Change one element at a time. Test a new summary for applications, then measure callback rate.
A reasonable baseline is % (one callback per applications). If you're below that, something needs to change. If above, keep what's working.
##### ###### ##### ### # # ### # # ###### ## ## ## ## ## ## ## # # # # # ## ##### #### ##### # # # # # # # #### ## # ## ## ## ## # # # # # ## ## # ###### ## ### # ### # ######
##### ###### ##### ### # # ### # # ###### ## ## ## ## ## ## ## # # # # # ## ##### #### ##### # # # # # # # #### ## # ## ## ## ## # # # # # ## ## # ###### ## ### # ### # ######
How to systematically test different versions
Change one element at a time. Test a new summary for applications, then measure callback rate.
A reasonable baseline is % (one callback per applications). If you're below that, something needs to change. If above, keep what's working.