What counts as delivered
A delivery succeeds when your endpoint answers with any2xx status within 10 seconds. The response body is ignored.
Everything else is a failed attempt:
Retry schedule
The first attempt usually happens within a minute of the event. After a failed attempt, it waits before trying again:
Attempts can start a little after their scheduled time. If the 7th attempt fails, the delivery is marked dead. It stays in the delivery log, and you can redeliver it once your endpoint is fixed.
Every attempt is signed fresh, with a new
X-Quotable-Timestamp. The event id, X-Quotable-Event-Id and X-Quotable-Delivery-Id stay the same across attempts.
Automatic disabling
If 50 attempts in a row fail across an endpoint, counting all of its deliveries, Quotable disables the endpoint and shows the reason in Settings → Webhooks. Any successful delivery resets the count. While an endpoint is disabled:- New events aren’t queued for it.
- Deliveries that were still waiting for a retry stay queued, and are sent when you resume it.
Pausing
When you Pause an endpoint, new events keep being queued for it but nothing is sent. Select Resume to send everything that queued up while it was paused. Use this during a planned migration or maintenance window instead of letting deliveries fail.Delivery guarantees
At least once
An event can arrive more than once, for example when your server processed it but answered after the 10-second timeout, or when someone redelivers it. Deduplicate on the event
id.Not ordered
Deliveries go out roughly in the order events happened, but a retry can land after a newer event. Use
occurred_at to order events, not arrival time.When the endpoint’s creator loses access
An endpoint delivers on behalf of the person who created it, within the brands that person can access. If that person leaves the workspace, or loses the role or brand access the endpoint needs, its deliveries stop: each one is marked dead without being sent. Settings flags these endpoints with a Creator left badge when the creator has left. To restore deliveries, have a current owner or integration admin add a new endpoint, then remove the old one.Build a resilient receiver
- Respond fast, work later. Verify the signature, store the event, answer
2xx, then process it from a queue or background job. - Make processing idempotent. Record each event
idwith a unique constraint so a duplicate is a no-op. - Treat events as hints. When the exact current state matters, fetch it from the API using
resource.href. - Monitor the delivery log. Dead deliveries and a disabled endpoint mean your receiver missed events.
Next steps
Manage endpoints
Pause, resume, rotate, redeliver and read the delivery log.
Verify signatures
Check every delivery before you act on it.