ai Case study
n8n CRM Notification Router
The failure mode of operational notifications is not missing them, it is getting all of them. A channel that receives every event becomes a channel people mute, and then the one notification that mattered is the one that was missed.
The business problem
CRM and commerce events were going to a single destination regardless of what they were or who needed to act. Everything looked equally urgent, which means nothing did. The fix is not fewer notifications, it is routing: the event type has to determine who sees it and whether it needs approval before anything happens.
What I delivered
- n8n routing logic that classifies each CRM and commerce event and directs it by type.
- Routing to a specific team member or channel rather than to a general destination, so a notification arrives with an owner.
- Approval queues for the events that should not proceed automatically, separating what is informational from what requires a decision.
- Error branches, so a failure in the routing surfaces rather than silently dropping the event.
- Credentials held in provider vaults or server-side configuration rather than in the workflow definitions themselves.
Technical approach
- Classification comes before routing. Every rule about who sees what depends on the event type being determined in one place rather than inferred repeatedly downstream.
- Notifications are addressed to someone. An alert with no owner is a notification that everyone assumes somebody else is handling.
- Approval gates are placed where a mistake would be expensive, and deliberately not everywhere, because a workflow that asks for approval constantly gets approved without being read.
- Every branch has an error path. Automation that fails silently is worse than none, because people have already stopped doing the manual version.
Result and evidence
Events reach the person who needs to act on them, and the ones that require a decision wait for it. The general channel stopped being the place notifications go to be ignored.
Commercial value
Operational awareness is a routing problem, not a volume problem. Getting the routing right is what makes the alerts worth having at all.
Readable implementation brief
implementation_brief {
project: "n8n CRM Notification Router"
order: "classify -> route -> notify/queue"
addressing: "specific person or channel; never a general
firehose"
gates: "approval only where a mistake is expensive"
failure: "explicit error branch on every path; nothing is
dropped silently"
secrets: "provider vaults or server-side config"
}