It is uses a Recursive Loop. And every round of the loop the digits are stripped based on the current itaration value of the loop. Matches are collected, non matches are input for the next round of the loop.
I tested the flow, and it worked; however, in the final result, it was keeping the options from Table 2 and excluding the not-found entries from Table 1. I made some adjustments to reverse the loop, which allowed me to retain all options from Table 1 while pulling the averages for the first reference found.