@jerry you need to plot all accounts in a formal statistical
#ControlChart, in which you determine via statistical analysis the upper & lower control limits, which for instance you might set at mean aura + three std devs for the UCL, & ofc mean aura - three std devs for the LCL. then as the data keeps accruing over time, you note those accounts tending dangerously near the UCL or LCL, but you only need take official action once an account's data line has exceeded the UCL or LCL for seven consecutive samples, as that denotes statistically significant system change in that account. once identified, urgent decisive action is required by you, lest other accounts also get uppity [or lowerly].