Choose a recurring loop.
A useful starting point repeats often enough to observe: a new enquiry, quote preparation, booking, invoice follow-up or a handoff between people. Name its trigger and its finish. “Improve operations” is too broad to tell you what to change.
Follow the last real example.
List the people, tools, messages and decisions involved. Mark where the information is copied, where work waits and where someone has to chase. A screenshot or a short recording of the actual process is often more useful than a perfect process diagram.
Measure the baseline.
Estimate the number of occurrences in a typical week, then observe time spent on several representative examples. Include checking, exceptions and rework. Keep the evidence source visible. Treat an estimate as an estimate until you have observed enough to rely on it.
Simplify before connecting.
Ask whether a step can disappear, whether one person can own the handoff, or whether one clear form can replace a chain of messages. Automation can repeat an unnecessary step faster. Removing the step may be the better improvement.
Protect the important decisions.
Name the actions a person should approve: a customer promise, an unusual quote, a sensitive response. Decide what happens if the connection fails or required details are missing. A dependable workflow has a visible owner and a fallback.
Compare a normal week.
Use the same measurement method after the change. Count exceptions and the work of maintaining the system. Keep it if the result is useful. Adjust it if the burden has only moved somewhere else.
Map your workflow