This document discusses an approach for detecting performance anti-patterns in applications developed using Object-Relational Mapping (ORM). It presents a framework that can detect and rank performance anti-patterns based on their expected impact. As an example, it describes how the framework can detect an excessive data anti-pattern where ORM configurations eagerly retrieve data from the database that is never used. Repeated measurements are used to quantify the actual performance impact of anti-patterns by fixing the issues. The framework was evaluated on several open-source systems where it identified hundreds of potential excessive data anti-patterns.