How (Not) To Fix Online Dating - An Empirical Assessment Using Computational Experiments

With roughly 40% of new couples meeting online, dating platforms have become commonplace. Yet, they share a common pitfall: a skewed gender ratio, with men substantially outnumbering women. Combined with prevailing gender norms—where men are expected to make the first move—this creates market congestion. Women are overwhelmed by messages and often face harassment, while many men struggle to receive any attention. We use computational experiments to examine how dating-platform designs affect individuals’ mating choices. We replicate popular app designs using a self-built web application and generate artificial profiles by combining images with socio-demographic data. Building on past research, we compare a “control” design with two interventions that reveal profiles’ popularity and capacity. Based on 66.132 profile ratings from participants in four countries, we find that the experimental conditions alone do not significantly shift user choices. We then analyze treatment heterogeneity using our large sample. Results show that visual and textual capacity indicators reduce the number of likes for highly attractive profiles, particularly when rated by men. However, these interventions do not increase engagement with less attractive profiles. These findings suggest that instead of redirecting attention to less attractive users, the intervention mostly benefits those who are already popular.