The analysis and processing of learning traces through Learning Analytics, although essential to help improve learning, requires addressing strong ethical issues. Tools provided to various educational stakeholders must ensure fair learning, and take into account the diversity of students and their behavior while avoiding bias and discrimination. In that context, descriptive cards are an emerging tool allowing for a representative description of the diversity of behaviors adopted within a class. However, their design raises a lot of questions about equity and fairness. To contribute to the need for fair learning, we present a study conducted in collaboration with teachers and propose a new design of descriptive cards, limiting as much as possible the introduction of bias.