Skip to main content
AI Automation With Zapier, Make and n8n

LESSON 6 OF 6

Keep it running

BY THE END OF THIS LESSON

Stop a working automation from quietly rotting.

Automations rot from the outside

3 stages, each leading to the next.
  1. A column is renamed
  2. It keeps running
  3. Gradually stops being correct

Nothing inside your automation changes on its own. Everything around it does: a form gains a field, a column is renamed, a team changes what a category means, someone starts using a different inbox.

The automation keeps running throughout. It just gradually stops being correct, and because it never errors, nothing tells you.

Write down what it assumes

Next to the automation, with 0 lines flagged.
Next to the automation
What it does
What it depends on
Who owns it
What would make it wrong

Next to each automation, keep a short note: what it does, what it depends on, who owns it, and what would make it wrong.

That last item is the one that matters. "This assumes the form has exactly these fields" is the sentence that saves someone an afternoon when the form changes — and without it, whoever inherits this has to reconstruct your reasoning from the steps.

Look at the output, not the dashboard

The weak version and the strong one.
AvoidEvery run succeedednothing about whether they were right
Do thisRead ten real outputsonce a month

A dashboard showing every run succeeded tells you nothing about whether the runs were right. Both look identical when a category has quietly drifted.

Once a month, read ten real outputs. You are not looking for errors — you are looking for the category that no longer means what it did, and the draft that no longer sounds like you.

WORKED EXAMPLE

LEFT ALONE

It has been running for eight months without errors, so it is fine.

REVIEWED

Monthly: read ten real outputs. Last month two were routed to a category the team stopped using in the spring, so the list was updated. Beside it, a note: depends on the intake form fields, owned by ops, wrong if the form changes or the category list does.

Eight months without errors is exactly what a rotting automation looks like from the outside. Nothing errored — the categories had simply stopped matching how the team works, and only reading the output shows that.

YOUR TURN

Write the assumptions note for something you already run.

RUN THIS

Here is an automation I have running: [describe it]. Write the note that should sit beside it: what it does, what it depends on, who owns it, and specifically what changes elsewhere would make it wrong without breaking it. Then tell me what I should look at in a monthly review, and what I would be looking for.

What a good result looks like

The "what would make it wrong" list should name real things in your setup — a specific form, a specific column, a specific category list. Anything generic has not been thought about yet.

KNOWLEDGE CHECK

Answer all 3 correctly to complete this lesson.

  1. 1. How does a working automation become wrong?

  2. 2. Why is a dashboard of successful runs not enough?

  3. 3. What is the most valuable line in the note beside an automation?

REMEMBER THIS

Automations rot from the outside — write down what they assume, and read the output monthly.