Firebase Predictions make it possible to detect and create user segments that are the most likely to make (or not to make) a particular action in the application (for instance, make a purchase or watch a video). Firebase Predictions are based on machine learning: the system analyzes user behavior and predicts their further steps on the basis of the findings.

Predictions are available for iOS and Android apps. Predictions create dynamic user audiences that will complete (or will not complete) a certain event. These audiences can be used in Remote Config, Cloud Messaging, A/B testing.

By default, Firebase Predictions feature two main predictions: churn, a probability that the user will leave the app, and spend – a probability that the user will make a purchase (a conversion). Besides, it’s possible to make own predictions by choosing other analytics events.

Examples of using Firebase Predictions:

  • Analyzing user behavior and predicting their further steps. Making own predictions based on any analytics events;
  • Remote Config, Cloud Messaging and A/B testing integration, which makes it possible to beforehand identify user interactions and use these tools more efficiently. For instance, the Spend user segment can be offered new goods, while the Churn users can be shown an ad;
  • Increasing user retention. Users who can intentionally leave the app may be involved and retained before they actually leave;
  • Personalizing the app. For instance, users who better respond to blue color can be offered a blue interface, etc;
  • Building complicated PUSH-notifications campaigns for income increase and user retention.

If we drop Predictions details, on Firebase user’s end, it’s a clear user-friendly tool that will help increase the app’s income and improve its major indicators: retention and LTV.