Footfall data in Marketing Mix Modeling (MMM)
How to provide store visitor counts as the business outcome signal for Sellforte modeling.
Sellforte can use footfall — the count of people entering a physical location — as the business outcome signal for modeling. This makes sense when driving visits is the primary goal of marketing, for example in furniture retail, grocery, DIY, cinemas, quick-service restaurants, and any other network where the store visit itself is the outcome you want to influence.
Footfall can be modelled alongside sales or on its own. When the goal is understanding what drives people through the door, footfall gives a cleaner and earlier signal than revenue, since a visit is registered the moment someone enters, independent of whether they buy anything on that trip.
Dataset definition
Daily visitor counts per physical location, produced by an in-store people counter, a third-party footfall provider, or an equivalent source, used as the business outcome signal for modeling.
Because footfall data varies substantially by source system, Sellforte does not prescribe a fixed schema. You provide the counts in whatever format your counting system produces, and the description below covers the information each row should carry rather than a fixed set of column names.
Grain & sample
Each row represents the footfall count for one date, one location, and (optionally) one visitor segment such as entrance, hour band, or customer type.
Example rows (illustrative):
| date | location id | location name | country | store format | open flag | footfall |
|---|---|---|---|---|---|---|
| 2025-01-01 | 1042 | Berlin Mitte | DE | flagship | 0 | 0 |
| 2025-01-02 | 1042 | Berlin Mitte | DE | flagship | 1 | 1834 |
| 2025-01-02 | 1108 | München Nord | DE | standard | 1 | 1256 |
| 2025-01-02 | 1211 | Hamburg Hafen | DE | mall | 1 | 942 |
Dimensions
- date
- location id
- location name (for readability)
- country
- region or city (if you model by geography)
- store format (flagship, standard, mall, outlet, or your equivalent)
- open flag (1 if the location was open that day, 0 if closed)
Metrics
- footfall (the visitor count for the row)
- transactions (optional, if the same counting system also captures purchases)
- conversion rate (optional, transactions divided by footfall)
- dwell time (optional, average minutes on-site)
Closed days and network changes
Please include closed days explicitly with footfall set to 0 and the open flag set to 0. This lets Sellforte distinguish a real zero from missing data. Similarly, when stores open, close, or relocate during the modeling window, make sure the location table reflects those events so the timeseries can be interpreted correctly.
Counting method note
Because footfall counts depend on the counting system in use — door sensor, camera-based counter, WiFi probe, or a third-party provider such as Sensormatic, RetailNext, or a mobile-location panel — please document which method produced the data and flag any mid-period changes such as sensor upgrades, recalibrations, or provider switches. Different methods can produce systematically different counts, and knowing when the method changed helps Sellforte read a step change in the series as a measurement artifact rather than a real change in visits.