Choose a task based on repetition, rules, and risk

Don’t automate something just because you can

A suitable task is often repetitive, has steps that can be described, and produces a result that can be checked. But that’s not enough. Also consider what could happen if something goes wrong, how often exceptions occur, and whether the task involves personal or sensitive data. The greater the potential impact of a mistake, the stronger your checks should be and the more cautiously you should introduce the change.

Choose one task and find out how often it’s performed, how much time it takes, and which parts involve repeated waiting or retyping. Ask the people who do the work: What’s routine, what varies, and which decisions depend on context? Their experience can reveal exceptions that aren’t included in a brief written guide.

Weigh the benefits against the costs of implementation, maintenance, and correcting mistakes. A small task with low risk may be a good one to test. Decisions about hiring someone, making a financial payment, or granting access to important services require much greater caution and human judgment. If you can’t assess the security or data-handling requirements, consult the person responsible before connecting the tool. A good candidate is one where you can explain both why you’re automating it and where automation’s limits are.

Also consider whether the rules change depending on the season, the client, or the type of request. If they change often, it may be safer to automate only the stable part and leave the rest to a person.

What should you check when choosing a task to automate?

Automation in Practice: Choose Tasks and Check Results

Continue the training

Enter your email address to continue the training.