I started by sketching some low-fidelity wireframes in my notebook. I like to use a paper notebook at the ideation stage as it lets my creativity run free without any digital constraints and lets me go through dozens of options before I find the best one! The same story happened here, I sketched out a bunch of different versions of the feature, and then eliminated the unsuccessful ones, keeping in mind John's journey, priorities, and pain points. For example, one of the options contained the feature of global editing the roles of the user, but I felt like it required a change in the overall logic of the interface (introducing a completely new feature instead of creating a way to complete the needed tasks with an already existing interface with some additions), making the learning curve steeper for the user. I also tried out different placements for the "filtering" and the "selecting" buttons, and different ways of their design (e.g. an icon or text).