Short answer
Look for tasks that repeat, consume attention and produce a checkable outcome. The best candidate is not always the most annoying one; it is where frequency, time and impact accumulate.
What to observe
For one week, log each occurrence, approximate time, who waited and what an error caused. Then separate necessary work from unnecessary movement. The opportunity is often not doing a task faster but removing transfers between spreadsheets, messages and systems.
The team copies and pastes to keep tools aligned.
An absence or volume spike immediately creates delay.
The outcome always follows the same structure.
A concrete example
A monthly report may require three downloads, renamed fields, joined columns and the same summary. Before automating the report, check whether the sources can be connected and whether the cleaning rules are stable.
How to move safely
- Count occurrences and minutes for five working days.
- Include waiting, hand-offs and corrections, not only clicks.
- Choose one point with a measurable outcome for the first test.
Test your process
Tick what happens today. The result changes as you answer.
There is not enough evidence to automate yet. Observe the process for a few days and record volume, exceptions and time spent.
Frequently asked questions
How many repetitions justify automation?
There is no universal number. Low frequency may matter when every case blocks sales, service or compliance; high frequency may not matter when a task takes seconds.
Should we start with the task people dislike most?
Use frustration as a clue, not the decision. Compare time, risk, impact and ease of validation.