Abstract: Technologies are provided for generation of quality scores for brand curation. A quality score represents a quality assessment of a brand. To generate quality scores, a machine-learning (ML) model can be trained to classify brands as either pertaining or not-pertaining to a group of select brands. The ML model can be trained using the group of select brands as a training set and performance signals for those brands as feature inputs. A select brand can be an expert curated brand, a brand carried by a particular merchant, or a brand identified using social media. A performance signal defines a value of a quality metric of a brand, where the quality metric defines an aspect of interaction between a customer and the brand within a digital marketplace. A quality score of a brand can be generated by applying the trained ML model to performance signals for the brand.