As long as you do have two readings, do they agree? And if the value of the measurement is to recognize improvement when it is achieved, isn't comparison more important than methodology?

However, if you are including rest time, there's always the question of whether the rest time was excessive and hence you took away from your average by larding around rather than pedaling.

I always think the "fish stories" of cycling are the length and grade of local hills, and "we didn't go very fast only 21 mph average."