Stop getting woken up at 3 a.m. by homegrown webhooks
Webhook delivery is an always-on distributed system. Endpoints time out, retries pile up, and one customer's outage backs up everyone else. Svix runs it for you, so failures turn into retries and alerts instead of pages.
Six ways homegrown webhooks wake people up, and what Svix does instead
Sending the POST is the easy part. Running delivery for real customers is what ends up on the pager.
Endpoints that time out
Every attempt gets 15 seconds, then moves on. A slow customer endpoint never holds your workers open or stalls the rest of the queue.
Retry storms
Retries back off exponentially (5 seconds, 5 minutes, 30 minutes, then hours) instead of hammering an endpoint the moment it comes back.
One customer's outage backs up everyone
Delivery is isolated per endpoint, so one dead endpoint never delays webhooks to your other customers.
A recovered endpoint falls over again
Set a throttle per customer or per endpoint, so a backlog drains at a rate their servers can take.
Endpoints that never come back
Endpoints that keep failing for five days are disabled automatically, and you get an endpoint.disabled webhook instead of an ever-growing retry queue.
“Did you send it?” tickets
Every attempt and response is logged, and customers check and replay deliveries in an embedded portal instead of asking an engineer to grep.
Get told about failures, not paged for them
Svix sends you operational webhooks as delivery degrades: message.attempt.failing after four failed attempts, message.attempt.exhausted when retries run out, and endpoint.disabled when an endpoint is switched off.
Route them to Slack or your incident tool, and page a human only when one needs to act.
{
"type": "message.attempt.failing",
"data": {
"appId": "app_2KWPBgL1Afxdpx2AI54p",
"appUid": "customer_123",
"endpointId": "ep_2KWPBgL1Afxdpx2AI54p",
"msgId": "msg_2KWPBgL1Afxdpx2AI54p",
"lastAttempt": {
"id": "atmpt_2KWPBgL1Afxdpx2AI54p",
"responseStatusCode": 502,
"timestamp": "2026-10-07T03:14:07Z"
}
}
}
“As for uptime and reliability, in our years of using Svix we never had any issues, concerns or incidents. We just set it up and forgot it existed.”
Uptime you can hold us to
Uptime SLAs are 99.999% on Enterprise, 99.99% on Professional, and 99.9% on Basic, with live status for every region.
Startups get $12,000 in free credits
Incorporated in the last three years and raised $7M or less? Get $1,000 a month in Svix credits for a year, plus hands-on onboarding from our team. Companies in the current YC batch get $50,000.
What happens when a customer's endpoint is down?
Svix retries on a fixed schedule: immediately, then after 5 seconds, 5 minutes, 30 minutes, 2 hours, 5 hours, 10 hours, and 10 hours, which covers about a day. If the endpoint keeps failing for five days, it's disabled and you get an endpoint.disabled webhook. Customers can replay everything that failed once they're back.
How long does Svix wait for a response?
15 seconds. Any 2xx response within that window counts as delivered. Anything else, including redirects and timeouts, is retried.
How do we get alerted?
Through operational webhooks: message.attempt.failing after four failed attempts, message.attempt.exhausted when retries run out, message.attempt.recovered when a failing message gets through, and endpoint.disabled. Send them to Slack or your incident tool.
What uptime does Svix commit to?
Uptime SLAs are 99.999% on Enterprise, 99.99% on Professional, and 99.9% on Basic. Live status is at status.svix.com.
Can we self-host Svix?
The Svix server is open source and you can run it yourself. Most teams use Svix Cloud so they don't have to operate it, which is the point of this page.
