How it works
The whole mechanism, stated plainly.
Phones looking for WiFi broadcast short probe frames whether or not they ever connect. We listen to those frames, anonymise them on the sensor, and turn them into the numbers a venue, exhibitor or organiser can act on.
The sensor
Hardware equivalent to a WiFi access point, running our firmware, passive and listen-only — it never transmits. Internal battery, 4G connectivity (local network optional), encrypted local storage for offline operation. Detection range is roughly 10 metres per sensor.
Location
One sensor tells you a device is nearby. A device within range of three or more sensors simultaneously can be placed approximately by triangulation — which is why sensor positions are fixed at deployment and must not move. We call the result perimeter detection, not positioning: sensors define perimeters, and we classify devices inside or outside them.
The classification
Every zone visit resolves into one of three shapes from two measurements — signal strength and time: faraway (weak signal, periphery), passer-by (present, briefly), and engaged (strong signal with at least 30 seconds of dwell — a platform-defined threshold we state rather than hide). Patterns are cohort-level, never individual paths.
Anonymisation
The device identifier in each frame is hashed with a unique random salt on the sensor itself, in 1–3 seconds. Raw identifiers are never stored — on the sensor, in transit, or on the server, which never receives and cannot reconstruct them. The salt is destroyed when your event closes, or at campaign end by agreement; destruction irreversibly eliminates linkability in production systems. The full pipeline is on the privacy page.
Daily cleaning
At the end of each day, devices that behave like fixtures — staff phones, exhibits, infrastructure — are removed, and the final daily report publishes. What you compare day to day is visitors, not furniture.
The HITmap
Hotspots, Interest, Traffic: interest as density around single sensors, traffic as preferred paths between sensor pairs, hotspots where three or more sensors see the same visits. Positions on the map are density functions, not located individuals — it shows where attention concentrates, and it makes zone-to-zone conversion measurable on large stands.
The honesty row
Counts are unique devices, not people. Phones with WiFi off are not seen; some visitors carry two devices; identifier randomisation on modern phones is factored into the statistical model. We publish definitions, not multipliers.
The pipeline