An automation is not a product you buy once. It connects programs that keep changing. Anyone who knows this and plans ahead rarely has trouble.
What typically breaks
- Expired access: passwords or permissions are renewed and the workflow doesn't know.
- Changed interfaces: a provider switches versions.
- Changed forms: a new field on the website and the mapping no longer fits.
- Unexpected data: an umlaut, an empty field, an unusual file format.
- Server issues: updates, disk space, certificates.
Good monitoring
- Error notification: if a run fails, a message goes immediately to the person responsible.
- Sign of life: if a workflow stops running altogether, nobody notices without a check. A daily checkpoint helps.
- Retry: an automatic second attempt catches short disruptions.
- No data loss: failed entries are collected and processed later.
Maintenance at a glance
- Updates of automation platform and server
- Checking access and expiry dates
- Backup of workflows and settings
- Short report: what ran, what was fixed
With self-hosted n8n the server is part of it too. With cloud services such as Zapier the provider looks after the platform; the workflows remain your responsibility.
FAQ
How often does an automation fail?
Well-built workflows usually run stably for a long time. Failures mainly come from changes outside: new versions, expired access, changed forms.
Can I monitor this myself?
Yes, with error notifications by email or chat. What matters is that someone is responsible and knows what to do in case of failure.
