n8n Workflow Monitoring: Get Alerts for Failures and Silent Failures
By Muhammad Usman · · 7 min read
The three ways n8n workflows break
- Failed executions: a node throws an error (an API is down, credentials expired, data is in the wrong format).
- Silent failures: the execution finishes with "success", but did nothing useful. For example, an order sync returns zero orders because an API changed a field name.
- Workflows that stop running: a trigger breaks, a webhook URL changes, or someone deactivates the workflow, and nothing runs at all.
Most teams only watch the first kind. The second and third are usually found days later, by a customer or a colleague asking "why is this data missing?"
Instant alerts for failed executions: the Error Workflow
n8n has a built-in feature for this:
- Create a new workflow that starts with the Error Trigger node.
- Add a notification node (Slack, email, Telegram or an HTTP request).
- In each production workflow, open Settings → Error workflow and choose the workflow you just made.
Now every failed execution sends an alert with the workflow name and error message. It's free and takes five minutes, so do it first.
Catch silent failures
An Error Workflow can't see silent failures, because nothing failed. To catch them:
- Check the output count. For each important workflow, decide what "normal" looks like (for example "at least 1 order per hour on weekdays") and alert when a successful execution produces zero items.
- Compare with history. A run that processes 3 items when it usually processes 300 deserves a look.
- Watch duration. A run that takes 10 times longer than normal is often about to fail.
Catch workflows that stopped running
Set a heartbeat per workflow: "alert me if this workflow hasn't run for 2 hours". This catches broken triggers, deactivated workflows and a stopped n8n instance. Pick the window based on the schedule: a workflow that runs every 15 minutes might get a 1-hour window, a daily report a 26-hour window.
Watch error rates, not just single errors
One failed execution in a thousand can be noise. A jump from 1% to 30% failed runs within an hour is an incident. Alerting on the error rate avoids alert fatigue while still catching real outages.
A practical n8n monitoring checklist
- Error Workflow connected to every production workflow
- Zero-output alert for workflows that must always produce data
- Heartbeat alert for scheduled and webhook workflows
- Error-rate alert per workflow
- Alerts sent where you'll actually see them (Slack, WhatsApp or SMS, not only email)
- A weekly summary of success rates
How ProofMyAI monitors n8n
ProofMyAI connects to n8n with an API key (n8n → Settings → n8n API), reads your executions every 15 minutes and alerts you on failures, silent failures, error-rate spikes, slow runs and workflows that stopped running. For faster alerts, it can also receive executions through a webhook. See the setup docs.
✅ Key takeaways
- n8n workflows fail in three ways: errors, silent failures and simply not running.
- Set up an Error Workflow first: it's free and catches failed executions instantly.
- Alert on zero output and unusual counts to catch silent failures.
- Use a heartbeat per workflow to catch broken triggers and stopped instances.
❓ Frequently asked questions
Does n8n have built-in alerts?
Yes. The Error Trigger node lets you build an Error Workflow that runs whenever another workflow fails. It doesn't catch silent failures or workflows that stop running, so add output and heartbeat checks.
What is a silent failure in n8n?
An execution that finishes successfully but produces nothing useful, for example a sync that returns zero records because an API response changed.
Does monitoring work with self-hosted n8n?
Yes. Monitoring through the n8n API works with n8n Cloud and self-hosted n8n, as long as the API is reachable from the monitoring service.