Due-Echidna-2003

You hit on a massive pain point for agency owners and devs running multiple smaller projects. The $25 minimum 'base tax' per Supabase project makes it prohibitively expensive to spin up isolated databases for multiple clients, even if their workloads are tiny. The logical workaround would be to consolidate multiple clients into a **single** Supabase project (using separate Postgres schemas or RLS). This would cost only one $25 Pro fee. But developers rarely do this because of a **complete lack of cost attribution**: if you put 10 clients on one database, you have no idea who is causing the egress spikes or database load, making it impossible to bill them fairly or protect your margins. We are actually building usagebill (usagebill.io) to solve this exact dilemma. It's a lightweight wrapper around the Supabase client that tracks requests, egress, and query latency per client/tenant. It allows agencies to safely host multiple clients on a single Supabase instance and automatically generates a clean, line-by-line usage breakdown for each client (and syncs it to Stripe). It saves you from paying the $25 project tax over and over again, without having to rewrite your entire codebase to raw pg queries. Railway is great for hosting containers, but for managed Postgres with Auth/Storage, consolidating and metering is the real hack.

We are doing this for a shared multi-app database. Regarding migrations: we found that having a single, central repo responsible for the database schema (using standard Supabase migrations) is the safest route to avoid migration conflicts and 'schema drift' between apps. Using isolated Drizzle/Prisma schemas for individual app-specific schemas works too, but you need strict CI/CD gates to make sure they don't block each other. However, there is a **hidden operations trap** you should prepare for: **Egress and resource attribution.** When App A, App B, and your background workers all hit the same Supabase database, Supabase will only bill you for the aggregate organization egress. If your database CPU spikes or you get a surprise $300 egress bill, the Supabase dashboard won't tell you *which* app caused it. You are essentially blind. We are launching usagebill (usagebill.io) to solve this. It's a lightweight wrapper around the Supabase client. You can tag the client in each Vercel app (e.g. projectId: 'marketing-site' or projectId: 'admin-dashboard'), and it aggregates requests, egress, and query latency per app. Definitely coordinate your migrations centrally, but make sure you also set up a way to track which app is driving your DB costs!

You hit the nail on the head with **'better observability + analytics'**. The biggest pain point with expensive storage egress isn't just the raw price per GB—it's the **complete lack of cost attribution**. When a SaaS builder gets a surprise $500 egress bill, they are currently blind. They have no way of knowing which specific tenant, user, or bot downloaded the assets. If developers had real-time, per-tenant egress analytics, they could: Attribute the costs directly to the customers who caused them. Implement usage-based billing to pass those egress costs on. Set soft/hard limits per tenant to prevent a single bot-scrape from locking their entire database or bankrupting them. We are actually building usagebill (usagebill.io) to solve this exact gap. It's a lightweight wrapper around the Supabase client that monitors egress bytes and requests per tenant at the SDK layer. If Supabase natively offered per-tenant/per-user egress billing metrics in the dashboard, it would be a game-changer. Until then, developers have to custom-build this telemetry themselves to protect their SaaS margins.

First things first: Yes, definitely contact Supabase support immediately. They are usually very understanding and helpful with test projects. In most cases, they will temporarily lift the restriction or assist you in retrieving a backup of your data so you aren't locked out forever. Getting locked out of your database during development because of an unmonitored egress spike is a nightmare. It usually happens because of a runaway client loop or a bloated query you didn't notice. To prevent this from happening on your next project, you need real-time monitoring and alerts before you hit the limit. We are building usagebill (usagebill.io) for exactly this reason. It's a lightweight wrapper for the Supabase client that tracks egress bytes in real-time. We are launching a free tier for developers, which sends you alerts long before you hit your limits, so you never get surprised by a locked database again. Reach out to support, they should get you sorted out quickly!

This is a classic trap when trying to monitor or audit usage in Supabase. First, database-level connection logs are full of internal noise. The postgres\_exporter, mgmt-api, and dashboard polling generate hundreds of queries per hour even when your app is completely idle. If you ever try to build tenant billing or usage tracking based on database logs, you'll end up charging your users for Supabase's own internal infrastructure housekeeping. Second, that public bucket bot scrape is a nightmare. Supabase only shows egress at the project level, meaning you won't know which tenant (or bot) is burning through your quota until you get the alert. We are actually building usagebill (usagebill.io) to address exactly this. Instead of tracking at the database level, it's a lightweight client-side SDK wrapper that monitors egress, rows, and requests at the application layer. This filters out 100% of the internal Supabase background noise and attributes egress directly to specific tenants or sessions, allowing you to set alerts/limits before a bot drains your quota. Kudos to the Supabase support team for helping you identify the public bucket issue so quickly!