Every automation conversation starts with an assumption about where the time goes. In our experience that assumption is wrong more often than it is right — which is exactly why the first step is measurement, not tooling.
Re-typing data that already exists
Someone reads a number off a PDF invoice and types it into a spreadsheet, then types it again into the accounting system. This is the single most common one, and it is usually invisible because it is spread across many small moments.
Routing that a human does by memory
An enquiry arrives and a person decides who should handle it. The rule they apply is consistent and unwritten. Consistent and unwritten is exactly the shape of work a machine can take over — once you write the rule down.
Reports assembled by hand
A weekly report that takes forty minutes of copy-paste is thirty-five hours a year. It is also the task most likely to be skipped when things get busy, which means you lose the report precisely when you need it.
How we measure it
For one week, every repetitive task gets a line: what it was, how long it took, whether the output was ever wrong. No tooling decisions yet. At the end of the week the ranking is usually obvious, and it usually surprises the owner.
Then automate exactly one thing
Pick the top item and ship it. One automation with a measurable result beats a platform migration that never finishes. If it does not save measurable hours in the first month, it was the wrong thing to automate.