Abstract This article aims to improve our understanding of scientometric data in a Benfordian context. Recently, Benford’s law has been used to detect scientific fraud. However, we need to better understand its application to scientometric data. Through the implementation of Benford’s law and the generalized Benford’s law, we propose a categorization of science products and metrics. To this end, we have performed chi-square, MAD, and Max tests on data sets from WoS and Scopus as well as on historical data. This enables us to better understand the behavior and characteristics of these objects in a Benfordian context, and invites us to discuss the nature of bibliometric indicators in this particular context.