In recent years, there has been an upsurge in intelligent mobility solutions that provide door-to-door services. Although these services offer convenience to certain individuals, it is frequently overlooked that they can lead to welfare losses when accounting for the reduced health benefits that result from reduced physical activity. In this paper, we derive a welfare function of introducing first- and last-mile public transport services. By comparing possible health gains from walking with corresponding accessibility losses, we identify the distance boundaries under which the service fails to be socially beneficial. The results are based on a simulation study and draw on further insights from a recent agent-based model from Copenhagen focusing on first- and last-mile public transport. Although the model is intentionally stylized and may not apply universally to all scenarios featuring diverse population densities, demographic profiles, or transport network layouts, the fundamental conclusion presented in the paper is that first-mile services have minimal welfare impact for average trip distances below 1 km, appears robust even under conservative assumptions. In this case, the probability of failure is almost 100% for any realistic parametrization. This finding implies that planners and researchers should focus on the design of main transit networks and the access and egress of active modes to and from the stations. In particular, door-to-door services covering shorter distances should not be the priority of public funding unless in particular situations or contexts.