How it works

What our numbers are, where they come from, and what they can and cannot tell you — so you can defend the decision they support.

Measured vs modeled

We show two different kinds of number, and we never blur them.

Measured — real counts from deployed sensors (public and open-data pedestrian counters, our own video studies). Each figure is tagged with its source, timestamp and confidence. Measured data is free.
Modeled — an estimate produced by our model where no sensor exists. This is the paid, per-point analysis: an absolute footfall estimate with confidence and the drivers behind it (POI mix, street class, weather, transit, competition), plus sun/shade at the exact point.

How the modeled estimate is built

An OpenStreetMap POI census around the point (retail, food, nightlife, transit, offices, tourism) sets the local context.
The street’s class and structure, matched to a calibrated archetype, gives a base volume and a 24-hour shape.
Weather (Open-Meteo), transit proximity and nearby competition adjust the estimate for the specific day.
Every output number carries a tier — measured, calibrated, or uncalibrated — and, where the model is uncalibrated, an honest range instead of false precision. Where public data cannot answer, we refuse rather than fabricate.

The free / paid line

Free: the street preview, measured real-sensor data, the typical-week pattern (relative shape only — no absolute numbers), availability links out to the local property portals (we link you to the portals; we don't list units ourselves, and their searches are city- or postcode-level, not street-level), the “possibly vacant” overlay (units mapped as vacant on OpenStreetMap — evidence, not a listing, and often stale), this methodology page, and re-opening anything you already own. Paid: opening a point (the modeled analysis), the assembled Street Report, and video/camera enrichment. A point you have opened is saved to “My analyses” and never charged again while it is still valid.

The honest limit

A modeled figure is an estimate produced by our model, not a guarantee. It is decision-support material, not a promise of commercial outcomes. When you need ground truth, add video or camera enrichment at the point to turn the estimate into a measured count.