A retail AI case study: 96% accurate people counting, a counterintuitive discovery that changed staffing plans, and what it means for any brand making decisions based on gut feel.
The marketing team at a leading Indian EV brand had a theory. Weekends drive showroom traffic. Friday evenings are the peak. Staff up on Fridays and Saturdays. Run your big campaigns for the weekend. It made intuitive sense — people research big purchases on weekends when they have time. Every retail brand in India operates on some version of this assumption.
The problem: the assumption was wrong. Not slightly wrong. Significantly wrong. At one of their flagship showrooms, Tuesdays were consistently outperforming Fridays in qualified visitor traffic — the kind of traffic that converts to test drives and purchases. The brand had been understaffing on Tuesdays and overstaffing on Fridays for years. The cost of this misallocation, across a network of showrooms, was substantial.
They only discovered this because they installed AI footfall analytics.
Before explaining what the EV brand discovered, it is worth clarifying what AI footfall analytics is — and what it is not. This matters because the most common objection to footfall analytics is privacy: "Are you filming our customers?"
The system Swaran Soft deployed uses 3D depth sensors, not cameras. A 3D sensor measures the volume of objects in its field of view. It detects that a human-shaped volume of approximately 0.3 cubic metres has passed through the entrance. It counts that as one person. It does not capture an image. It does not record a face. It cannot identify who the person is. There is no personally identifiable information collected at any point.
This distinction is critical for two reasons. First, it means the system is GDPR compliant and meets India's DPDP Act 2023 requirements without any special consent mechanisms. Second, it means the system achieves 96%+ accuracy — significantly higher than camera-based systems — because it is not affected by lighting conditions, clothing colours, or the angle at which a person enters.
The demographic analysis layer (age and gender estimation) uses on-device processing with aggregated, anonymised outputs. The system can tell you that 60% of your Tuesday visitors are in the 25–35 age bracket. It cannot tell you who any of those individuals are.
The EV brand's pilot started with five showrooms across two cities. Within the first four weeks, the data revealed patterns that contradicted every assumption the marketing and operations teams had made.
The Tuesday effect was the most striking. At the flagship showroom, Tuesday afternoon traffic between 3 PM and 6 PM was consistently 23% higher than Friday evening traffic in the same time window. More importantly, the conversion rate (visitors who spent more than 15 minutes in the showroom — a proxy for serious interest) was 31% higher on Tuesdays than Fridays.
The hypothesis the team developed: Tuesday afternoon visitors are typically professionals who have taken time off specifically to visit the showroom. They have already done their research online. They come with intent. Friday evening visitors are more likely to be browsing — stopping by after work without a specific purchase intention. Higher volume, lower intent.
This single insight changed the staffing plan for that showroom. Senior sales staff — the ones with the highest conversion rates — were shifted to Tuesday afternoon coverage. The impact on conversion was measurable within 30 days.
"We had been running our showroom operations on assumptions that felt right but had never been tested. The data showed us that our most valuable customers were coming on Tuesday afternoons, and we had been sending our junior staff to handle them."
— Head of Retail Operations, Leading Indian EV Brand
The Tuesday effect was just the beginning. Over 90 days, the footfall analytics system surfaced five additional insights that changed how the brand operated its showroom network:
| Value Driver | Mechanism | Estimated Impact |
|---|---|---|
| Staff optimisation | Right staff on right days | 15% labour cost reduction |
| Conversion improvement | High-intent visitor identification | 8–12% conversion uplift |
| Campaign ROI | Footfall attribution to campaigns | 20% better campaign spend allocation |
| Network management | Outlier identification and correction | Underperforming stores identified |
Every retail brand in India is making staffing, marketing, and expansion decisions based on assumptions. Some of those assumptions are correct. Many are not. The ones that are wrong are costing money every day — in misallocated staff, in campaigns that do not drive store visits, in expansion decisions made without data.
The question is not whether your assumptions are wrong. Some of them definitely are. The question is whether you want to know which ones — and how much it is costing you to be wrong about them.