Key Takeaways
- Connected vehicle GPS data often stops in parking lots, limiting direct POI visit attribution.
- Parking lot polygons provide the missing spatial link between vehicles and the POIs they serve.
- Combining connected car data with parking lot and POI geometry enables more accurate visitation models.
- Parking lot–to-POI relationships are especially critical for malls, strip centers, and multi-tenant locations.
- Polygon-based attribution reduces guesswork compared to radius- or centroid-based methods.
According to data from Statista, there were about 84 million connected cars on the roads in the United States in 2021. That number is expected to surpass 305 million by 2035, making the United States the biggest market for connected vehicles. More connected cars on the road means that connected car and managed fleet data are becoming more abundant. Businesses have already started using data from connected vehicles to optimize their fleet logistics, analyze traffic patterns, and understand consumer and fleet visitation behavior at POIs (visit attribution).
Using connected vehicle data alone for visit attribution presents a unique challenge: understanding where consumers have gone after they have parked their cars. While location data from mobile devices can typically be seen within a POI’s geofence (since people carry their phones with them), GPS data from connected vehicles usually ends well outside the POI in a nearby parking lot. Without accurate data on parking lots and the POIs those parking lots are serving, visit attribution with connected car data is guesswork. However, by combining connected car data with precise parking lot polygons and POI data, data scientists can create robust POI visitation models.
SafeGraph now offers precise polygon data on parking lots for this use case as a part of our Geometry dataset. SafeGraph Parking Lots matches parking lot polygons with the POIs they are serving, helping you understand what POIs a consumer visited based on where they parked their car.
SafeGraph Parking Lots
SafeGraph Parking Lots currently contains the parking lot(s) serving over 6M Places. These are available as a collection of premium geometry rows depicting the shape and size of surface parking lots across the US.
When combined with connected vehicle data, the SafeGraph Parking Lots and Places datasets can help you understand visitation behavior at POIs. For example, a small strip mall may consist of a number of mall tenants and a parking lot. These tenants and their parking lot would be linked in our data, so you would know that any connected vehicle that stopped in the parking lot was a patron at the strip mall.
Precise polygon data on parking lots helps complete the picture of a driver’s journey. SafeGraph’s Parking Lot dataset provides context on the relationship between a parking lot and the POIs it is serving. Combined with connected car data, parking lot polygons make it possible to broaden visit attribution models to specific places.
Learn more about SafeGraph Parking Lots by visiting our site and download a free sample to experiment with the data yourself.
FAQ’s
1. Why is connected vehicle data alone insufficient for POI visit attribution?
Connected vehicle GPS data typically ends when a vehicle parks, often outside a POI’s geofence, making it unclear which place was actually visited.
2. How do parking lot polygons improve attribution accuracy?
Parking lot polygons capture where vehicles stop and link those locations to the POIs they serve, enabling more reliable inference of visits.
3. How is this different from mobile device location data?
Mobile device data often enters a POI’s geofence because people carry phones indoors, whereas vehicle GPS data usually does not.
4. Which POI types benefit most from parking lot–based attribution?
Multi-tenant locations such as strip malls, shopping centers, and large retail complexes benefit the most.
5. Can parking lot polygons be used with other mobility datasets?
Yes. Parking lot geometry can complement multiple mobility sources where final destination ambiguity exists.