Abstract Professional sport enhances social interaction among athletes through meetings such as training groups, large-scale competitions and sponsored events. Among these, the Olympic Games represent the largest international competition, having a unique global impact. This event is assumed to promote international cooperation through the values of sport, becoming an opportunity for creating global connections between athletes. Here we want to understand the properties of the following activity, an indicator of the information flows, between highly successful Olympic athletes both at microscopic and macroscopic scales. To do so, we built a database including the Twitter usernames of Olympic medallists in Tokyo 2020, using it for creating the follower-followee network, a directed network representing who followed whom in this social network. From our database of 1052 athletes, we found 7326 connections among 964 athletes. The most popular athletes, those with highest number of followers, were Kevin Durant (basketball, USA), Allyson Felix (athletics, USA), Teddy Riner (judo, France), Alex Morgan (football, USA) and Simone Biles (gymnastics, USA). The macro-scale network structure displayed organized patterns related to microscopic—node—properties such as sex, country and sport, evidencing assortative connectivity patterns. These assortative patterns highlighted the need for a model quantifying the features’ importance in the followees choice, which we introduced through a gravity approach modelling the number of connections between homogeneous groups. Our research remarks the importance of datasets built from public exposure of professional athletes, serving as a proxy to investigate interesting aspects of many complex socio-cultural systems at different scales.