markets (like India and Brazil), than in the case of other countries (like Japan and USA).
Practice
When Fitify was launched, the APK size was about 99MB. As the limit was 100MB, it was clear that the size can discourage potential users to install it. The source of the size problem was not the code, libraries or UI assets. About 85% of the size was caused by workout video assets which were the cornerstone of the user experience. There were two solutions to consider — render videos with a higher compression rate or postponing the download of videos to the onboarding of the app. We decided to shrink the quality of videos slightly, and the size was decreased from 99MB to 57MB.
The day was 8th of October 2018, and because the APK size was slightly increased again (from 57MB to 67MB) on the 8th of November, I decided to compare 28 days before and 28 days after the change so the result won’t be affected by the change in November. The metric I was comparing was a conversion rate — a proportion of installers from all store listing visitors. All this data is provided in the Google Play Console.
Finally, here is the result based on more than 170 thousand store listing visitors:
Reducing the app size led to a relative drop in a conversion rate of 0.38%. Not the (57/6MB)*1%=9.5% uplift “promised” by Google 🤔
Not giving up (yet). India was the most “size sensitive” country in the paper by Google Play. Therefore I measured the influence solely for this country. Surprisingly, the relative CVR improvement was even lower, and most importantly negative as well.
Giving up. Conclusion here:
Shrinking APK size just may not be for us, or at least for our vertical. It may not be for you as well (despite Google is telling something different). I don’t think Google would purposely manipulate the data as there’s no objective reason for that. However, from my experience, I can only recommend you to save money and time and invest it in your app store listing, keyword optimization or most importantly — the product 💪
Next time on my Medium:
- Our experience with Firebase Analytics.
- Free tools to spy on your competitor’s revenue
- Localized App Pricing
- Push Notification Marketing
- Firebase Growth Features
Thanks for reading. Leave a clap, leave a follow and here is my Twitter 👀
