Scoring

Make The Calculation Visible Before Trusting The Result

The Weighted Score Assembly spreads each criterion as a transparent numerical segment — Weight, Observed Result, Threshold Result, Evidence Coverage, and Confidence Adjustment. Insufficient evidence is marked Unproven (never zero). Adjust a weight and the raw score, confidence-adjusted score, and domination risk update immediately.

Completion Rate35%

Handling Time30%

Escalation Count20%

Allowed Exceptions15%

Completion RateEscalations agents close without opening an engineering ticket

Weight %

Observed Result63%

Threshold ResultAbove Minimum

Evidence Coverage88%

Confidence Adj.+0

Score74

Contribution25.9

Handling TimeMedian minutes from escalation open to resolution

Weight %

Observed Result26 min

Threshold ResultAbove Minimum

Evidence Coverage76%

Confidence Adj.-4

Score70

Contribution21.0

Escalation CountWeekly escalation volume the workflow handles without backlog

Weight %

Observed Result128/wk

Threshold ResultAbove Minimum

Evidence Coverage52%

Confidence Adj.-8

Score56

Contribution11.2

Allowed ExceptionsEscalation types explicitly out of scope for this pilot Unproven

Weight %

Observed ResultIT reviewed · Security pending

Threshold ResultUnproven

Evidence Coverage34%

Confidence Adj.-12

Score—

Contribution0

Assembled score

Raw score68

Confidence-adjusted62

Unrated / Unproven15%

Watch domination risk — Completion Rate at 35% weight can swing the call alone if Coverage stays thin.

Weights must total 100% before a scoring model can publish. Unproven segments lower confidence instead of writing a zero.

Publish Scoring Model

Saves weights, formulas, missing-data handling, and model version. Every later score references this published version.