When is it worth automating a task?

When all three hold: it repeats often (daily or weekly), it has clear rules you could explain to a new employee as a checklist, and the process is stable. Then automation saves hours and stops the mistakes hiding in boring manual work.

Which tasks should I NOT automate?

Tasks that happen rarely (the cost never pays back), decisions that need human judgement, and processes that keep changing. Automate chaos and you get faster chaos.

How do I know if an automation will pay for itself?

Two numbers: how many hours the task eats per month and how stable its steps are. Many hours on stable, repetitive work almost always pays back.

← All posts 15 April 2026 Updated: 3 June 2026

When automating a task is worth it (and when it isn't)

The most common question I hear isn’t “can you automate it?”. Almost anything can be automated. The right question is “is it worth it?”.

Automate when all three are true

It repeats. A task done every day or every week quickly pays back the time you spend building it. Something done twice a year doesn’t.
It has rules. If you can explain the steps to a new hire with a checklist, it can be automated. If it needs judgment every time, it gets harder.
It’s stable. A process that changes every month will keep breaking. Stabilise it first, then automate.

When all three hold, automation doesn’t just save time. It also kills the errors that hide inside boring manual work.

Don’t automate (yet)

Tasks done rarely. The cost never pays back.
Decisions that need human judgement. An agent can prepare them, but the call stays with a person.
Processes that haven’t settled. Automate chaos and you get faster chaos.

The rule in practice

Before I automate anything, I measure two things: how many hours a month it eats, and how stable the steps are. Many hours on stable, repetitive work? It’s almost always worth it. If not, I’ll tell you honestly before you spend a euro.

Got a routine task eating your time? Tell me where you lose hours.

FAQ

When is it worth automating a task?: When all three hold: it repeats often (daily or weekly), it has clear rules you could explain to a new employee as a checklist, and the process is stable. Then automation saves hours and stops the mistakes hiding in boring manual work.
Which tasks should I NOT automate?: Tasks that happen rarely (the cost never pays back), decisions that need human judgement, and processes that keep changing. Automate chaos and you get faster chaos.
How do I know if an automation will pay for itself?: Two numbers: how many hours the task eats per month and how stable its steps are. Many hours on stable, repetitive work almost always pays back.