Errors, Alerts, and Run Logs
Plan for failure so a broken automation gets noticed and fixed, not ignored for weeks.
Objective
By the end of this lesson, you can add an error alert and a run log to your automation, and write a short plan for what to do when it breaks.
Every automation breaks eventually
Passwords change. An app updates and renames a field. Someone edits the form and deletes a question. These are normal.
The danger is not that an automation breaks. The danger is that it breaks quietly. For two weeks, requests go nowhere, and nobody knows.
Three protections
1. Error alerts. Set your tool to email or text a named person when a run fails. Zapier, Make, and n8n each have error notification settings. Turn them on, and send them to someone who will read them, not a shared inbox nobody checks.
2. A run log. Add a last step that writes one row to a "Run log" tab: time, item ID, and result. If the log goes quiet on a busy day, something is wrong.
3. Retries and fallbacks. Many tools can retry a failed step a few times. If it still fails, the fallback should send the item to a person, not drop it.
A heartbeat check
A heartbeat is a simple daily check that the automation is alive. For example, every morning at 9, a small automation counts yesterday's rows in the run log. If the count is zero on a weekday, it texts the owner: "No catering requests logged yesterday. Please check the automation."
Zero might be real. A slow day happens. But the text costs the owner five seconds and can save a week of lost requests.
A local example
Write a break plan
A break plan is a few lines in the handoff document that answer:
- Who gets the error alert?
- Where do they log in to see what failed?
- What is the manual backup while it is broken? For example, "Check the form responses tab directly each morning."
- Who can they call for help?