Urban public space is often provided for freight delivery operations in the form of on-street (un)loading zones (LZ). Since public space is scarce and demanded by several users, city authorities have the challenge of managing LZ by gaining knowledge about freight curbside needs and utilization. Although technological solutions and enforcement practices have become popular among policymakers to capture curbside dynamics, there is still an open and promising research field for designing analytical frameworks that shape LZ decision-making processes. This fact has motivated the authors to define the concept of Smart Loading Zones (SLZ) as the involvement of technology and data analytics in the planning and management of LZ in a responsive and user-oriented way. Besides proposing a conceptual approach for the study of SLZ, this paper implements data analytics tools for enhancing decisions on LZ network design, using the City of Vic (Spain) as a case study. The machine learning techniques k-means++, DBSCAN, and integer linear programming prescribed the LZ number, location and service assignment based on establishments' coordinates, walking distances and freight demand. Results from the case study showed how an optimized number, location, and size of LZ improved occupation levels, i.e., from 18 % to 80 %, while freeing up curbside space for other users. Service coverage was also improved by allocating LZ to establishments within walking distances no greater than 75 m. Further development of methods and tools for SLZ at tactical and operational decisions are recommended for future studies.