If you run n8n workflows for clients in production, you probably have an error workflow. An Error Trigger connected to Slack or email, maybe with the workflow name and error message in the notification. It is the first thing every guide tells you to set up, and it is genuinely useful.
It is also about 40% of what you need.
This is not a criticism of error workflows — they do exactly what they are designed to do. The problem is that what they are designed to do is narrower than what production actually throws at you. Here are the gaps, in the order they usually bite.
An error workflow that is set to inactive is skipped in silence. n8n writes one line to the server log and moves on. Every failure after that goes unreported. You believe you have alerting. You have none.
Worse: there is no global error workflow in n8n. It is set per workflow, in each workflow's own settings. So every workflow you add later is uncovered by default, and nothing counts them for you. In a 30-workflow instance, it is a matter of when, not if.
A node set to continueOnFail (or onError: continueRegularOutput)
can fail, and the execution still finishes green. The error workflow never fires, because
n8n does not consider the execution failed. The run status says success.
The work did not happen.
This is the most expensive class of failure in production, because nobody finds out until a human on the client side notices — and by then it has been days or weeks.
Catching these requires walking data.resultData.runData per node, not just
reading the execution status. An error workflow has no mechanism for this — it only fires
when the execution itself fails.
Full technical write-up here.
An error workflow gives you:
What it does not give you:
An expired Google OAuth credential, a client renaming a spreadsheet column, and a bug in your expression all arrive looking identical: "workflow failed." Two of them are a message to the client. One is you opening your laptop. Working out which, every time, by hand, is where the margin on a retainer goes.
If your error workflow itself has a bug — a Slack webhook that expired, a channel that was renamed, a rate limit on the notification service — you get silence. The error workflow fails trying to report a failure, and nothing reports that.
This is the "who watches the watcher" problem, and it is not theoretical. It was one of the first bugs found in real testing of production n8n setups: the error handler had been failing for two weeks, and not a single alert had gone out.
An error workflow is reactive — it fires after something breaks. It cannot tell you that a deployed workflow has no trigger, is set to inactive, has nodes that swallow errors, or is missing retry configuration on network calls. Those are risks that exist right now in your instance, waiting for the right moment to become incidents.
| Capability | Error workflow | Full monitoring |
|---|---|---|
| A run failed | ✓ | ✓ |
| Someone gets notified | ✓ | ✓ |
| Silent failure caught | ✗ | ✓ |
| Which client it belongs to | ✗ | ✓ |
| Whose fault it is | ✗ | ✓ |
| What to do about it | ✗ | ✓ |
| Risk flagged before failure | ✗ | ✓ |
| Error workflow itself failing | ✗ | ✓ |
| Client can check status | ✗ | ✓ |
Error workflows are good. They are the right first layer. The problem is treating them as the only layer — which is what most production n8n setups do, because until recently, there was nothing else that worked at the client-management level.
An agency managing 10+ client workflows needs three things an error workflow cannot provide:
"I check my executions manually every morning."
That works until it doesn't. You miss a morning, you are on holiday, you add a new workflow and forget to check it. Manual checks do not scale, and the failure you miss is always the one the client finds first.
"I use Prometheus and Grafana for monitoring."
Prometheus tells you execution counts and error rates. It does not tell you which client, whose credential expired, or what to say to the client. It is infrastructure monitoring, not client-operations monitoring. Both are useful; they answer different questions.
"I built my own monitoring with a heartbeat workflow."
Heartbeat workflows catch the "workflow stopped running" case — which is real and important. They do not catch silent failures (run finished, work didn't happen), and they do not classify failures by cause. If you have built one, you are ahead of most. The gap is in what happens after the alert.