Most discovery calls we run don't mention software until at least twenty minutes in. We spend the first part of the call mapping the actual workflow as it exists today — every handoff, every place a lead or a record sits waiting on a person.
That map is more useful than any feature list. It tells us where automation will actually change a number, versus where it would just move the same bottleneck one step downstream.
By the time we do talk about the system itself, the scope is usually narrower than the client expected — and more precisely aimed at the thing that's actually costing them time.