# Limits, access and retention

Understand finite request limits, plan transitions and bounded recovery windows.

## Read the effective limits

Standard engineering defaults are 120 reads and 30 writes per minute per key and per tenant, 5 active keys and 10 subscriptions. GET /me and the integrations dashboard show the limits effective for your account. Enterprise limits are finite and server-controlled.

Limits apply across keys, so creating another key does not increase tenant capacity. Follow the returned rate-limit headers and Retry-After. Pages contain at most 100 records and requests are bounded to 64 KiB.

Guest creation still consumes the existing plan allowance. API use does not bypass credit accounting, booking capacity, messaging eligibility or opt-outs.

## Handle changes in plan access

Starter and Growth require an upgrade for production API, Zapier and webhook access. Business and Enterprise eligibility does not override a disabled account, demo mode, unpaid access or a release gate.

Existing lower-tier integrations require an explicit migration decision before enforcement. If a migration entitlement is recorded for your account, the dashboard shows its end date. The proposed transition period is not an automatic grant to every account.

After access ends, managers retain redacted metadata and shutdown controls. Restore eligible access before using unrevoked keys again, and explicitly resume paused subscriptions for new events.

## Plan for bounded recovery

Completed idempotency results are kept initially for 30 days; the event feed recovery window is initially seven days. Delivery diagnostics are redacted and bounded. Application privacy and account-deletion requirements can remove retained personal data sooner.

Export only the permitted records your workflow actually needs. Do not depend on an old entry snapshot, event cursor or delivery diagnostic as a permanent system of record.