In this paper, we propose a general framework for creating such an interpretable composite quality metric. The components or individual KPIs are discretized on a Fibonacci scale 8, 5, 3, 2, 1 such that a score of 8 can be interpreted as the best, 3 as bad and 1 as the worst experience. The weights for the components are selected via numerical methods such that the correlation with the bottom-line KPI is measured for each weight combination. The weight combination with the best correlation and non-skewed distribution can be selected as optimal. The result then is a single composite metric constructed from multiple components that is predictive of a bottom-line KPI.