How to find the manual work worth automating

Every business has work that feels like it should not need a person: copying an order from an email into a spreadsheet, chasing a late timesheet, retyping the same figures into a second system. Automating it sounds obviously right.

It is not always. Automation costs money to build and effort to maintain. The skill is in choosing the right tasks.

Start with a list, not a tool

For one week, ask each team to note the tasks they repeat. For every task, record three things:

• How often it happens. Daily, weekly or monthly.
• How long it takes. Minutes each time.
• How predictable it is. Does it follow the same steps every time, or does it need judgement?

That is all you need to make a first decision.

What makes a good candidate

The best tasks to automate are frequent, rule-based and dull. A task done twenty times a day, the same way each time, is ideal. So is anything where a typing mistake causes real trouble, such as invoicing.

A task done once a quarter is usually a poor candidate, however tedious. It would take years to repay the cost of building the automation.

A task that needs judgement, such as deciding how to answer an unhappy customer, should stay with a person. Software can still help by gathering the information that person needs.

A quick sum

Multiply the minutes a task takes by the number of times it happens in a year. That gives the hours it costs you annually. Twenty minutes a day is about eighty hours a year, or two full working weeks. Put the tasks in order by that figure and the priorities are usually obvious.

Fix the process first

Automating a bad process gives you a fast bad process. Before building anything, ask whether the task needs to exist. A report nobody reads should be stopped, not automated. An approval that is always granted may not need to be an approval.

Start small

Choose one task near the top of the list, automate it, and run it alongside the manual way for a couple of weeks. Once people trust it, move to the next. Small steps that work build more confidence, and more savings, than one large project that tries to change everything at once.