The most common question retail operations teams ask is "how many people walked through". It's the wrong question. The right question is "for how long, in front of what, and what got in their way". Footfall is a denominator. Dwell is the story. We don't have a retail pilot running yet. This post is about how we'd run the first one.
01 · WHAT A SINGLE-STORE PILOT WOULD COVERWhat a single-store pilot would cover
The smallest useful retail pilot is one store. Two entrances, a window display, four numbered aisles and a till bank, each instrumented with a single edge sensor. Counts are anonymous, dwell is measured per zone, and till queue time is measured by zone-entry-to-zone-exit. No cameras stream off the device; no faces are recognised.
After four weeks of trailing data, the report would surface:
- Median dwell per aisle, with delta against the store average, the merchandising lever.
- Till wait by hour of day, day of week, the operations lever.
- Window-display attention curve, the visual-merchandising lever.
None of those numbers require knowing who anyone is. All of them are decisions a merchandiser or store manager can act on.
02 · THE TILL BOTTLENECK A CONVERSION REPORT MISSESThe till bottleneck a conversion report misses
The conversion report a head of stores receives every Monday flattens the day to a number. Dwell numbers expose the texture. A store can be in front-of-house excellence and back-of-house failure simultaneously: full aisles, considered baskets, four minutes at the till. The most likely outcome at four-minute till wait is a basket put down, not a basket bought.
Dwell lets the conversation move from "did the campaign work" to "did the till staffing match the wave". One is a marketing question. The other is a roster question. The data points at which.
03 · TWO PATTERNS WE'D EXPECT TO FINDTwo patterns we'd expect to find
A few patterns are common enough across the literature that we'd expect to find them on any first deployment:
- End-cap dwell beats centre-aisle dwell. Window-display end-caps consistently outperform mid-aisle merchandising for attention. Quantifying the gap is what dwell adds.
- Off-peak dwell is the better merchandising signal. Peak hours are noisy, people are bumping through. Off-peak dwell exposes which aisles earn considered attention.
04 · THE PRIVACY ANSWER FOR RETAILThe privacy answer for retail
Retail is the industry where the privacy answer matters most for the brand. Shoppers carry their phones and their assumptions through the door. The platform's answer is the same as in every deployment: the model runs on the sensor, frames never leave, counts are anonymous. There is no face matched to a basket, no demographic inferred from a dwell, no PII produced anywhere along the pipeline. A retailer who measures NPS can deploy occupancy analytics without spending any of it.
05 · WE'RE LOOKING FOR THE FIRST RETAIL PILOTWe're looking for the first retail pilot
If your store has been measured on footfall for a decade and nobody has ever shown you the dwell, talk to us about a pilot. Single-store pilots welcome.