Guide
n8n workflow monitoring that looks at the work, not the spinner.
The executions list tells you a workflow ran. It does not tell you the client still received leads. This is the monitoring setup we use for n8n instances that feed CRMs.
What to watch on every production workflow
Import the workflow over the public REST API and keep four rules on by default: no successful execution in N hours, a success that output zero items, the workflow flipped to inactive, and an error rate above a threshold you choose.
N depends on the schedule. A form webhook might allow 6 quiet hours overnight. A daily digest should allow a little more than 24. A lead-ads poller that usually runs every 15 minutes should not be silent for an afternoon.
Poll on a schedule, alert once
LeadPulse pulls /api/v1/workflows and /api/v1/executions on a cron, every 10 minutes. Each problem opens one incident. Later checks update that incident instead of sending another email. When the condition clears, you get a single recovery.
That is the difference between a monitor and a log drain. Agencies do not want a Slack channel that scrolls. They want a sentence: this workflow stopped delivering, and it has been since Tuesday 06:00 UTC.
Give the client proof
A status dot in your own dashboard is for you. The Monday Automation Health Report is for the client: uptime, runs, items processed, and the incidents from the week, under your logo.