Every time someone proposes a fix to a flow or UI... you should be running a benchmark test to know A) How bad is this flow, really? and B) When you change the flow... did it improve, get worse, or stay the same?
These tests are VERY cheap to do—often you can do them in a hallway with non-engineering staff to get a general sense of issues... and then bring in users to verify this. It's honestly the #1 way you measure design success.
Hi there Andrew,
These are very good points you raise! I agree, this type of testing doesn't cost anything but it really is worth it.