What to Do When an Automation Suddenly Stops Working

Sooner or later, an automation that was working fine may suddenly stop. It could be something as simple as an expired credential, or something more complex like a change on an external platform. What matters is having a clear process to diagnose it quickly, instead of reacting haphazardly.

First steps to diagnose the problem

1. Identify when it stopped working — check whether there’s an activity log showing the last time it worked correctly.
2. Check for a recent change — an update to a connected tool, a password or permissions change, or a modification to the process the automation doesn’t account for.
3. Check connections and integrations — many automation failures come from an external platform changing something (an API, a data format) without notice.
4. Check whether data volume changed significantly — sometimes a system starts failing simply because the volume it has to process grew more than expected.

How to keep the impact from piling up while you fix it

While you diagnose the problem, it’s worth having a manual contingency plan for the most critical parts — for example, if customer service automation fails, having someone on the team manually cover the most urgent inquiries until it’s resolved.

Why active monitoring beats purely reactive monitoring

Many failures get caught late because no one is actively monitoring whether the automation is still working — the problem gets discovered when someone notices something didn’t happen (a customer left unanswered, a piece of data not updated). An active monitoring system, with automatic alerts on abnormal behavior, catches the problem well before someone has to notice it on their own.

What to do once the immediate problem is fixed

Fixing the symptom isn’t enough — it’s worth understanding the root cause to keep the same problem from happening again. If it was a change on an external platform, for example, it’s worth checking whether the automation can be made more resilient to that kind of change going forward.

The expected result

A clear process to react quickly when something fails, instead of losing valuable time trying to figure out what happened from scratch every time. This is part of what maintenance & continuous improvement covers — not just preventing failures, but having a clear process to fix them fast when they happen.