How to work out whether automating is worth it

Time saved multiplied by frequency, against the cost of building and keeping it working.

Automation is often justified by enthusiasm rather than arithmetic. The calculation is simple and frequently says no.

Step-by-step

1. Enter time the task takes and how often it happens.
2. Enter the cost of the person doing it.
3. Enter build cost, tool cost and expected maintenance.
4. Read the payback period.

Frequency dominates

A task taking two minutes daily is eight hours a year. A task taking a day but happening annually is one day. The second feels far more worth automating and is not.

Small frequent tasks are where automation pays, and they are the ones nobody notices.

Maintenance is the forgotten cost

Automation is not built once. Systems change, formats change, an API is deprecated, an edge case appears. Budget ongoing maintenance — often 15–25% of the build cost annually — or the payback calculation is fiction.

A fragile automation that breaks monthly and needs investigating each time can cost more than the manual task it replaced.

What is not in the arithmetic

• Consistency. Automation does not have off days, which matters for anything where errors are expensive.
• Speed, where responsiveness has value beyond the hours saved.
• Scale. Manual work has a ceiling; automated work usually does not.
• Fragility. When automation breaks, does anyone still know how to do it by hand?

Frequently asked questions

Which tasks are worth automating?

Frequent small ones. A two-minute daily task is eight hours a year; a full day happening annually is one day, and rarely worth it.

How much maintenance should I budget?

Commonly 15–25% of build cost annually. Automation that touches other systems breaks when those systems change, and that time is a real cost.

Is automation always better than doing it manually?

No. Low-frequency tasks, work needing judgement, and anything changing constantly are often better left manual. And automating a task that should not exist entrenches it.