November 2017 Summaries

Firebase introduced multi-database support for Realtime Database projects to simplify scaling through sharding without requiring separate Firebase projects. Available on the Blaze plan, the feature lets developers create multiple database instances within one project, access secondary instances using absolute URLs, and retain a shared Firebase Authentication session rather than relying on custom authentication. Applications can distribute data using strategies such as a master shard that maps data locations, separate databases by data type, or isolated databases for individual customers in multi-tenant services. Each instance maintains independent security rules and can be tested in the console, while Cloud Functions for Firebase can trigger events from a specified database instance or use the default instance when none is named. Existing applications must update their SDKs to supported versions before using the feature.

Firebase Cloud Messaging (FCM) is a free cross-platform service that delivers more than 400 billion messages daily, and its new RESTful HTTP v1 API is designed to make message delivery safer, simpler, and more adaptable. The API uses OAuth2 access tokens that expire after roughly an hour, reducing the risk associated with exposed credentials, while requiring no client-side changes for existing FCM applications. It also improves cross-platform messaging by allowing developers to combine shared notification content with Android-, iOS-, and web-specific settings in a single request, rather than creating separate messages for each platform. Its structured platform-specific payload blocks support current iOS, Android, and web capabilities and provide a flexible foundation for future platforms or protocols, positioning the HTTP v1 API as a more secure and extensible replacement for the legacy FCM API.

Firebase announced console improvements that incorporate strengths from Fabric following its integration with Firebase, aiming to help developers manage app development, stability, analytics, and growth more efficiently. The redesigned navigation groups products into Develop, Stability, Analytics, and Grow, while a new Project Home highlights 30-day crash-free user rate, 30-day crashes, daily active users, and monthly active users with trend graphs. The console also adds Fabric’s Latest Release dashboard under Analytics to provide a quick view of recent release performance and potential issues. Analytics has been reorganized into question-based cards focused on topics such as user engagement and retention rather than technical terminology, a design preferred by 90% of surveyed users. Realtime crash and active-user data is now available in the Analytics dashboard and Latest Release section, with plans to extend realtime reporting further across the console.

Firebase announced the beta launch of Crashlytics as its primary crash-reporting tool, integrating Fabric’s realtime stability monitoring capabilities into the Firebase console and recommending that existing Firebase Crash Reporting users upgrade. Crashlytics groups similar crashes, displays crash-free user rates, supports release filtering, and uses significance badges to identify issues tied to specific devices, operating systems, or jailbroken phones, helping developers prioritize fixes. Developers can add custom logs and key-value data to capture user actions and app state before crashes, while realtime dashboard updates and email alerts notify teams of new, regressing, or increasingly impactful problems. Crashlytics data can also trigger Cloud Functions for Firebase, enabling automated escalation workflows such as routing critical purchase-flow failures to engineers or Slack channels, and its stability metrics appear across the Firebase project overview, Analytics dashboard, and release pages. Firebase cited Doodle’s use of Crashlytics and Remote Config to improve retention and engagement, noted that Fabric Crashlytics users do not yet need to migrate, and said further integrations are planned.

Firebase introduced a beta A/B testing layer integrated with Remote Config and Firebase Notifications, enabling developers to test multiple app variations among randomly selected users and evaluate outcomes such as revenue, retention, crash reduction, or custom Analytics events. Developers can control audience size, experiment variants, and success metrics, while Firebase delivers versions, tracks user behavior, and identifies stronger results using Bayesian statistical models also used by Google Optimize. Habit-building app Fabulous used the feature to simplify its onboarding flow by testing the removal of several steps, finding that eliminating both a letter and a routine-commitment prompt increased onboarding completion by 7% without harming retention. The platform also supports testing notification wording to measure opens and downstream actions, and requires apps to use updated Remote Config and/or Firebase Cloud Messaging libraries.

Firebase Predictions launched in beta to bring Google machine learning capabilities to Firebase developers by creating daily updated user groups based on predicted behavior. Built-in groups identify users likely or unlikely to churn or make an in-app purchase within seven days, while adjustable tolerance settings balance audience size against the risk of false positives. Developers can target these segments through Firebase Cloud Messaging notifications or Remote Config, such as offering incentives to at-risk users or adapting a game’s difficulty for likely churners. The service also supports custom predictions based on app-specific conversion events, enabling teams to identify and engage users likely to complete—or not complete—particular actions. Firebase described the product as an evolving beta and invited feedback through its support channels.