The useful question is not “How much time can this save?”

A task that takes ten minutes can still be a poor automation target. It may happen rarely, be costly to get wrong, or require more time to maintain than it returns. A baseline makes the decision less impressionistic.

Count the whole system

Record how often the task happens, how long it takes, who does it and what errors or delays matter. Use a range where the timing varies. Then include implementation, subscriptions, training, exception review and ongoing maintenance.

Be precise about whose time changes. A process might reduce one person’s admin while creating review work for someone else. It might also move effort into setup or make a once-visible decision harder to understand.

Run a small, reversible trial

Some saved minutes return as useful focus. Others fragment the day or create new maintenance. A small trial can reveal which is true before the process becomes dependent on a new tool.

Choose a period and a simple measure in advance. Compare like with like, include failures and review the result with the people who use the system. If the evidence is mixed, change the design or stop.