Measuring understanding is notoriously difficult. Indeed, in formulating learning outcomes the word 'understanding' is usually avoided, but in the sciences, developing understanding is one of the main aims of instruction. Scientific knowledge is factual, having been tested against empirical observation and experimentation, but knowledge of facts alone is not enough. There are also models and theories containing complex ideas and interrelationships that must be understood, and considerable attention has been devoted across a range of scientific disciplines to measuring understanding. This case study will focus on one of the main tools employed: the concept inventory and in particular the Force Concept Inventory (FCI). The success of concept inventories in physics has spawned concept inventories in chemistry, biology, astronomy, materials science and maths, to name a few. We focus here on the FCI and ask how useful concept inventories are for evaluating learning gains. Finally, we report on recent work by the authors to extend conceptual testing beyond the multiple-choice format.