It has long been known that the length of a given section of coastline increases with increasing resolution of measurement. This is termed the self-similarity property of a shoreline. Similarly, there exist multiple spatial and temporal scales of geomorphic activity on coasts. The given spatial scale at which a vectorized line delineates the coastline is therefore inherently attached to the geomorphic processes which operate within this spatial scale. Choosing a specific coastline dataset to investigate rates of erosion and erosion outputs means that a geomorphic process is preferentially emphasized which can prove problematic in systems operating at multiple scales such as the coastal one. It also means that erosion outputs (i.e. volumes), which are generally computed using the length of the coastline, are also tacitly associated with a specific spatial scale, that is, with the geomorphic processes taking place at these scales. Any volume loss estimation from the coasts will thus always be relative. It seems nevertheless necessary to provide the scientific community with numbers, especially at the global scale to compare sources/sinks of sediments from different types of environments. We therefore ask the question: Which scale is relevant to the calculation of volumes eroded form the coasts? Is it dependent on their lithological nature? On their morphology? Or is it intrinsically bound to the nature of the dynamics affecting the surfzone and/or the shoreface?We provide an example taken from the Arctic Coastal Dynamics online Geographical Information System, a circum-arctic inventory of coastal environments with specific emphasis on the geomorphological description of the coastal tract and the carbon contents of the subaerial coastal units. Results show that the self-similarity of Arctic coasts is highly dependent on both the type of coastal environment and the nature of the processes affecting the subaqueous part of the coastal profile. Additionally, the results show a substantial change in volume estimation with scale prompting the need for a cautious interpretation of volume losses estimations, especially at the global scale.