It is logical that the generalization of digital approaches in history is leading to a democratization of the graphic representation of the data produced by these processes. Rather than presenting long series of examples, this very cursory chapter seeks to fuel reflection on our uses: why do we visualize historical data? Is it for illustrative purposes, to "show" our historical object and make it understandable to a large audience? Or is it, on the contrary, because the raw data is unintelligible to us, and visualization is therefore a heuristic tool intended for their exploration? The central point of my argument is based on a typology of sources and uses, a double entry table which is intended as a kind of decision-making aid for those seeking to make their data speak in the right way to the right audience.