A retail people counting system automatically counts the people passing through a store entrance, using cameras or sensors, and reports footfall hour by hour. Combined with the number of till transactions, it gives you the store's conversion rate, so you can tell whether weak sales come from low traffic or from shoppers lost inside, and plan staff around real demand.
Key takeaways
- A retail people counting system counts the people passing through each entrance, hour by hour, using cameras or sensors.
- Conversion rate is transactions divided by visitors; it shows whether weak sales come from low traffic or from inside the store.
- There are four main technologies: infrared beams, overhead 3D sensors, IP cameras with computer vision, and Wi-Fi/Bluetooth signals.
- Measure counting accuracy at your own door with a manual count after installation, not from the manufacturer's brochure.
- Under Türkiye's KVKK, camera counting should process video locally, avoid facial recognition and inform visitors at the entrance.
What store managers cannot see today
Most stores close the day with the till report: how many transactions, what turnover, what average basket. Those numbers matter, but they only tell half the story. When turnover falls, two very different situations look identical: either fewer people came in, or the people who came in left without buying. The fix for the first is the window, the location and the campaign; the fix for the second is staffing, stock and the in-store experience.
In a chain the problem grows. If two branches take the same turnover but one has twice the footfall, that branch is actually missing sales. Without footfall data, the area manager compares branches on turnover alone and holds up the wrong one as the example.
Staff rotas are often guesswork too. Shifts are built on "Saturdays are busy", yet nobody knows exactly when the peak starts and ends, or how long the weekday lunchtime rush lasts, until it is measured.
What does not measuring store footfall cost?
Physical stores that do not measure footfall compete with a channel that measures every visit. An online shop knows how many people arrived, how many added something to the basket and how many paid; a physical store usually sees only the last step.
According to the Turkish Ministry of Trade's announcement of its E-Commerce Outlook Report 2025, e-commerce accounted for 19.3% of total trade in Türkiye in 2025.
Sales is also where businesses use AI most. According to TurkStat's AI Statistics 2025 bulletin, 46.5% of Turkish enterprises using AI use it for marketing or sales, making it the most common purpose. A store that does not measure has to judge a campaign, a window change or a new branch location by instinct, while competitors who measure can compare.
Then there is loss. The US National Retail Federation's National Retail Security Survey 2023 found that the average retail shrink rate rose from 1.4% to 1.6% in FY2022, representing $112.1 billion in losses. The figure is American and shrink has many causes, but it is clear that too few staff at peak times can mean both lost sales and weaker control. To plan staff around real demand, you first have to measure that demand.
People counting technologies compared: sensor or camera?
There are four main technologies behind a retail people counting system. Because accuracy depends on door width, lighting, crowding and mounting height, it is more useful to look at where each one is strong than to quote a single figure:
| Technology | How it counts | Strength | Watch out for |
|---|---|---|---|
| Infrared beam | Counts breaks in a beam across the door | Simple and inexpensive | Counts people side by side as one; weak at telling direction |
| Overhead 3D/stereo sensor | Tracks heads using depth information | Good in crowds and at direction | Mounting height and door width matter |
| IP camera + computer vision | Detects people in the image and counts crossings of a virtual line | Can reuse existing cameras; heat maps, dwell time and child/staff separation can be added | Privacy design and processing power must be planned |
| Wi-Fi/Bluetooth signals | Detects phone signals | Trends in returning visitors | Misses phones that are off or configured differently; unreliable for counting on its own; phone identifiers also need separate assessment under KVKK, Türkiye's data protection law |
Camera-based counting with computer vision has become the most flexible option, because the same camera can later answer questions about heat maps, till queues or window dwell time. The logic is the same detect-and-classify approach we describe in computer vision quality inspection; the object is simply a person rather than a product.
Setting up a retail people counting system: seven steps
- Write down the question. "How many came in?" is not enough. Are you measuring conversion, staff planning, campaign impact or an occupancy limit? The question drives the sensor choice.
- Define the counting points. Every entrance must be counted separately. In a shopping centre unit that means the door; on the high street you may also count passers-by at the window.
- Choose the technology. If your existing cameras are suitable, assess them first; for managing that camera estate centrally, see our guide to ONVIF IP camera central management.
- Exclude staff and children. Staff movements inflate the count. Decide up front whether staff use a separate entrance or are excluded by uniform or badge. Decide too whether children should be counted separately; some overhead sensors and computer vision models can do this with a height threshold.
- Measure accuracy in your own store. After installation, count by hand for a few hours at different levels of busyness and compare with the system. The figure in the brochure may not hold for your door.
- Combine with sales data. Match footfall with transaction counts from the till or enterprise resource planning (ERP) system in the same time slots; that is the only way to calculate conversion. We explain the technical side in API integration explained.
- Tie it to decisions. Use the data in the weekly meeting: shift times, campaign timing, window changes and branch comparisons. Our guide to BI dashboard KPIs helps decide which indicators belong on the dashboard. To turn the weekly figures into a written commentary, see AI management reporting.
Which indicators and decisions does people counting data feed?
The core indicators for a retail people counting system are hourly footfall, conversion rate (transactions / visitors), revenue per visitor, visitors per member of staff and the share of passers-by who come in. Footfall trends can also feed AI demand forecasting models, since store traffic is a leading indicator of sales.
Once the data is there, decisions become concrete. If footfall rises at lunchtime while conversion drops in the same hours, the problem is not traffic but shoppers waiting at the till or the fitting rooms, and the fix is to move staff breaks. If a campaign lifts footfall but leaves conversion unchanged, the message is drawing people in but the layout or stock is not holding them. And comparing branches on revenue per visitor rather than turnover reveals which one is genuinely well run.
If till queues are a persistent problem, a self-checkout kiosk is one way to reduce waiting at peak times.
Putting footfall data next to in-store screen play logs shows which content works at which time of day; we cover the screen side in our guide to retail digital signage management.
Footfall data becomes useful when it meets sales and staffing data on the same dashboard; we cover that set-up in our article on retail chain store management system.
How do you make people counting comply with KVKK?
Any people counting project that uses cameras must be designed with data protection in mind. In Türkiye this is governed by KVKK, the Personal Data Protection Law. Even if video is only processed in real time and never stored, images of identifiable people are being processed, so this can count as processing personal data and the duty to inform still applies.
The Turkish Data Protection Authority's announcement of 8 June 2026 was written for workplace security cameras, but it is a useful guide for counting cameras too: the purpose of the cameras should be clearly defined, footage kept for the shortest possible period with automatic deletion, access limited to authorised people, and people informed that the area is being recorded. The Authority's announcement also does not accept monitoring employee productivity as a legitimate purpose, so a counting system must not turn into staff performance tracking.
The practical design principles for camera-based people counting are these: process video on the camera or a local device where possible and send only counts to the centre; do not use facial recognition; do not store footage, or keep the retention period short and in writing; put a privacy notice at the entrance. It is best to have the legal assessment for your specific project done with a data protection specialist. You will find more detail in our articles on CCTV recording under KVKK and AI and personal data under KVKK.
How does Digital Bridge run people counting projects?
At Digital Bridge we treat people counting not as a sensor sale but as a data project that should end in a changed decision. We do not sell off-the-shelf packages; after a needs analysis we prepare a written proposal setting out scope, phases and cost. We work across all of Türkiye's provinces, remotely and on site:
- Discovery and needs analysis. We review your entrances, existing cameras, till and ERP systems. Through our retail and chain store solutions we agree in writing which problem will be measured and which indicators will reach the decision meeting.
- Pilot. We start with one or two branches. If your existing ONVIF-compatible cameras are suitable we build counting on your IP camera monitoring infrastructure; if not, we use a suitable sensor, and we validate accuracy with a manual count alongside your team.
- Computer vision. Where you need staff exclusion, heat maps or queue measurement, our computer vision service adapts the model to your store's lighting, door angle and uniforms. Where possible, we recommend an architecture that processes video locally and passes on only the counts.
- Integration and reporting. We join counting data with till and ERP transactions through API integration and show conversion by branch, hour and day on a BI dashboard.
- Privacy design. We prepare the privacy notice, retention period and access rights together with our data protection compliance consultancy.
Counting data can also talk to other screens in the store. For digital signage displays that change campaign content at busy times, see our digital signage guide.
For the link between footfall and stock, read our article on retail inventory optimisation. If store reports still live in spreadsheets, our article on moving from Excel reports to BI explains that transition.
Starting AI in retail with people counting
Retail people counting is a good first AI project in retail: the scope is narrow, the result is measurable and it feeds decisions directly. Our article on where to start with AI in business explains how to choose a first project, and you will find other use cases on our Artificial Intelligence topic page.
Next step
To prepare for a people counting project, pull out the last three months of hourly transaction counts by branch and put them next to your staff rota; that table shows where counting data will fit. Then contact us: we will assess your entrances and camera infrastructure with you and set out the pilot branch and the indicators to measure in a written proposal.