The document discusses applying the 80/20 rule to optimize a loyalty miles program. It describes a case study of an airline using data from its 15 million customer loyalty program to forecast customer tier levels in real-time and optimize resource allocation. The airline was able to improve its linear forecasts of customer tier levels from 42% to 47% accuracy by using various machine learning techniques like generalized boosting machines and generalized linear models on clustered customer data, with a focus on the 80/20 rule to optimize non-value adding activities and resources. It concludes that opportunities exist in applying the 80/20 rule to identify and reduce waste, and that properly defining the problem question is key for success with big data projects.