Why developers leave Firebase: Firestore data model can't be normalized (you live with denormalization forever), pricing on reads and bandwidth surprises growing teams, and migration off is genuinely painful. The 2024+ rise of Postgres-native BaaS provides credible alternatives with relational data models.
Supabase is the Postgres-native answer to Firebase — Auth, Storage, Realtime, Edge Functions in one platform. Standard Postgres means joins, foreign keys, RLS policies. Self-hostable, open-source. Mobile SDKs less mature than Firebase's but growing. For full-stack web apps Supabase has caught up and pulled ahead on data-model fit.
Convex is TypeScript-end-to-end backend with reactive queries. Type inference flows from schema to React hooks. Built-in scheduler/cron. The trade-off vs Firebase: TypeScript-only (no mobile SDK), proprietary execution model. For TS-first teams the DX is the best in the BaaS category.
Neon is Postgres-only (no auth/storage/realtime bundled) — pair with Clerk for auth and S3 for storage. Scale-to-zero economics make preview-environment databases nearly free. For teams that prefer best-of-breed services over BaaS bundling, Neon + Clerk + R2 is a serious alternative architecture.
PlanetScale is MySQL-via-Vitess. Schema-change deploy requests prevent prod migration disasters. Branching for migrations. Trade-off: MySQL-only, no foreign keys (Vitess limitation), no scale-to-zero. For teams already on MySQL at scale the schema-change safety is the moat.
Most reviewers report 6-12 weeks for a non-trivial migration. The data shape rewrite (denormalized Firestore → relational Postgres) is the hard part. Auth migration is mechanical. SDK swap is straightforward. Plan for a sprint per major collection.
Firebase's iOS, Android, and Flutter SDKs are still platform standards. For mobile-primary products with realtime sync and offline requirements, Firebase remains the pragmatic choice. The data-model trade-offs are real but mobile SDK maturity offsets them for many teams.
Yes — many teams keep Firestore for live mobile sync while moving server-rendered web and analytics queries to Postgres. The dual-write pattern adds complexity but lets you migrate critical paths first. Plan for the dual-write code being temporary; commit to a final migration date.
