Webhook retries and handling failures

Endpoints go down. We expect that and retry failed deliveries automatically, but designing your handler with retries in mind keeps your data complete.

When a delivery is considered failed

Any response that is not a 2xx, or that does not arrive within the timeout, counts as a failure and triggers a retry.

The retry schedule

We retry with exponential backoff over roughly a day — quick attempts at first, then progressively longer gaps. After the final attempt the delivery is marked failed and surfaced in the dashboard.

Designing for retries

• Make your handler idempotent by deduplicating on the event id.
• Acknowledge fast, then process asynchronously, so a slow downstream system does not cause retries.
• Use the Replay button in the dashboard to re-send events you missed during an outage.

If a whole endpoint is failing repeatedly, we may pause it and email you. Fix the endpoint, then replay the backlog from the dashboard.