Analyzing pedestrian dynamics has since long been an active and practical field of interest. Since the introduction of, in particular, smartphones, various organizations saw a simple means for automatically measuring pedestrian dynamics. The basic idea is simple: network packets sent by WiFi-enabled devices can be collected by sensors and by extracting the unique MAC address from each packet, it should be possible to count how many devices are detected by a single sensor, as well as how devices move between sensors. Although this approach has been commercially deployed for many years, it is now largely forbidden (at least in the EU) due to obvious privacy infringements. In this paper, we address challenges and some potential solutions to automated measurement of pedestrian movements while protecting privacy. The results come from learning the hard way: having run experiments extensively over the past years, we have gradually gained considerable insight in what is possible and what may lie ahead.