Measuring foot traffic at a shopping center involves distinguishing between two metrics that should never be confused: the center’s metrics, which count entries into the mall and foot traffic between sections, and a retailer’s metrics, which count entries into its own store. The former is the responsibility of the mall operator, the latter of each tenant, and the two are most often based on different systems.
This is why the figures reported by a mall and those recorded by a retailer almost never match. Understanding where the discrepancy comes from is more important than trying to eliminate it: the discrepancy is normal; it is its magnitude and consistency that provide insight.
The mall operator measures entries by entrance, foot traffic in the concourse, the distribution across floors and sections, and, depending on the location, parking lot occupancy. These metrics are used to manage operating hours, cleaning, security, and the leasing of retail spaces.
Retailers measure foot traffic into their sales areas, foot traffic by interior zone, and cross-reference these figures with their sales receipts to calculate a conversion rate. These metrics are used to determine staffing levels and evaluate a store’s performance.
Between these two lies a metric that is rarely utilized: foot traffic past the storefront, also known as storefront traffic. It is of interest to both parties because it distinguishes between the attractiveness of a location and the attractiveness of the store itself.
Four metrics cover the essentials of operational decision-making.
These metrics are derived from cameras already installed in the mall, provided that the camera views cover the desired counting lines.
For a tenant, three metrics are sufficient for decision-making. Store entries, by time slot, which inform scheduling. The capture rate—that is, the percentage of passersby in front of the storefront who enter—which evaluates the window display and signage. Time spent in each interior zone, which evaluates the store layout.
The conversion rate—the ratio of sales receipts to visitors—is meaningful only if staff are excluded from the visitor count. Without this exclusion, stores with a high staff density automatically show a lower rate, without any commercial difference to explain it.
Four factors combine, and none of them is a measurement error.
First, the scope: a person who enters the mall does not enter every store, which makes the mall’s figure structurally higher.
Next, the counting point: a line placed at the store’s threshold and a line placed past the entrance area do not count the same thing, because people who turn back at the entrance are not counted in the latter case.
The counting convention also plays a role: some systems count crossings, others count visits, with a rule that merges consecutive entries by the same person. These two conventions produce significant discrepancies in a store with many exits and returns.
Finally, there is the issue of staff handling, which is almost never standardized between the mall and its tenants.
The practical consequence is that a comparison is only valid when based on a homogeneous data set. Comparing foot traffic for one month to that of the previous month—using the same system and the same counting convention—is reliable. Comparing the mall’s figures to those of a single store at a given point in time is not.
Three precautions are sufficient in most cases. Document the exact position of each counting line and do not move it without recording the date of the change. Establish a single convention for distinguishing between crossings and visits, and document it. Explicitly address secondary entrances, parking lot exits, and connections between levels, which are the primary source of double counting in a shopping center.
The fourth precaution concerns historical data: a change in the convention or the counting line’s position creates a break in the data series. It’s better to flag this in the dashboards than to let a team interpret a technical variation as a commercial effect.
A shopping mall is a place open to the public. Video surveillance there is governed by the Internal Security Code: prior authorization from the prefect, public notices informing visitors, and limited retention of footage—a maximum of one month unless an investigation is underway.
Automated foot traffic analysis adds a specific purpose to the system and therefore requires data processing to be documented under the General Data Protection Regulation (GDPR): legal basis, information provided to data subjects, data minimization, retention period for metrics, and entry in the data processing register. An impact assessment is required since the system amounts to large-scale, systematic surveillance of a publicly accessible area.
The National Commission for Information Technology and Civil Liberties (CNIL) distinguishes between uses for statistical analysis and uses leading to an individual decision. An aggregated foot traffic measurement—without facial recognition and without re-identification across cameras—falls under the former category. Finally, as soon as employees are working in the field, the Labor Code requires that employee representative bodies be informed and consulted.
Five written responses are needed for a meaningful comparison: the definition of each metric, the distinction between crossings and visits, the treatment of staff and the technology used, the data retained and its retention period, and the location of the servers.
CORE, the platform developed by XXII, reads the video feeds from a center’s or store’s existing cameras and produces aggregated metrics on foot traffic, density, and dwell time—without facial recognition. For both a facility operator and a retailer, the value of a foot traffic figure depends first and foremost on the consistency of its calculation methodology—and this methodology must be clarified before signing the contract, not after.