Receive signed events for every delivery, bounce, open, click and complaint, and replay what you missed.
AI summarises event streams and flags unusual bounce patterns as they appear.
{
"event": "bounced",
"message_id": "msg_01H...",
"recipient": "user@example.com",
"reason": "550 5.1.1 mailbox unavailable",
"timestamp": "2026-10-13T08:15:00Z"
}
Delivered, deferred, bounced, opened, clicked, complained and unsubscribed.
Verify every request came from MigoSMTP.
Failed deliveries are retried automatically.
Re-send events from the log after an outage.
Subscribe to only the events you need.
See request, response and status for every attempt.
Enter your HTTPS URL and choose events.
Check the signature header on each request.
Return a 2xx response to confirm receipt.
Switch between stages to see what you control at each step. Illustrative interface with sample data.
Update your database, trigger workflows and alert your team without polling.
Mark orders as notified when delivered and flag contacts when a hard bounce arrives. Webhooks remove the need for polling jobs and keep application state accurate.
On bounce or complaint events, add the address to your own do-not-mail table straight away, so other systems that email the same person stop as well.
Send a Slack or ticket alert when a password reset bounces. Support can then contact the customer by another channel before they give up. It saves hours of manual work.
If your endpoint was offline during a deployment, replay the missed events from the log and bring your records back to the correct state. Nothing is lost in the meantime.
A copilot condenses thousands of events into a short digest, such as which provider is deferring and which campaign drives complaints, linking to the messages behind each point.
Models learn normal bounce, complaint and deferral rates and alert you when a stream departs from them. You set who is notified and whether any automation runs.
Paste a failed delivery log and the assistant explains the likely cause, such as a signature mismatch or timeout, and suggests a fix to test. You make the change.
Through an MCP server, AI agents can read event history and delivery logs with read-only scopes. Write actions need explicit scopes and, if you wish, human approval.
AI features are assistive: they summarise and alert, and any automated action is something your team configures and approves. [Confirm which AI features are available at launch.]
Reject requests whose signature header does not match. Rotate the secret periodically and straight away if it may have been exposed.
Acknowledge with a 2xx as soon as the payload is safely stored, then process it from a queue. Slow handlers cause timeouts and duplicate retries.
Deliveries can repeat and arrive out of order. Use the message ID and timestamp to deduplicate and to ignore stale state changes.
Filter events per endpoint. Fewer irrelevant requests means less load on your servers and simpler monitoring. Keep integrations lean and quick to debug.
Alert on repeated failures in the delivery log. A silent outage means silent data drift until someone replays the gap.
An email webhook is an HTTP request your email provider sends to your server when an event occurs, such as a message being delivered or bouncing. It lets your application react in real time without checking the API repeatedly. This replaces polling jobs and keeps your records current.
Each request includes a signature header generated with your endpoint's secret. Recompute the signature from the raw request body on your side and compare. Reject the request if they differ. The documentation has code samples. [Confirm signature scheme.] Sending a sample event from the dashboard confirms your handler works.
Delivered, deferred, bounced, opened, clicked, complained and unsubscribed. You can subscribe each endpoint to a subset, so one service gets bounces and another gets engagement events. Pick only those your systems act on, such as bounces for suppression and clicks for analytics, to keep traffic to your endpoint small and relevant.
Failed deliveries are retried with backoff, and you can replay events from the log once you recover. [Confirm retry schedule.] Each attempt, with request and response, is recorded so you can see what failed. Check that your handler returns quickly and logs failures.
[Describe ordering and duplicate handling.] As a general rule, build handlers that tolerate duplicates and out-of-order events by using the message ID and timestamp. Event ordering can differ across providers and retries, and a single message may generate several events in a short time. Handle each event idempotently, compare timestamps before changing state, and keep the original event for troubleshooting purposes.
Some mail systems fetch images automatically, which registers as an open even if no person read the message. We separate those from human opens where we can, but open tracking remains an estimate. Treat opens as a rough indicator. Count clicks and replies as stronger signals of engagement.
Send a sample event from the dashboard to your URL and check that your handler verifies the signature and returns a 2xx. Review the delivery log to see the request and your response. Use a staging endpoint with a test secret before production. Run it before pointing production at the endpoint.
AI can summarise event streams, flag unusual bounce or complaint patterns and help debug failed deliveries. It recommends actions but does not change your configuration unless you set up a rule. [Confirm which AI features are available at launch.] Review recent logs before connecting new endpoints.
Some mail systems fetch images automatically. We separate those from human opens where we can.
Send a sample event from the dashboard.
Create an account, add your domain and send your first email.