How broad or narrow to go with a test hypothesis for comparing old designs with a redesigns

Say I am given an open brief like 'improved the sign-up journey for product X' or 'improve the search experience for product Y' etc.

I want to do some user testing around the current experience so this will be the baseline metrics and when I user test my designed solution I can compare the old with the new.

Is a general test hypothesis like the following too broad?

We believe that by redesigning the sign-up journey we can improve the overall sign up experience We will know this to be true when we increase the success rate, decrease time on task and increase the attitudinal score for happiness (1 - 5) and ease (1-5)?

Or is it better to break down hypotheses like:

We believe that by ? simplifying/reducing the number of steps? in the sign-up journey, we can reduce time on task We will know this to be true when the mean time on task is lower on the redesign

I see a floor in the latter. Your opening hypothesis, on which will form your benchmark tests, is assuming that reducing the number of steps in the sign up journey is the right solution but you've barely begun your discovery phase so maybe this is wrong?