Indivd
•
•
Intro
A fashion retailer in Central Europe achieved at least 97% absolute counting accuracy in each of its 35 stores across 80 counting lines.
A fashion chain in Central Europe had its people count verified at 35 stores, each against an independent count covering the same periods, between November 2024 and April 2026.
Every location achieved at least 97% absolute accuracy. This is the floor across the set, not an average, based on 13,076 visitors and 1,099 verified trading hours.
A November weekday at a fashion store in Central Europe. One doorway, winter coats and shopping bags crossing in both directions, one camera carrying an entrance line and an exit line. Across the region, thirty-four more stores in the chain run on the same kind of count. The staffing and buying decisions that shape each store depend on what lines like these record.
Between November 2024 and April 2026, all 35 stores went through the same check.
The stores
One doorway carries the whole store’s number
A fashion chain makes decisions by store and by hour. How many colleagues does the Saturday floor need? Does a new window display bring people through the door, or only up to it? Which locations earn investment when the next like-for-like review comes?
Every one of those decisions divides sales by visits. At most of these locations, the visits side of that calculation rests on a single entrance and exit pair. Where the count rests on one pair, there is no second measurement point to dilute an error. The pair is right, or the store’s denominator is wrong.
A fashion store entrance is also a difficult place to count. The doorway is narrow, so groups arrive abreast. Window shoppers hover at the threshold without entering. For five months of the year, heavy outerwear changes every silhouette that crosses.
A count that drifts at one doorway misrepresents the whole store. Accuracy is not something a model simply has. It is something a deployment produces. A deployment that has produced it once has proven one building. Thirty-five buildings are a different claim.
The check
One standard, applied thirty-five times
The chain decided where, when, and how much was tested at each store. Some locations were assessed during winter weeks, while others were assessed during spring and summer periods. The checks covered single days or multi-day windows, depending on each store’s calendar.
Across the programme, 13,076 visitors were included in the comparison over 1,099 verified trading hours. Indivd’s count was compared with an independent count covering the same periods.
The check covered 80 counting lines on 38 cameras. At each location, the complete tested set was assessed together.
The measure is strict. It counts every miss and every overcount in full, with no netting between them. On that measure, the minimum across the 35 locations was 97.00 percent absolute accuracy.
The result is published as a floor, not an average, because the minimum is the only figure that applies to every store at once.
Absolute accuracy is 100 percent minus the absolute error rate. Undercounts and overcounts are both errors. A missed entry, a missed exit, and a counted person who never crossed are each counted in full. Errors in opposite directions are never permitted to cancel one another.
DEFINITION
Absolute accuracy is 100 percent minus the absolute error rate.
Undercounts and overcounts are both errors. A missed entry, a missed exit, and a counted person who never crossed are each counted in full. Errors in opposite directions are never permitted to cancel one another.
What the number carries
Thirty-five stores dividing by tested numbers
Verified between November 2024 and April 2026: at least 97.00 percent absolute accuracy at every one of 35 fashion stores in Central Europe.
This is the minimum across the set, not an average. The programme covered 13,076 visitors, 1,099 verified trading hours, and 80 counting lines on 38 cameras.
What the chain gains is a like-for-like comparison that means something. Hourly conversion at one store can sit against hourly conversion at another because both denominators were tested in the same way, using the same measure.
The Saturday rota, a window change read through its effect on entry rate, and the investment case for one location over another all depend on a count checked against an independent reference during that store’s selected trading periods.
Without a consistent standard, errors can compound quietly. One overcounting pair can make a healthy store appear to have a conversion problem. One undercounting pair can lead to Saturday staffing based on fewer visitors than actually arrived. A chain comparing stores with unequal denominators risks applying the wrong lessons across its estate.
In a separate engagement, KPMG audited Indivd’s counting accuracy at a major retailer’s store against manual ground truth. Every camera tested came within 1.17 percent, the strongest result of any system included in that audit.
What the chain can ask next is what one tested store never could: run one change across the estate and trust every number that moves.
More verifications in this series: a hardware and home-improvement chain in Northern Europe and a home-improvement store verified during its garden season. The full collection is available under Quality assurance stories.
FAQ
What is footfall in retail?
Footfall is the number of people entering a store, counted at the entrance. It is the denominator beneath conversion, sales per visitor and most store performance comparisons, which is why its accuracy determines the reliability of everything built on it.
What should a fashion chain look for in a retail people counting system?
Evidence that accuracy was measured at locations like theirs, under real trading conditions, against an independent count, and under a stated error measure. Across 35 verified fashion stores, every location reached at least 97% absolute accuracy on 80 counting lines.
How accurate are people counters in fashion retail?
Advertised figures for the category run from 95% to 99.9%, but those describe controlled tests rather than a specific store. In this verification of 35 fashion stores, the floor across every location was 97% absolute accuracy against an independent count of 13,076 visitors.
How do retail chains compare store performance fairly?
By making sure every store's visitor count has been verified the same way before comparing anything built on it. If two stores are miscounted by different amounts, the comparison partly measures their counters rather than their trading.
Can people counting accuracy be verified across an entire chain?
Yes, by testing each location individually and publishing the floor rather than the average. This programme covered 35 stores between November 2024 and April 2026, each during periods the retailer selected.
Can a third party verify Indivd's counting accuracy?
Yes. The customer decides where, when, and how much is tested, can run the check without Indivd present, and can put the same test in anyone else's hands.