FAQs about car passage and footfall data

In this article, you will find various questions and their answers that you may have about our data sources regarding footfall and car passage.

What is the timeframe of the data?

• The footfall & car passage set are based on 12 months data. We often add new data if it becomes available.
• Footfall & car passage predictions are on weekly basis.

What is the source of the footfall data?

The footfall & car passage dataset is based on a wide selection of mobile app data. We get our mobile app data from different providers. Our providers collect mobile location data via an SDK anonymously (Mobile app publishers). The selection of apps is very broad and contains apps of various types (weather, traffic, TV guide, entertainment, etc.), ensuring that we do not have over  or under representation of certain profiles.

How do we translate raw mobile app data to a footfall value for road segment?

This paragraph describes the processing of mobile data through an advanced algorithmic developed by RetailSonar. Here are the key steps:

• Data Filtering: High-quality data points are retained by evaluating sampling frequency, GPS accuracy, and outliers. Devices with movement typically provide around 100 measurements daily, covering about 15 hours.
• Population Extrapolation: Weights are assigned to devices based on ZIP-code penetration rates to represent the entire population and account for local variations. If a Zipcode has 10% penetration, each device from that area is weighted accordingly.
• Data Aggregation: Weighted trajectory data is aggregated to create passage maps, detailing movement patterns on road segments.
• Validation and Calibration: Car passage statistics are validated against induction loop data (90% accuracy). Footfall and visitor data are verified using the RetailSonar Activity Map, which estimates annual visits to points of interest using various proxies (e.g., hospital beds, retail area).
• Mode Classification: Movement data is classified into "stay" or "move" episodes. For movement, transport modes (car or bike/walk) are identified using attributes like speed and distance.
• Trajectory Mapping: Discrete location data is converted into continuous trajectories via map-matching and interpolation, considering the transport mode.

The datapacks can be consulted in the platform via our customers. They provide us valuable user feedback as well to increase the data quality and acknowledge the large added value compared to manual counting.

Why do not all road segments get a footfall or car value?

We only predict footfall within shopping areas. If a road segment is not located within a shopping area, we will not show a footfall value. Typically footfall below 2000 passants/week is not included in the map.

For car passage; if the vehicles/ week is below a threshold value it also won't be indicated as a main road segment on the map.

Why is a year-over-year comparison not available for car passage and footfall figures?

RetailSonar's car passage and footfall figures are produced by proprietary models that combine multiple data sources and are continuously refined, with periodic updates to the algorithms, data inputs, and calibration. This refinement is what allows accuracy to improve over time, but it also means the methodology behind a given figure evolves from one year to the next.

As a result, a current-year and prior-year figure for the same location cannot be directly compared: the difference between them would reflect both any actual change in traffic or footfall and the effect of the methodology having evolved, and the two cannot be separated after the fact. Each figure remains calibrated and validated against real-world data at the time it is produced. As a result, the current methodology does not support a reliable year-over-year comparison for a given location.