Not which reason resolves worst — that is set in the reason book, and reporting it back would be circular. This is what the failures cost: every call-back is a contact handled twice, and those hours come off the same roster.
Hours per FCR point is what one percentage point of first-contact resolution on that reason would be worth across this window, in agent hours. It is the number to aim a training budget at.
Scored inside the complexity band. Complex reasons resolve far less often than simple ones by design, so a raw comparison measures the work rather than the person. Each agent is judged against the median of their own band, using a robust z-score that the outliers themselves cannot inflate.
The gap between these medians is the entire reason the comparison has to happen inside a band.
Emergent, not configured: a minority of customers generate a disproportionate share of contacts. This is the shape a bank would size a relationship-management tier against.
A Gini of 0 would mean every customer calls equally often; 1 would mean a single customer makes every call.
Week on week, tested rather than eyeballed. Reason shares wobble on thousands of calls a day, and a half-point move is noise.