A regional operations director reviews Saturday's numbers: 4,200 visitors, 9.8% conversion. She cuts two Saturday shifts at that store because conversion looks healthy and traffic looks manageable. The problem: the counter over the main entrance was over-counting by roughly 15% because staff crossed the threshold constantly and a promotional banner created shadow triggers. Real traffic was closer to 3,650, real conversion above 11%, and the store was actually understaffed at peak. She just made a rational decision on irrational data — and the cost of that decision will never appear on any report as a footfall error. It will show up as lost sales, and someone will blame the store team.
That is the defining feature of inaccurate footfall data: the cost is real, recurring and almost always misattributed. Below is where it actually lands on the P&L.
Conversion rate is transactions divided by visitors. Transactions come from the POS and are essentially exact. Visitors come from counters, so every percentage point of counting error passes straight into conversion — in the opposite direction. Over-counting deflates conversion; under-counting inflates it.
This matters most when you compare stores. If Store A counts at 97% accuracy and Store B at 84% (a common gap when hardware generations, mounting heights or entrance layouts differ), your conversion league table is fiction. Bonuses get paid to the wrong managers. "Underperforming" stores get intervention visits they never needed. One retail director put it plainly after an audit: the store they had been holding up as the conversion benchmark was simply the one with the worst under-counting.
The financial consequence compounds. If conversion looks like 12% but is really 10%, headquarters concludes the traffic-to-sales machine works and pushes budget into driving more footfall through marketing. The actual problem — in-store conversion — goes unfunded.
Most scheduling tools, and most experienced managers, build rosters from historical traffic curves. If the hourly curve is distorted, you get two expensive failure modes:
For a CFO, this is the fastest place to quantify the damage. Take one store, estimate the labour hours allocated against phantom traffic, multiply by loaded wage cost, multiply across the estate. In chains with 50+ locations, the annual figure is usually large enough to fund the entire counting infrastructure several times over.
Here is the observation that separates people who have actually run counting programmes from people who have read about them: in a mid-sized store, employees can account for 10–20% of all entrance crossings. Stock runs, smoke breaks, till changeovers, back-of-house trips through the front. Every one of those crossings inflates footfall and drags reported conversion down. Older systems ignore this entirely. Modern platforms — Vemco's among them — use staff-exclusion algorithms, typically identifying employees by tags or patterns, and strip them from the count. Retailers switching on staff exclusion for the first time are routinely startled: conversion jumps two or three points overnight, and years of historical benchmarks turn out to have been measuring the stockroom rota as much as the customers.
Footfall data increasingly sits inside lease conversations — turnover rents, break-clause arguments, expansion decisions. If your traffic data is soft, you negotiate from weakness. A landlord's counter says one thing, yours says another, and neither party can prove calibration. Worse, portfolio decisions get made on the numbers: a location that appears to have declining traffic (but actually has a degrading sensor) can end up on the closure shortlist. Closing a viable store because of a miscalibrated device is the single most expensive footfall error available, and it has happened.
Be suspicious of anyone quoting a flat, guaranteed accuracy figure. Real-world accuracy depends on lighting, entrance layout and how visitors behave — groups, children, trolleys, revolving doors all interact with sensor performance. The honest framing, and the one Vemco contracts on, is a contractual minimum of 96%, with 98–99% typically achieved when conditions allow. Modern 3D AI sensors such as Xovis handle occlusion and group separation far better than legacy beam or thermal units, and can distinguish children from adults — which matters if your conversion denominator should reflect buying-capable visitors.
Two practical points for anyone evaluating their current setup:
You do not need a perfect model to justify fixing this. A defensible back-of-envelope: estimate your counting error rate (audit two stores), apply it to labour hours scheduled against traffic, add the marketing spend allocated on distorted conversion figures, and add a risk line for any lease or closure decision currently resting on traffic data. For most multi-site retailers the first two lines alone exceed the cost of accurate counting within the first year. The third line is where careers are protected.
Inaccurate footfall data never sends an invoice. It just quietly reroutes wages, marketing budget and capital toward the wrong stores, the wrong hours and the wrong conclusions — and lets your best-measured competitor make better decisions with the same information you thought you had.
Want to know what your counting error is actually costing you? Vemco's team can audit your current footfall accuracy, benchmark it against contractual standards and show you where the distortion sits in your conversion and scheduling data. Get in touch for a data accuracy review — bring your worst-performing store's numbers, and let's find out whether the store or the sensor is the problem.