Why my Supabase bill increases: database, storage, and transfer
Published on 2026-11-14 by frfrand
Learn to identify billable Supabase resources, distinguish normal consumption from leaks, and decide when to optimize or migrate.

When a Supabase invoice rises, the cause is not always PostgreSQL. Growth can come from database compute, stored files, egress traffic, backend operations, or additional services depending on your plan. To make decisions, first separate usage, included quota, and overages.
1. Create an invoice inventory
In the Supabase dashboard, review project and organization usage and billing. Compare two full periods and note each billable resource separately. Confirm current limits and prices on your plan's official page: do not extrapolate an old amount to 2026.
| Resource | Useful metric | Diagnostic question |
|---|---|---|
| PostgreSQL database | Size, CPU, connections, instance | Did the load grow or did the plan change? |
| Storage | GB stored and traffic | Were duplicates or large files uploaded? |
| Egress | Outbound GB | Are there repeated downloads or uncached assets? |
| Functions and API | Invocations and duration, if applicable | Is there a call loop? |
| Other services | Specific invoice line | Are there add-ons or forgotten projects? |
2. Distinguish storage from transfer
Saving 20 GB of images and serving them millions of times are different consumptions. A thumbnail strategy and CDN cache can reduce traffic even if the stored volume remains the same. Analyze MIME types, average weight, download frequency, and cache policies. If private URLs are signed, be careful not to turn sensitive resources into public ones to optimize costs.
3. Investigate PostgreSQL
Review slow queries, indexes, table and index sizes, long transactions, connections, and event retention. PostgreSQL should not be replaced by another technology before determining if an unindexed query or a repeated task is causing spikes.
To understand execution plans you can use EXPLAIN (ANALYZE, BUFFERS) on safe queries; remember that ANALYZE executes the query and can alter data if the statement is a write operation. Perform controlled tests and avoid running heavy commands during critical hours.
4. Set limits on unexpected consumption
Add metrics for bytes served, requests, and file size per user or company. Establish upload quotas, allowed formats, and compression where appropriate. Use budget alerts available in your plan and verify if they are informative or if they actually stop consumption.
5. Optimize, change plan, or migrate?
Optimize first if the problem is a query, duplicate file, or abusive usage. Compare other plans if you need predictable resources and specific support. Evaluate your own infrastructure when stable load, team capacity, and data portability justify it.
Do not confuse migrating the database with migrating the entire application: Supabase can include Auth, Storage, functions, RLS, and access tokens. Exiting requires inventorying dependencies.
You can expand on RLS and isolation in Supabase, how to migrate Lovable Cloud to Supabase and the ArduMaker database services.
6. How to read Supabase charges without mixing categories
The Supabase billing documentation distinguishes an organization subscription, compute resources per project, and variable consumptions shared at the organization level, according to the category. This means that creating another project can increase Compute cost even if its database is barely used.
In plans that include installments, the overage is billed after exceeding the applicable limits. Do not mix database GB with Storage GB or with egress: they are different counters. Supabase also distinguishes outgoing traffic served from cache and without cache, with potentially different rates.
In October 2026, their Compute documentation further explains that Compute Hours are not covered by the Spend Cap. Do not rely on a limit alert as a guarantee that all charges will be frozen. Check which concepts are covered in your organization.
To detect an increase, create this weekly table: date, active projects, instance size, disk GB, Storage GB, uncached egress, cached egress, and contracted add-ons. Cross-reference changes with deployments and campaigns and you will find the category that changes.
Specific references: Compute de Supabase, Egress de Supabase.
Frequently asked questions
### Does more users necessarily mean a more expensive database? Not linearly; it depends on queries, allocated resources, and the plan.
Is storing and downloading a file billed the same?
Not necessarily; they are metrics that must be inspected separately.
Is it worth migrating just for a spike?
Before deciding, determine if it was exceptional or represents sustained consumption.
Reference: Supabase Pricing and official documentation.