← Guides

How to back up an Airtable base: snapshots, CSV exports and the API

Updated 2026-11-22. Sources are listed at the end.

Airtable gives you three ways to keep a copy of a base. They differ in what they keep, where the copy lives and how much work they take.

Airtable backup methods compared
MethodWhere the copy livesScheduleEffort
SnapshotsInside AirtableAutomatic, based on usage; manual snapshots on demandNone
CSV downloadYour computerManualOne view of one table at a time
Web API scriptWherever you chooseYour schedulerWrite and maintain a script

Option 1: Snapshots

Airtable takes snapshots of a base automatically, based on how much the base is used, and Editors or Creators can take one by hand. Scheduling snapshots at set intervals is not supported. How long snapshots are kept depends on the plan:

Snapshot retention by Airtable plan
PlanSnapshots kept for
Free2 weeks
Team1 year
Business2 years
Enterprise Scale3 years

Restoring a snapshot creates a new base and leaves the existing one untouched. Snapshots are stored in Airtable, so they are not a copy outside Airtable.

Option 2: Download a CSV

From a grid view, open the menu next to the view name and choose Download CSV (web and desktop apps). The file holds the field values visible in that view, so remove filters if you want every record. Record comments, field descriptions, the base guide and data kept only in extensions are not included. Attachments appear as a file name and a link that expires after about two hours, so download the files promptly if you need them. Details: how to export Airtable to CSV.

Option 3: A script using the Web API

  1. Create a personal access token with only the scopes data.records:read and schema.bases:read, limited to the bases you want to copy. Airtable does not let a script read back a token's scopes, so take care when creating it.
  2. Save the schema: GET https://api.airtable.com/v0/meta/bases/{baseId}/tables.
  3. Page through each table, 100 records per request, following the offset value until there is none:
BASE=appXXXXXXXXXXXXXX
TABLE=tblXXXXXXXXXXXXXX
OFFSET=""
PAGE=0
while true; do
  URL="https://api.airtable.com/v0/$BASE/$TABLE?pageSize=100"
  [ -n "$OFFSET" ] && URL="$URL&offset=$OFFSET"
  curl -s -H "Authorization: Bearer $AIRTABLE_TOKEN" "$URL" > "page-$PAGE.json"
  OFFSET=$(jq -r '.offset // empty' "page-$PAGE.json")
  PAGE=$((PAGE + 1))
  [ -z "$OFFSET" ] && break
  sleep 0.25   # stay under 5 requests per second per base
done

Limits a backup script must respect:

  • 5 requests per second per base. After a 429 response, wait 30 seconds before the next request.
  • Monthly call caps per Airtable workspace: 1,000 calls on Free and 100,000 on Team; Business and Enterprise Scale have no monthly cap. See Airtable API limits explained.
  • Attachment links expire after about 2 hours, so download attachment files during the same run.

A rough budget: a base with 20,000 records needs about 200 list calls per full copy (100 records each), so a daily copy uses about 6,000 calls a month. That fits within Team's 100,000 but not Free's 1,000.

Keep several days of copies, not just the latest: a bad bulk edit or a deleted table is copied too on the next run.

Option 4: A backup service

Paid services also exist, for example ProBackup and On2Air. Depending on the service and plan, copies are kept in the provider's storage or sent to a cloud drive such as Google Drive (per their pricing pages, 22 November 2026).

Lindenkit Backup for Airtable: coming soon

We are building a scheduled backup of the Airtable bases you choose (records, attachments and schema) into a Cloudflare R2 bucket you own, using a token with read scopes only. It is not available yet and has no release date.

Read about it and join the waitlist

Sources