Boost your app reviews with Firebase Predictions!
Are you wondering when to show ‘rate app’ dialog? After 10th app opening? After using all app functions? Both? How about letting AI decide on that — now with Firebase Predictions it is super easy and efficient!
What is Firebase Predictions?
It is one of nifty features of Firebase — lately introduced at Firebase Summit, it allows you to predict if particular user is going to meet some event in the future (ex. next 7 days) — recalculated every 24h for each user. It uses events from Google Firebase Analytics to do so, so no additional work is needed. Pretty neat!
App rating is the king
With all ASO strategies to rank higher in the search results of Google Play/App Store ratings are crucial — mostly because they are visible on search result page. Also with a direct competitor it can be a deciding factor between apps. Having high rating is even more important than having higher number of reviews all together. Our goal is to ask at moment when user is not busy and as well is aware of all benefits of our app. Card mixed inside app content is much better than dialog itself.
Use Firebase Prediction to… well… predict
My new criteria to show rating dialog would look like this:
- 10+ app open
- Wifi connection (probably at home/work)
- Firebase Prediction is true for clicking “Rate”
or (as you can define only one predicted value in Remote Config)
- Firebase Prediction is false for churn in next 7 days
Previous were much more complicated and less ‘aggressive’ (20+ app opening, using at least 3 features…).
Results… well —are way better than expected!
Ok, but maybe it is because it is more ‘aggressive’ (showing dialog earlier for more users) as less criteria has to be met? Actually yes — this is the case, my click ratio of ‘Rate’ jumped from 3,21% to 4% so more users click ‘Rate’ than before(good!) but at same time drastically more users are clicking ‘Cancel’ or ‘Later’ (1,6% -> 7% and 5,3% -> 13%) — those number doesn’t add to 100% as ‘rate dialog’ can be showed multiple times to one user, until she/he clicks or other criteria is no longer met (ex. churn in next 7 days is true).
Also with that statistic I know that 63% of clicking “rate” ends with leaving app review in Google Play. In your case results might be different as probably it depends on app category and also on platform — my analysis is based on Android app Canary — air pollution monitoring app, currently ranked as second popular world-wide in Google Play.
Conclusion
Firebase Predictions is great tool to simplify such problems, also in my case being a bit more aggressive with showing popup really paid off. What is more important now we take into account not only existing values as criteria but also future predictions of them so if somebody is probably going to uninstall our app soon she/he won’t get that notification — great!
With all above you should keep in mind things that you are not allowed to do to get positive user rating.
What’s next?
Firebase Predictions is still in the Beta stage, and can be used to many more features in your app. From my point — Webhooks/HTTP API is missing so you would be able to trigger some backend action if users is going to meet some prediction.
Also with Neural Networks API in Android 8.1 you can start developing your own and local AI models to solve such problems!
