Confidence is derived from evidence quality—not assigned as a fixed marketing score. BillzWise evaluates sample size, independent contributors, provider diversity, and recency.
Core factors
Size · diversity · recency
Matching
Item · unit · context · attributes
Weak evidence
Explicit reasons, not hidden
Before confidence can support a fairness verdict, a cohort needs at least three qualifying observations, two distinct bills, and two independent contributors. Falling below any minimum produces an insufficient-data result.
More qualifying observations, distinct bills, independent contributors, and represented providers improve confidence. A cohort concentrated at one provider receives an explicit reduction reason.
Only the most recent 90 days are eligible. If the newest observation is more than 30 days old, the result explains that recency weakened confidence.
Results identify whether comparison used a canonical item, normalised unit, provider or area context, and relevant diagnostic attributes. This helps users understand what was—and was not—held constant.
A cohort may clear the minimum while still having limited evidence quality. BillzWise uses cautious copy, shows the reduction reasons, and does not charge for a low-confidence actionable insight.
