Problem
A shop owner can see the floor but not the pattern. The owner places staff from experience, without data on when shoppers actually arrive; some sections hold attention much longer than others, and shoppers can stand in a section with nobody nearby while staff are pulled to the till.
Users
The owner and floor managers of the Madhar Sha showroom.
Constraints
- The preview is not yet connected to the showroom's own cameras, so every figure shown is illustrative.
- Until then, the footage and annotations come from public datasets: CARR (Customer Activity Recognition in Retail, CC BY 4.0) and retail_gaze.
- The readers are shop staff with no analytics background.
What I built
A static dashboard with a "yesterday at a glance" summary (visitors, fullest moment, typical stop in a section) and insight cards for the busiest hour, the section that holds shoppers longest, time a section had shoppers but no staff, and billing-counter pressure by hour.
Metrics
No real metrics yet. The preview's figures are illustrative.
Trade-offs I made
- Every number comes with advice.Each one has a sentence saying what to do, such as putting the most knowledgeable staff near the section shoppers linger in.
- Demo figures are labelled.The preview states that its figures are illustrative until the showroom's cameras are connected.
What I'd do next
- Connect the pipeline to the showroom's own cameras so the figures are real.