In emerging Internet-of-Things (IoT) realities, sensor nodes near roads will leverage opportunistic connections to vehicles to forward their data to the cloud. In planning such IoT platforms, node placement must be informed by an assessment of infrastructure-to-vehicle (I2V) transfers at a tentative location. It is not feasible to measure I2V volumes at all potential locations, so predictive models are necessary. We propose that qualitative characteristics of a potential site, in particular the existence of traffic and fleet-related points-of-interest (POI), can inform about the vehicles’ mobility patterns and can ultimately be related with the quality of I2V service. In this work, we analyze a real-world dataset of WiFi I2V link measurements in an urban setting with an urban fleet. We observe that most connections occur with the vehicles stopped, and show that stopping regions are related with POIs. Our conclusions include that traffic lights and fleet POIs account for a considerable part of the collected I2V samples, whereas crosswalks account for few transfer occasions.