cx-redress-consistencylisted

# Redress consistency Two customers, same failure, same circumstances, different outcomes. It is one of the few CX findings that is indefensible on its face — and unlike most conduct questions, it is measurable from data most firms already hold. This audits **outcomes for consistency**. Designing the framework that produces the outcomes is a different job; without a framework, expect the variation to be large and the finding to be that one is needed. ## Compare like with like, or find nothing Raw variation in redress amounts is meaningless — cases genuinely differ. The audit only works once you have defined comparable groups. Build the comparison on: - **The failure type**, at the level of cause rather than category. - **The loss** — the actual detriment, where recorded. Two customers with the same failure and different losses should get different amounts. - **Whether redress was owed or discretionary.** These follow different rules and pooling them destroys the analysis. Owed redress should show *very* little variation for a given loss; goodwill legitimately shows more. - **Product, market and period**, since policy changes over time. Then look for variation *within* those groups. Between-group variation is expected; within-group variation is the finding. **If the data does not let you group by loss, say so.** Without it you can still detect gross disparities and cannot distinguish justified variation from unjustified. That is a useful, limited result — report