Does the Supabase free plan include backups?

No. Automated backups start on the Pro plan ($25/month, as of August 2026) — daily, kept for 7 days. On the Free plan there is no backup running for you: drop a table tonight and there is no restore button waiting anywhere in the dashboard. If you do not export the data yourself, no usable copy of it exists anywhere. The official backup docs say so plainly, recommending that free-tier projects regularly export their data with the CLI and keep off-site copies. This page is the practical version of that advice: what Free actually includes, the pause clock that makes it urgent, and three ways to get a real backup — two of them free.

What each plan actually gives you

Prices as of August 2026. One more asterisk that surprises people on every plan: even where automated backups exist, they cover the database only — Storage files are never in them. And on newer projects the platform’s own backups cannot be downloaded — they restore in place or not at all.

The pause clock makes this urgent, not optional

Free projects are paused after 1 week of inactivity, and you get at most 2 active projects. Pausing itself takes a snapshot, and that snapshot is what lets you restore a paused project from the dashboard — for up to a year. Note what it protects you from: the pause, not yourself. It appears when the project winds down, not when a migration goes wrong. Past the year, in-place restore is off the table; the backup and Storage files remain downloadable for migration to a new project, but you are now recovering on a schedule Supabase picked, not one you did. The free tier is where side projects, prototypes that quietly became real, and the weekend tool a client depends on all live. Nobody is on call for those. That is the argument for a backup that runs without you, not against one.

Option 1 — take a dump right now (two minutes)

Works on every plan, costs nothing, and future-you gets a copy that restores into any Postgres with the extensions your schema uses. Grab the Session pooler connection string from the Connect button at the top of the dashboard — not the Direct connection string, which runs over IPv6 and fails on most home and CI networks:

export DB_URL="postgresql://postgres.<project-ref>:<password>@aws-0-<region>.pooler.supabase.com:5432/postgres"

pg_dump "$DB_URL" \
  --format=custom \
  --schema=public \
  --file="backup-$(date +%F).dump"

--schema=public captures your application schema; Supabase-managed schemas (auth, storage metadata) stay out, by design. If you reach for the official supabase db dump instead, know its default output contains no rows — schema definitions only. The .sql export guide covers both commands and that trap in detail.

Option 2 — put it on a schedule (fifteen minutes)

A one-off dump ages fast. The GitHub Actions guide has a complete weekly workflow that dumps your database — and optionally Storage files — into a bucket you own, for free. Weekly is the right frequency here: every backup counts against the Free plan’s 5 GB/month egress allowance, and free databases are capped at 500 MB, so weekly stays around 2 GB/month even at the ceiling while daily would blow through the allowance three times over. Use a private repo — GitHub disables scheduled workflows in public repos after 60 quiet days, and a backup repo is exactly the kind that goes quiet.

Option 3 — let a service run it (also free)

BackupDrill’s free plan covers one project with weekly backups into your own S3, R2, or B2 bucket — the bytes never touch our disks — with an email if a backup fails. Your first backup also gets a restore drill: the snapshot is restored into a throwaway Postgres and verified, so you see proof it actually restores rather than a green checkmark. Scheduled weekly drills are what the paid plans add. Prefer to self-host the whole loop on a machine you run? That exists too — the tools comparison covers the honest options, including the free ones.

FAQ