AutomationLearning
Automations That Don't Break at 3am
Building n8n workflows that survive real traffic: idempotency, retries, and knowing when something failed before your client does.
May 28, 20261 min read
Automation is easy to start and hard to trust. A workflow that runs once in a demo is not the same as one quietly processing real events for weeks. Here is how I make n8n workflows I can stop babysitting.
Make every step safe to retry
Networks fail and webhooks fire twice. If a step is not idempotent, a retry creates duplicate records or double-sends an email. I key writes on a stable id so running the same step twice is a no-op.
Fail loud, not silent
- Wrap external calls with error branches, not just the happy path.
- Send yourself a notification on failure so you hear about it before the client does.
- Capture the input payload on error so you can replay it after the fix.
Building something in this space?
I take on select builds when the work is worth doing right.