Image segmentation is a complex task that aims to simultaneously satisfy various quality criteria. In this context, topology is being increasingly considered. Guaranteeing correct topological properties is indeed crucial for objects presenting challenging (e.g. small, elongated, numerous) shapes. This is especially true in medical imaging. Designing topology-aware metrics is then relevant, both for assessing the quality of segmentation results and for designing losses involved in learning procedures. In this article, we introduce ccDice (connected component Dice), a topological metric that generalises the popular Dice score. By contrast to Dice, that acts at the scale of pixels, ccDice acts at the scale of connected components of the compared objects, thus leading to a topological assessment of their relative structure and embedding. ccDice is a simple, explainable, normalized and low-computational topological metric. We provide a formal definition of ccDice, an algorithmic scheme for computing it, and we assess its behaviour by comparison to other usual topological metrics, thus emphasizing its relevance.